+1 (208) 254-6996 [email protected]

 250-350words content notes for EACH Article. SEE PDF (GOOD ARTICLE EXAMPLE) !    

Bell, D. R., Gallino, S., & Moreno, A. (2014). How to win in an omnichannel world. MIT Sloan Management Review, 56(1), 45-53.

Don't use plagiarized sources. Get Your Custom Essay on
Article Notes
Just from $13/Page
Order Essay

Dawar, N., & Bendle, N. (2018). Marketing in the age of Alexa: AI assistants will transform how companies and customers connect. Harvard Business Review, 96(3), 80.

4/24/2020 onlinetext.html

file:///Users/matthewfisher/Downloads/MKTG064903-S20R-Article Notes The Metrics that Marketers Muddle-359053/Marina Hamagaki_504182_assignsubmission… 1/3

Article Notes 3

Bendle, N. T., & Bagga, C. K. (2016). The metrics that marketers muddle. MIT Sloan Management Review, 57(3), 73-82.

Despite their widely acknowledged importance, some popular marketing metrics are regularly misunderstood and misused. One major reason for marketing’s diminishing role is the difficulty of meaning its impact: The value marketers generate is often difficult to quantify. The main goals of this article are to understand how these marketing metrics are used and understood and to develop ideas to help marketers unmuddle their metrics. The authors conducted surveys from managers from all functions across the business-to-business and business-to-consumer industries.

5 Best Known Marketing Metrics:

–       Market share

–       Net Promoter Score (NPS)

–       The Value of a ‘Like’

–       Consumer Lifetime Value (CLV)

–       Return on Investment (ROI)

Market Share

Market share is a popular marketing metric. One reason for why manager value market share is that research from the 1970s suggested a link between market share and ROI; however, the linkage may be less clear: the studies have found it is often correlational rather than causal. The survey found that there were two ways managers used market share: as an ultimate objective or as an intermediate measure of success. Increasing market share is not a meaningful ultimate objective for maximizing shareholder value and stakeholder management: If the aim is to maximize the returns to shareholders, increased market share offers no benefits unless it eventually generates profits. In some markets, bigger can be better; however, economies of scale do not automatically apply all markets.

Unmuddling Market Share:

The authors suggest a simple set of rules for the appropriate use of the market share metric:

–       Managers should not consider market share as the ultimate objective or as a proxy for absolute size.

–       Managers should evaluate it from the competitors’ and consumers’ point of view. If an increase in market share is not going to get positive feedback from competitors and consumers, then an increase in market share will not lead to a productive result.

–       Managers should analyze whether market share drives profitability in your industry. Companies with superior products tend to have high market share and high profitability because product superiority causes both.

4/24/2020 onlinetext.html

file:///Users/matthewfisher/Downloads/MKTG064903-S20R-Article Notes The Metrics that Marketers Muddle-359053/Marina Hamagaki_504182_assignsubmission… 2/3

This means that the two metrics are correlated, BUT it does not necessarily mean that increasing market share will increase profits.

Net Promoter Score (NPS)

This metric is used to measure customer loyalty to a firm. Companies among diverse industries have embraced NPS as a way to monitor their customer service operations while NPS also has been seen as a system that allows managers to use the scores to shape managerial actions.

One of the advantages of NPS is its simplicity: It is easy for managers and employees to understand the goal of having more promoters and fewer detractors. However, there are weaknesses: E.g., in the net promoter literature, a customer’s worth to Apple has been described as the customer’s spending, ignoring the costs associated with serving the customer. It is also easy to imagine how to increase the net promoter score (such as making customers happier) while destroying even to-line growth (by slashing prices). Another problem with NPS as a metric is the classification system: The boundaries between scores of 6 and 7 (detractors and passives) and 8 and 9 (passive and promoters) seem somewhat arbitrary and culturally specific.

Unmuddling NPS:

The value of NPS depends on whether a manager sees it as a metric or as a system. The authors suggest that the NPS metric cannot change the marketing performance. However, they advise using this metric as a part of a system employed in evaluating the performance which might lead to a cultural shift within the organization.

The Value of a ‘Like’

This metric is used for measuring the social media capital of the company. New approaches are being developed all the time and they have the potential to aid understanding of how social media creates value. It is measured as the difference between the average value of customers endorsing the company and the average value of the customers who are not endorsing the company. The majority of managers link between their social media spending the value of a ‘like’. However, it does not mean that the cause of the differences in users’ value is attributable to a company’s social media strategy. And the reason that social media strategy shouldn’t be seen as the driver of value difference between fans and nonfans is because customers who are social media fans will differ from nonfans for reasons unrelated to the company’s social media strategy.

Unmuddling the Value of a ‘Like’:

This difference between two groups of consumers does not suggest an effect of online marketing activity or lack thereof. It should be investigated thoroughly by the managers. If the management is using the revenue to measure customer value, then this marketing metric does not give a good estimate. However, if the company does want to understand the impact of social media marketing, they should use randomized control experiments to derive causal answers.

Consumer Lifetime Value (CLV)

Consumer lifetime value (CLV), which is the present value of cash flows from a customer relationship, can help managers in decision making related to investment in developing customer relationships, as it is used to measure the value of the current customer base. If the management is using the customer value in their decision-making process, then CLV is a useful tool for them.

Unmuddling CLV:

4/24/2020 onlinetext.html

file:///Users/matthewfisher/Downloads/MKTG064903-S20R-Article Notes The Metrics that Marketers Muddle-359053/Marina Hamagaki_504182_assignsubmission… 3/3

The authors suggest that CLV calculations should not include the customer acquisition cost and the estimated CLV should be compared to the estimated acquisition cost to derive conclusions. The bigger the difference between the estimated CLV and the estimated acquisition cost, the better the acquisition campaign.

Return on Investment (ROI)

Return on investment is a popular and potentially important metric allowing for the comparison of disparate investments. A critical requirement for calculating ROI is knowing the net profit generated by a specific investment decision. According to the authors, there is confusion within management over the use of ROI. However, as ROI is understood across disciplines, it is a powerful metric to communicate across the organization.

Unmuddling ROI:

The authors advise that if a manager is assessing the financial return on an investment, then ROI is an appropriate metric and can be calculated by dividing the incremental profits by the investments. Agribusiness marketing managers who are passionate about establishing the credibility of the value created through marketing should be thorough in their use of metrics. Most importantly, they should be able to understand the metric, its use and what it represent

R E P R I N T N U M B E R 5 6 1 1 5

F A L L 2 0 1 4 V O L . 5 6 N O . 1

How to Win in an Omnichannel World By David R. Bell, Santiago Gallino and Antonio Moreno

Please note that gray areas refl ect artwork that has been intentionally removed. The substantive content of the article appears as originally published.

PLEASE NOTE THAT GRAY AREAS REFLECT ARTWORK THAT HAS BEEN INTENTIONALLY REMOVED. THE SUBSTANTIVE CONTENT OF THE ARTICLE APPEARS AS ORIGINALLY PUBLISHED.

PAULA CUNEO, A TEACHER IN Ashland, Massachusetts, ordered 10 pairs of corduroy pants in a range of sizes and colors from Gap Inc.’s website, and later returned seven of them, ac-

cording to a 2013 Wall Street Journal article.1 Ms. Cuneo is, perhaps unwittingly, an exemplar of a

key challenge in today’s omnichannel retail environment — an environment where customers shop

through a variety of online and offline channels. The challenge omnichannel retailers face is this:

How can retailers provide consumers with information (about what products best suit them) with-

out incurring downside on product fulfillment (delivery of products)?

The omnichannel environment presents new challenges and opportunities for both information

and product fulfillment. This is equally true for “traditional” retailers like the Gap, which began busi-

ness with physical stores, and “new” retailers like New York-based eyeglasses brand Warby Parker,

which started out by selling online. While all

retailers need to effectively and efficiently

manage fulfillment and information provi-

sion, there are important nuances to how this

happens — depending on where and how

the retailer got started and what kinds of im-

provement create the most leverage.

This article delivers a customer-focused

framework showing how to win in the omni-

channel env ironment throug h critical

innovations in information delivery and

product fulfillment. The framework emerged

from our research with both traditional and

nontraditional retailers. To thrive in the new

environment, retailers of all stripes and

origins need to deploy information and ful-

fillment strategies that reduce friction in

every phase of the buying process. This

means simultaneously providing, in a cost-

effective and nar rative-enhanc ing way,

FALL 2014 MIT SLOAN MANAGEMENT REVIEW 45

How to Win in an Omnichannel World Retail customers are now “omnichannel” in their outlook and behavior — they use both online and offline retail channels readily. To thrive in this new environment, retailers of all types should reexamine their strategies for delivering information and products to customers. BY DAVID R. BELL, SANTIAGO GALLINO AND ANTONIO MORENO

THE LEADING QUESTION How can retailers effectively adapt to an omnichannel environment?

FINDINGS �Consumers’ omni- channel behavior is spurring innova- tions in the ways retailers provide information and products.

�Both traditional and online retailers should consider hybrid online-offline approaches.

�Hybrid approaches include inventory- only showrooms and “buy online, pick up in store” options.

R E T A I L I N G

46 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

information that removes initial uncertainties and

barriers to purchase — as well as fulfillment options

that allow retailers to get their products to customers

in the most convenient and cost-effective way.

Our research relies on detailed customer-behav-

ior data (such as visits, purchases and returns) from

omnichannel retailers. We then used these data to

perform statistical tests of the impact of manage-

ment interventions (such as website enhancements

and showroom openings) on overall demand and

fulfillment efficiency. (See “About the Research.”)

We explain why the best way to navigate the om-

nichannel environment is to: (1) take a customer

perspective and (2) view the activities of the com-

pany through the lens of the two core functions of

information and fulfillment. Last, and most impor-

tant, we elaborate on each of the core elements of

our information and fulfillment matrix in detail

and highlight the key implications for omnichan-

nel retail practice.

A Customer-Focused Framework We adopted a customer-focused perspective on

omnichannel strategy, as our research and experi-

ence tells us this that is best way to ensure the

formation of cohesive and effective initiatives. The

framework asks two simple yet fundamental ques-

tions: (1) How will customers get the information

they need to facilitate their purchase decisions? and

(2) How will transactions be fulfilled? (See “The

Information and Fulfillment Matrix.”)

When it comes to fulfillment, customers either

visit the store to pick up items or the “store comes to

them” when products are delivered. This is true with

information, too, as customers either visit stores to

obtain (offline) information or seek information

remotely, either online or perhaps through catalogs.

Prior to the advent of the Internet, there were re-

ally only two generic types of retailers. The first type

was traditional retailers, indicated by the upper-left

quadrant in our matrix (quadrant 1), wherein all

product information is delivered offline through

physical stores, and customers visit stores to take ful-

fillment. Many retailers still operate exclusively in

this quadrant, as exemplified by Ross Stores or

HomeGoods. The second type was catalog retailers,

which can be considered an early precursor to to-

day’s pure-play online retailers (quadrant 4), in

which information is delivered directly to customers

via the Internet (instead of a catalog) and product

fulfillment takes place via delivery.

The development of the commercial Internet

spurred growth in the number of pure-play Internet

retail companies (quadrant 4), with online transmis-

sion of information and fulfillment via delivery, as

exemplified by Amazon.com or Overstock.com. The

great promise of the “omnichannel revolution,”

however, lies not simply in the new retail businesses

ABOUT THE RESEARCH We conducted our academic research on omnichannel issues by using large customer databases from Crate & Barrel, Bonobos.com and WarbyParker.com and supplementing them with other data from external public sources as necessary (discussed below). We worked closely with executives to elaborate research issues that were not only of theoretical or aca- demic interest but also of practical economic importance to retailers.

To facilitate our research, management pro- vided us with the following kinds of data fields: unique customer ID (disguised for confidential- ity), transaction date, transaction items, transaction value and customer (shipping) ZIP code. We were also privy to information on other important kinds of customer activity, including website visits and data on product sampling and product returns. Management made available detailed information on the

timing and nature of specific interventions (for example, website improvements) so that we could assess the impact, if any, they had on sales, returns and other customer behaviors.

Furthermore, since it is well known that sales through online channels vary dramatically by geographic location in accordance with the types of customers living there and their local shopping options,i we appended a rich set of geodemographic data to the sales data provided by the companies. One nice feature of this data is that it is freely or cheaply available from gov- ernment and commercial sources, such as the U.S. Census and the geographic information system provider Esri.

We characterized each U.S. location (ZIP code) according to several local features, including the age, income, education and ethnicity of local residents; total population; and population density. We were also able to describe several aspects of the local offline

retailing environment, including the number of offline stores likely to compete with the web- sites of the companies we were studying, offline expenditures on the product category, travel distance to offline retailers, and so on.

After assembling the customer and manage- ment intervention data, we estimated econometric models to assess the impact of key management interventions: specifically, whether they worked (for example, whether a “buy online, pick up in store” program increased sales at the website) and, if so, whether there were any unintended consequences, either pos- itive or negative. In many instances, we were analyzing so-called “natural experiments,” in the sense that the management interventions took place in the field and a number of real cus- tomers with real buying experiences were exposed to them, while other customers were unaffected by these changes and could be used as a control group.

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 47

made possible by Internet connectivity, but also

more subtly and profoundly in the emergence of the

retail strategies located in quadrants 2 and 3.

One observation immediately apparent from our

information and fulfillment matrix is that a retailer

has the potential to operate in any of the four quad-

rants. However, to develop intuition for what kinds

of combinations of information delivery and fulfill-

ment will succeed for what kinds of businesses and

why, it helps to first focus on the exemplar cases for

quadrants 1 and 4 and think about the products and

experiences for which they excel.

All “traditional” retailers began life in quadrant 1,

delivering information to customers via in-store ex-

periences and fulfilling demand there as well. This

includes retailers that stock others’ brands, as well as

vertical retailers that sell their own brands (typically

manufacturers such as Apple, Nike or Patagonia). To

win in an omnichannel world, however, a traditional

player needs to expand through quadrants 2, 3 and 4.

Similarly, we argue that a pure-play online retailer

needs to pursue quadrant 2 and 3 strategies, and

consider partnerships with traditional (quadrant 1)

retailers as well.2

Navigating the Framework The predominance of quadrant 1 retailing is an em-

pirical fact of developed and emerging markets alike.

In 2013, e-commerce (excluding travel) as a percent-

age of the total retail market was a mere 10% in the

United Kingdom, 8% in the United States and 6%

and 1% in China and India, respectively.3

Nevertheless, it is natural and most likely impera-

tive for most traditional retailers to participate in

quadrant 4 and build an e-commerce operation.

Quadrant 4 retailing is growing at a rapid pace, both

within the United States and abroad.4 We predict

that the “synchronized” experiences of quadrant 1

and quadrant 4 activities — where information and

fulfillment activities are accomplished in the same

channel — will continue to anchor retail and that

new players will start businesses in either quadrant,

but that new players’ operations will be greatly

enhanced by strategies and activities that fall in

quadrants 2 and 3. This is due to the opportunities

that arise from decoupling the information and ful-

fillment dimensions of quadrants 1 and 4. Let’s start

by looking at the information dimension.

Information: Remote vs. Direct Access When retailers operate in online quadrants (2 and

4), this dictates that they give customers informa-

tion about the products through some remote

means, such as a catalog or a website. This form of

information delivery is most suited to products

containing few, if any, “nondigital” attributes.5 A

nondigital attribute — for example, the fit and feel

of apparel and related categories or the taste and

texture of products — is difficult to fully observe

and assess without a physical inspection. Uncer-

tainty about nondigital attributes is a key barrier to

consumers’ willingness to buy online and is an

especially important deterrent for first-time online

purchases.6 Once a consumer has experience with

a brand or product, he or she may be willing to rely

on purely online information for subsequent on-

line purchases.

Conversely, when companies operate in the

upper “offline” quadrants (1 and 3), they give cus-

tomers direct access to product information via

physical access to products. This method of infor-

mation delivery is especially well-suited to retailing

products that have significant “high-touch” ele-

ments, important service requirements or significant

nondigital attributes. In fact, it is this very same

point of strength of quadrant 1 retailing that makes

retailers who only operate there especially vulnera-

ble to consumers “showrooming” — examining

merchandise onsite but purchasing at a lower price

THE INFORMATION AND FULFILLMENT MATRIX In an omnichannel retail environment, customers can either visit stores to obtain information, or they can seek information remotely. They can also either visit a store to pick up items, or the store can “come to them” when products are delivered.

Information Delivered

Fulfillment

Pure-Play E-Commerce • Amazon.com • Overstock.com

Traditional Retail • HomeGoods • Ross

Online Retail Plus Showrooms • Warby Parker • Bonobos

Shopping and Delivery Hybrid • Crate & Barrel • Toys “R” Us

Online

Pickup

Offline

Delivery

1

2

3

4

48 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

online.7 Mitigation or elimination of showrooming

is a vital management objective for traditional retail-

ers.8 It is highly detrimental to the showroomed

retailer if the consumer, after appropriating the in-

formation from that retailer, buys online from a

competitor.

Thus, if quadrant 4 is well suited to selling prod-

ucts for which customers either have a fair degree

of certainty about what to expect or can expect only

limited value from a live customer-service experi-

ence, and quadrant 1 is well suited to high-touch

products yet highly vulnerable to showrooming,

this raises intriguing possibilities for hybrid experi-

ences (quadrants 2 and 3). They can both enhance

the customer experience and improve performance

outcomes for retailers.

Fulfillment: Delivery vs. Pickup With regard to fulfillment, there are important dif-

ferences in the customer experience and business

impacts on the retailer between the quadrants on

the left (1 and 2) and the quadrants on the right (3

and 4). From a consumer’s point of view, obtaining

a product after a visit to a physical location (store)

has both advantages and disadvantages. The cus-

tomer does not have to pay for shipping or wait for

delivery of the product; however, the consumer

incurs travel costs. Similarly, delivery has both

advantages and disadvantages for consumers. Dis-

advantages include waiting time (and delayed

gratification) and perhaps shipping costs as well.

Advantages include mitigation of travel costs and

the ability to access products that would not neces-

sarily be displayed in a physical store.9

From the retailer’s point of view, fulfilling

orders in stores (quadrants 1 and 2) or via delivery

(quadrants 3 and 4) pose very different challenges.

When orders are fulfilled in stores, there are impor-

tant location and store-design decisions to be

made. Stores have to be accessible to customers and

large enough to hold inventory. These constraints

can translate into significant real estate costs.

Furthermore, in order to fulfill transactions in

stores, the retailer must carry the right products in

the right stores at the right time. In order to do that,

the company has to decide which products to carry

in each brick-and-mortar retail location and also

accurately forecast demand for each product and

store — something that is notably harder to do on

a per-store basis than at a higher level of aggrega-

tion.10 Higher demand uncertainty at the store level

results in higher supply-mismatch costs, which are

then manifested through excess inventory or lost

sales due to products being out of stock.

Conversely, fulfilling orders via delivery (quad-

rants 3 and 4) relaxes some of the design constraints

for the retailer’s physical locations. First, fulfillment

can be centralized from a distribution center located

in a less expensive area, although some orders can be

shipped from a conventional store. Second, central-

ized fulfillment makes forecasting demand easier

because it allows forecasts to be made at a more ag-

gregate level, reducing supply-demand mismatch

costs.11 This efficiency is especially important when

variety is high (with a large number of SKUs) and

the demand for each individual product is low.12

Information Online, Fulfillment Offline Crate & Barrel, based in Northbrook, Illinois, is a

traditional retailer of furniture and housewares

with strength and heritage in quadrant 1 and an

active presence in quadrant 4 — which makes it a

retailer that offers predominantly “synchronized”

experiences. Could it enhance its overall perfor-

mance though quadrant 2 strategies, which rest on

hybrid experiences? To test the potential of one form

of hybrid strategy, management implemented a

“buy online, pick up in store” (BOPS) option at

Crate & Barrel stores throughout the United States.

Other retailers such as Toys “R” Us and The Home

Depot have launched similar initiatives during

recent years.13

Crate & Barrel has numerous stores located in

the United States and Canada, and the BOPS op-

tion was offered to shoppers in the United States

only. To isolate the impact of BOPS, we looked for

differences in shopper behavior (for example, sales

levels for a particular product category or overall

store traffic) between the two countries.14 Since we

controlled for other differences between the United

States and Canadian stores, we could attribute any

remaining differences in sales and store traffic to

the availability of the BOPS option.

There are two reasons why BOPS, in theory,

offers shoppers a compelling value proposition.

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 49

First, they can very easily get accurate information

about prices and availability of items that they are

interested in buying before placing their order.

Second, since they can pick the item up, they get

gratification from immediate access to the pur-

chased product. Thus, BOPS eliminates a critical

search friction for shoppers — wondering whether

a particular item is available in a store and what it

costs. BOPS also counters a key deficiency of the

online shopping experience — waiting for pur-

chased items to be delivered. In a very real sense,

BOPS gives shoppers the best of both worlds —

full information delivered before purchase (no

search friction) and immediate fulfillment (no

waiting for delivery).

Management’s expectation was that post-BOPS,

online sales in the United States would increase.

Surprisingly, that didn’t happen. In fact, online

sales went down, even though traffic to the website

went up. To understand why, note that most of the

items sold at Crate & Barrel have nondigital,

“touch-and-feel” attributes that are hard to com-

municate online. Hence, even thoug h B OPS

allowed shoppers to fully resolve uncertainty about

price and availability prior to shopping, it did not

allow them to resolve uncertainty about products’

nondigital attributes.

Nevertheless, overall sales at the stores went up.

Shoppers, having confirmed that the products they

wanted were in stock and appropriately priced,

went to the store to inspect and buy. BOPS removed

search friction for information that can be effi-

ciently communicated online (price and in-stock

position) and eliminated post-purchase waiting

time. The online purchase option could not, how-

ever, help customers eliminate uncertainty about

nondigital attributes.

A variant of the BOPS shopping process is ROPO

(“research online, purchase offline”), or “reverse

showrooming.” For instance, traditional retailers

that commit to providing accurate price and inven-

tory information online, and that otherwise engage

customers effectively online as well, can see increased

traffic and sales in their physical stores. Furthermore,

in our experience, both BOPS and ROPO customers

can, after visiting the store in person, generate incre-

mental sales in product categories other than the

one(s) that drove the initial visit.

Information Offline, Fulfillment Online If it makes sense for Crate & Barrel to “move down”

from quadrant 1 into quadrant 2 and start provid-

ing a richer online-information experience, it may

also make sense for Warby Parker (and other ini-

tially pure-play Internet retailers) to expand from

quadrant 4 into quadrant 3. As we noted earlier, for

products that require service and/or have high

number of touch-and-feel components, offline

delivery of product information is likely to be val-

ued by customers. There might be other benefits as

well, including an increase in brand awareness and

positive reinforcement of the brand’s legitimacy.

Management at Warby Parker, well aware that

many customers like to touch and feel eyewear prior

to buying it, had, from the inception of their busi-

ness, offered a sampling program called “Home

Try-On,” which allows customers to order five

frames free of charge and keep them for five days,

with free shipping both ways.15 The site also offers a

virtual try-on system in which customers can upload

pictures of their faces and overlay different frames

onto their photos. Still, management realized that

these two channels might be insufficient for at least

some segment of customers; hence, they opened a

third channel — the inventory-only showroom.

(Warby Parker also participates in quadrant 1 for

part of its product line, offering in-store fulfillment

of sunglasses in their own-store retail locations in

New York City, Boston and Los Angeles.)

Inventor y-only showrooms are third-party

locations (typically stores that sell apparel and ac-

cessories) that display the full line of Warby Parker

frames. Customers can visit these stores, such as the

partner showroom in Old City in Philadelphia, try

the frames on and then have the product fulfilled

via delivery just as if they had ordered online. The

inventory-only showroom is an example of quad-

rant 3, as information is delivered offline, but

product sales are fulfilled via delivery.

We wanted to understand the impact of these

inventory-only showrooms on demand, brand

awareness and product returns. To assess any poten-

tial marketing and operational efficiency benefits of

this hybrid experience, we defined a “trading area”

for the showroom (typically a 30-mile radius

around the showroom) and analyzed the “natural

Furniture and housewares retailer Crate & Barrel was able to drive economically significant offline increases in traffic and sales by providing accurate price and inventory information online.

50 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

experiment” created when showrooms were opened.

A naive approach to assess the impact would be

to simply compare sales and returns in the trading-

area region before and after the introduction of the

showroom. This approach, however, would not

account for other factors at the location that could

also be driving sales and returns. Hence, we utilized

an econometric method called differences-in-

differences with propensity matching, in which we

compared the difference in sales between a “treat-

ment city” (a city with a showroom) and a “control

city” (a city without a showroom), after adjusting

for the fact that the locations with showrooms were

deliberately chosen by the Warby Parker manage-

ment team, rather than simply selected at random.

Using the adjusted data, we then compared the dif-

ference between treatment- and control-location

sales prior to the introduction of a showroom to

the difference between treatment- and control-

location sales after the showroom was opened.16

Our regression analyses highlighted several bene-

fits of Warby Parker’s showrooms. First, and perhaps

not too surprising, total sales increased about 9% in

the locations within the trading area of the show-

rooms. (Remember, this increase is relative to other

locations that are “matched” in other ways but that

do not have a showroom.) The company was able to

expand its sales by providing information offline.

Next, we found that website sales that had their origin

in the showroom trading area (as measured by the

ZIP code) increased significantly, too, by about 3.5%.

An offline showroom thus appears to confer aware-

ness and brand-legitimacy benefits such that new

customers show up in the online channel.

For the sampling channel, or Home Try-On pro-

gram, we analyzed not only sales but also the number

of customers who tried the program. After the show-

room opened, sales through this channel decreased

by 5.5% and the total number of customers trying

the sampling program declined by 8%. So, while this

channel now generated fewer sales, it became more

efficient as the conversion from ordering try-ons to

actual purchases increased significantly.

Hence, an online-first retailer that starts providing

product information offline can see improvements in

both realized demand and operational efficiency. A

key efficiency for Warby Parker was higher conversion

in the sampling program, but there were other

efficiencies as well. In locations within the trading area

of a showroom, online returns declined, as did the

probability of individual customers placing multiple

Home Try-On orders. Thus, for an online retailer,

adding showrooms is more than a mechanism for ex-

panding awareness and total demand. It also allows

customers to “sort” into their preferred channel on the

basis of their need for prepurchase information.17

The same positive effects can be tracked in tempo-

rary or “pop-up” stores. As we saw positive demand

impacts for fixed showrooms, we expected to see sim-

ilar benefits from increasingly popular pop-up or

movable stores as well.18 Warby Parker, for example,

has a retrofitted school bus that traveled throughout

the United States for several months, making stops in

numerous cities and towns.19 In analyzing its effect,

we found that in locations where the bus stopped,

sales increased, both in total and through the website,

implying that pop-up stores boost both sales and

awareness. Short-term pop-up retail is evolving rap-

idly as new platforms emerge to facilitate it. An

intriguing example is thestorefront.com, a website

that “connects artists, designers, and retailers with

beautiful, local retail space.”20

The Importance of Information for Customers Product information gets delivered to potential

customers not just by a retailer but also by other

customers of that retailer. Managers need to recog-

nize the impor tance of variation in physical

geography and types of offline environments. Prior

research has shown that online sales of commodity

products vary significantly with variations in real-

world factors, such as population density and

access to stores.21 But we found an equally impor-

tant impact on products with nondigital attributes.

This was reinforced through work we did with

Bonobos.com, a men’s fashion brand with e-com-

merce as its core distribution channel.

Bonobos, like Warby Parker, began life in quadrant

4 as a pure-play online retailer and later developed so-

called offline “guideshops.” These relatively small

(typically about 1,200 square feet), high-service loca-

tions provide sufficient inventory for the customer to

try on, but not to buy at the store and take home im-

mediately. That is, they utilize online fulfillment in the

same way that Warby Parker showrooms do.22 This

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 51

quadrant 3 strategy has been highly effective for

Bonobos, and aggressive expansion is planned.23

Moreover, several Bonobos items are available in

Nordstrom department stores throughout the United

States, and this traditional, quadrant 1 retailing has

been important for Bonobos’ growth.

Nevertheless, we speculated that offline delivery

of information by customers might be a critical fac-

tor as well. We specifically focused on offline

variation in what sociologists call “social capital” —

the extent to which colocated individuals share

information …

R E P R I N T N U M B E R 5 6 1 1 5

F A L L 2 0 1 4 V O L . 5 6 N O . 1

How to Win in an Omnichannel World By David R. Bell, Santiago Gallino and Antonio Moreno

Please note that gray areas refl ect artwork that has been intentionally removed. The substantive content of the article appears as originally published.

PLEASE NOTE THAT GRAY AREAS REFLECT ARTWORK THAT HAS BEEN INTENTIONALLY REMOVED. THE SUBSTANTIVE CONTENT OF THE ARTICLE APPEARS AS ORIGINALLY PUBLISHED.

PAULA CUNEO, A TEACHER IN Ashland, Massachusetts, ordered 10 pairs of corduroy pants in a range of sizes and colors from Gap Inc.’s website, and later returned seven of them, ac-

cording to a 2013 Wall Street Journal article.1 Ms. Cuneo is, perhaps unwittingly, an exemplar of a

key challenge in today’s omnichannel retail environment — an environment where customers shop

through a variety of online and offline channels. The challenge omnichannel retailers face is this:

How can retailers provide consumers with information (about what products best suit them) with-

out incurring downside on product fulfillment (delivery of products)?

The omnichannel environment presents new challenges and opportunities for both information

and product fulfillment. This is equally true for “traditional” retailers like the Gap, which began busi-

ness with physical stores, and “new” retailers like New York-based eyeglasses brand Warby Parker,

which started out by selling online. While all

retailers need to effectively and efficiently

manage fulfillment and information provi-

sion, there are important nuances to how this

happens — depending on where and how

the retailer got started and what kinds of im-

provement create the most leverage.

This article delivers a customer-focused

framework showing how to win in the omni-

channel env ironment throug h critical

innovations in information delivery and

product fulfillment. The framework emerged

from our research with both traditional and

nontraditional retailers. To thrive in the new

environment, retailers of all stripes and

origins need to deploy information and ful-

fillment strategies that reduce friction in

every phase of the buying process. This

means simultaneously providing, in a cost-

effective and nar rative-enhanc ing way,

FALL 2014 MIT SLOAN MANAGEMENT REVIEW 45

How to Win in an Omnichannel World Retail customers are now “omnichannel” in their outlook and behavior — they use both online and offline retail channels readily. To thrive in this new environment, retailers of all types should reexamine their strategies for delivering information and products to customers. BY DAVID R. BELL, SANTIAGO GALLINO AND ANTONIO MORENO

THE LEADING QUESTION How can retailers effectively adapt to an omnichannel environment?

FINDINGS �Consumers’ omni- channel behavior is spurring innova- tions in the ways retailers provide information and products.

�Both traditional and online retailers should consider hybrid online-offline approaches.

�Hybrid approaches include inventory- only showrooms and “buy online, pick up in store” options.

R E T A I L I N G

46 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

information that removes initial uncertainties and

barriers to purchase — as well as fulfillment options

that allow retailers to get their products to customers

in the most convenient and cost-effective way.

Our research relies on detailed customer-behav-

ior data (such as visits, purchases and returns) from

omnichannel retailers. We then used these data to

perform statistical tests of the impact of manage-

ment interventions (such as website enhancements

and showroom openings) on overall demand and

fulfillment efficiency. (See “About the Research.”)

We explain why the best way to navigate the om-

nichannel environment is to: (1) take a customer

perspective and (2) view the activities of the com-

pany through the lens of the two core functions of

information and fulfillment. Last, and most impor-

tant, we elaborate on each of the core elements of

our information and fulfillment matrix in detail

and highlight the key implications for omnichan-

nel retail practice.

A Customer-Focused Framework We adopted a customer-focused perspective on

omnichannel strategy, as our research and experi-

ence tells us this that is best way to ensure the

formation of cohesive and effective initiatives. The

framework asks two simple yet fundamental ques-

tions: (1) How will customers get the information

they need to facilitate their purchase decisions? and

(2) How will transactions be fulfilled? (See “The

Information and Fulfillment Matrix.”)

When it comes to fulfillment, customers either

visit the store to pick up items or the “store comes to

them” when products are delivered. This is true with

information, too, as customers either visit stores to

obtain (offline) information or seek information

remotely, either online or perhaps through catalogs.

Prior to the advent of the Internet, there were re-

ally only two generic types of retailers. The first type

was traditional retailers, indicated by the upper-left

quadrant in our matrix (quadrant 1), wherein all

product information is delivered offline through

physical stores, and customers visit stores to take ful-

fillment. Many retailers still operate exclusively in

this quadrant, as exemplified by Ross Stores or

HomeGoods. The second type was catalog retailers,

which can be considered an early precursor to to-

day’s pure-play online retailers (quadrant 4), in

which information is delivered directly to customers

via the Internet (instead of a catalog) and product

fulfillment takes place via delivery.

The development of the commercial Internet

spurred growth in the number of pure-play Internet

retail companies (quadrant 4), with online transmis-

sion of information and fulfillment via delivery, as

exemplified by Amazon.com or Overstock.com. The

great promise of the “omnichannel revolution,”

however, lies not simply in the new retail businesses

ABOUT THE RESEARCH We conducted our academic research on omnichannel issues by using large customer databases from Crate & Barrel, Bonobos.com and WarbyParker.com and supplementing them with other data from external public sources as necessary (discussed below). We worked closely with executives to elaborate research issues that were not only of theoretical or aca- demic interest but also of practical economic importance to retailers.

To facilitate our research, management pro- vided us with the following kinds of data fields: unique customer ID (disguised for confidential- ity), transaction date, transaction items, transaction value and customer (shipping) ZIP code. We were also privy to information on other important kinds of customer activity, including website visits and data on product sampling and product returns. Management made available detailed information on the

timing and nature of specific interventions (for example, website improvements) so that we could assess the impact, if any, they had on sales, returns and other customer behaviors.

Furthermore, since it is well known that sales through online channels vary dramatically by geographic location in accordance with the types of customers living there and their local shopping options,i we appended a rich set of geodemographic data to the sales data provided by the companies. One nice feature of this data is that it is freely or cheaply available from gov- ernment and commercial sources, such as the U.S. Census and the geographic information system provider Esri.

We characterized each U.S. location (ZIP code) according to several local features, including the age, income, education and ethnicity of local residents; total population; and population density. We were also able to describe several aspects of the local offline

retailing environment, including the number of offline stores likely to compete with the web- sites of the companies we were studying, offline expenditures on the product category, travel distance to offline retailers, and so on.

After assembling the customer and manage- ment intervention data, we estimated econometric models to assess the impact of key management interventions: specifically, whether they worked (for example, whether a “buy online, pick up in store” program increased sales at the website) and, if so, whether there were any unintended consequences, either pos- itive or negative. In many instances, we were analyzing so-called “natural experiments,” in the sense that the management interventions took place in the field and a number of real cus- tomers with real buying experiences were exposed to them, while other customers were unaffected by these changes and could be used as a control group.

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 47

made possible by Internet connectivity, but also

more subtly and profoundly in the emergence of the

retail strategies located in quadrants 2 and 3.

One observation immediately apparent from our

information and fulfillment matrix is that a retailer

has the potential to operate in any of the four quad-

rants. However, to develop intuition for what kinds

of combinations of information delivery and fulfill-

ment will succeed for what kinds of businesses and

why, it helps to first focus on the exemplar cases for

quadrants 1 and 4 and think about the products and

experiences for which they excel.

All “traditional” retailers began life in quadrant 1,

delivering information to customers via in-store ex-

periences and fulfilling demand there as well. This

includes retailers that stock others’ brands, as well as

vertical retailers that sell their own brands (typically

manufacturers such as Apple, Nike or Patagonia). To

win in an omnichannel world, however, a traditional

player needs to expand through quadrants 2, 3 and 4.

Similarly, we argue that a pure-play online retailer

needs to pursue quadrant 2 and 3 strategies, and

consider partnerships with traditional (quadrant 1)

retailers as well.2

Navigating the Framework The predominance of quadrant 1 retailing is an em-

pirical fact of developed and emerging markets alike.

In 2013, e-commerce (excluding travel) as a percent-

age of the total retail market was a mere 10% in the

United Kingdom, 8% in the United States and 6%

and 1% in China and India, respectively.3

Nevertheless, it is natural and most likely impera-

tive for most traditional retailers to participate in

quadrant 4 and build an e-commerce operation.

Quadrant 4 retailing is growing at a rapid pace, both

within the United States and abroad.4 We predict

that the “synchronized” experiences of quadrant 1

and quadrant 4 activities — where information and

fulfillment activities are accomplished in the same

channel — will continue to anchor retail and that

new players will start businesses in either quadrant,

but that new players’ operations will be greatly

enhanced by strategies and activities that fall in

quadrants 2 and 3. This is due to the opportunities

that arise from decoupling the information and ful-

fillment dimensions of quadrants 1 and 4. Let’s start

by looking at the information dimension.

Information: Remote vs. Direct Access When retailers operate in online quadrants (2 and

4), this dictates that they give customers informa-

tion about the products through some remote

means, such as a catalog or a website. This form of

information delivery is most suited to products

containing few, if any, “nondigital” attributes.5 A

nondigital attribute — for example, the fit and feel

of apparel and related categories or the taste and

texture of products — is difficult to fully observe

and assess without a physical inspection. Uncer-

tainty about nondigital attributes is a key barrier to

consumers’ willingness to buy online and is an

especially important deterrent for first-time online

purchases.6 Once a consumer has experience with

a brand or product, he or she may be willing to rely

on purely online information for subsequent on-

line purchases.

Conversely, when companies operate in the

upper “offline” quadrants (1 and 3), they give cus-

tomers direct access to product information via

physical access to products. This method of infor-

mation delivery is especially well-suited to retailing

products that have significant “high-touch” ele-

ments, important service requirements or significant

nondigital attributes. In fact, it is this very same

point of strength of quadrant 1 retailing that makes

retailers who only operate there especially vulnera-

ble to consumers “showrooming” — examining

merchandise onsite but purchasing at a lower price

THE INFORMATION AND FULFILLMENT MATRIX In an omnichannel retail environment, customers can either visit stores to obtain information, or they can seek information remotely. They can also either visit a store to pick up items, or the store can “come to them” when products are delivered.

Information Delivered

Fulfillment

Pure-Play E-Commerce • Amazon.com • Overstock.com

Traditional Retail • HomeGoods • Ross

Online Retail Plus Showrooms • Warby Parker • Bonobos

Shopping and Delivery Hybrid • Crate & Barrel • Toys “R” Us

Online

Pickup

Offline

Delivery

1

2

3

4

48 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

online.7 Mitigation or elimination of showrooming

is a vital management objective for traditional retail-

ers.8 It is highly detrimental to the showroomed

retailer if the consumer, after appropriating the in-

formation from that retailer, buys online from a

competitor.

Thus, if quadrant 4 is well suited to selling prod-

ucts for which customers either have a fair degree

of certainty about what to expect or can expect only

limited value from a live customer-service experi-

ence, and quadrant 1 is well suited to high-touch

products yet highly vulnerable to showrooming,

this raises intriguing possibilities for hybrid experi-

ences (quadrants 2 and 3). They can both enhance

the customer experience and improve performance

outcomes for retailers.

Fulfillment: Delivery vs. Pickup With regard to fulfillment, there are important dif-

ferences in the customer experience and business

impacts on the retailer between the quadrants on

the left (1 and 2) and the quadrants on the right (3

and 4). From a consumer’s point of view, obtaining

a product after a visit to a physical location (store)

has both advantages and disadvantages. The cus-

tomer does not have to pay for shipping or wait for

delivery of the product; however, the consumer

incurs travel costs. Similarly, delivery has both

advantages and disadvantages for consumers. Dis-

advantages include waiting time (and delayed

gratification) and perhaps shipping costs as well.

Advantages include mitigation of travel costs and

the ability to access products that would not neces-

sarily be displayed in a physical store.9

From the retailer’s point of view, fulfilling

orders in stores (quadrants 1 and 2) or via delivery

(quadrants 3 and 4) pose very different challenges.

When orders are fulfilled in stores, there are impor-

tant location and store-design decisions to be

made. Stores have to be accessible to customers and

large enough to hold inventory. These constraints

can translate into significant real estate costs.

Furthermore, in order to fulfill transactions in

stores, the retailer must carry the right products in

the right stores at the right time. In order to do that,

the company has to decide which products to carry

in each brick-and-mortar retail location and also

accurately forecast demand for each product and

store — something that is notably harder to do on

a per-store basis than at a higher level of aggrega-

tion.10 Higher demand uncertainty at the store level

results in higher supply-mismatch costs, which are

then manifested through excess inventory or lost

sales due to products being out of stock.

Conversely, fulfilling orders via delivery (quad-

rants 3 and 4) relaxes some of the design constraints

for the retailer’s physical locations. First, fulfillment

can be centralized from a distribution center located

in a less expensive area, although some orders can be

shipped from a conventional store. Second, central-

ized fulfillment makes forecasting demand easier

because it allows forecasts to be made at a more ag-

gregate level, reducing supply-demand mismatch

costs.11 This efficiency is especially important when

variety is high (with a large number of SKUs) and

the demand for each individual product is low.12

Information Online, Fulfillment Offline Crate & Barrel, based in Northbrook, Illinois, is a

traditional retailer of furniture and housewares

with strength and heritage in quadrant 1 and an

active presence in quadrant 4 — which makes it a

retailer that offers predominantly “synchronized”

experiences. Could it enhance its overall perfor-

mance though quadrant 2 strategies, which rest on

hybrid experiences? To test the potential of one form

of hybrid strategy, management implemented a

“buy online, pick up in store” (BOPS) option at

Crate & Barrel stores throughout the United States.

Other retailers such as Toys “R” Us and The Home

Depot have launched similar initiatives during

recent years.13

Crate & Barrel has numerous stores located in

the United States and Canada, and the BOPS op-

tion was offered to shoppers in the United States

only. To isolate the impact of BOPS, we looked for

differences in shopper behavior (for example, sales

levels for a particular product category or overall

store traffic) between the two countries.14 Since we

controlled for other differences between the United

States and Canadian stores, we could attribute any

remaining differences in sales and store traffic to

the availability of the BOPS option.

There are two reasons why BOPS, in theory,

offers shoppers a compelling value proposition.

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 49

First, they can very easily get accurate information

about prices and availability of items that they are

interested in buying before placing their order.

Second, since they can pick the item up, they get

gratification from immediate access to the pur-

chased product. Thus, BOPS eliminates a critical

search friction for shoppers — wondering whether

a particular item is available in a store and what it

costs. BOPS also counters a key deficiency of the

online shopping experience — waiting for pur-

chased items to be delivered. In a very real sense,

BOPS gives shoppers the best of both worlds —

full information delivered before purchase (no

search friction) and immediate fulfillment (no

waiting for delivery).

Management’s expectation was that post-BOPS,

online sales in the United States would increase.

Surprisingly, that didn’t happen. In fact, online

sales went down, even though traffic to the website

went up. To understand why, note that most of the

items sold at Crate & Barrel have nondigital,

“touch-and-feel” attributes that are hard to com-

municate online. Hence, even thoug h B OPS

allowed shoppers to fully resolve uncertainty about

price and availability prior to shopping, it did not

allow them to resolve uncertainty about products’

nondigital attributes.

Nevertheless, overall sales at the stores went up.

Shoppers, having confirmed that the products they

wanted were in stock and appropriately priced,

went to the store to inspect and buy. BOPS removed

search friction for information that can be effi-

ciently communicated online (price and in-stock

position) and eliminated post-purchase waiting

time. The online purchase option could not, how-

ever, help customers eliminate uncertainty about

nondigital attributes.

A variant of the BOPS shopping process is ROPO

(“research online, purchase offline”), or “reverse

showrooming.” For instance, traditional retailers

that commit to providing accurate price and inven-

tory information online, and that otherwise engage

customers effectively online as well, can see increased

traffic and sales in their physical stores. Furthermore,

in our experience, both BOPS and ROPO customers

can, after visiting the store in person, generate incre-

mental sales in product categories other than the

one(s) that drove the initial visit.

Information Offline, Fulfillment Online If it makes sense for Crate & Barrel to “move down”

from quadrant 1 into quadrant 2 and start provid-

ing a richer online-information experience, it may

also make sense for Warby Parker (and other ini-

tially pure-play Internet retailers) to expand from

quadrant 4 into quadrant 3. As we noted earlier, for

products that require service and/or have high

number of touch-and-feel components, offline

delivery of product information is likely to be val-

ued by customers. There might be other benefits as

well, including an increase in brand awareness and

positive reinforcement of the brand’s legitimacy.

Management at Warby Parker, well aware that

many customers like to touch and feel eyewear prior

to buying it, had, from the inception of their busi-

ness, offered a sampling program called “Home

Try-On,” which allows customers to order five

frames free of charge and keep them for five days,

with free shipping both ways.15 The site also offers a

virtual try-on system in which customers can upload

pictures of their faces and overlay different frames

onto their photos. Still, management realized that

these two channels might be insufficient for at least

some segment of customers; hence, they opened a

third channel — the inventory-only showroom.

(Warby Parker also participates in quadrant 1 for

part of its product line, offering in-store fulfillment

of sunglasses in their own-store retail locations in

New York City, Boston and Los Angeles.)

Inventor y-only showrooms are third-party

locations (typically stores that sell apparel and ac-

cessories) that display the full line of Warby Parker

frames. Customers can visit these stores, such as the

partner showroom in Old City in Philadelphia, try

the frames on and then have the product fulfilled

via delivery just as if they had ordered online. The

inventory-only showroom is an example of quad-

rant 3, as information is delivered offline, but

product sales are fulfilled via delivery.

We wanted to understand the impact of these

inventory-only showrooms on demand, brand

awareness and product returns. To assess any poten-

tial marketing and operational efficiency benefits of

this hybrid experience, we defined a “trading area”

for the showroom (typically a 30-mile radius

around the showroom) and analyzed the “natural

Furniture and housewares retailer Crate & Barrel was able to drive economically significant offline increases in traffic and sales by providing accurate price and inventory information online.

50 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

experiment” created when showrooms were opened.

A naive approach to assess the impact would be

to simply compare sales and returns in the trading-

area region before and after the introduction of the

showroom. This approach, however, would not

account for other factors at the location that could

also be driving sales and returns. Hence, we utilized

an econometric method called differences-in-

differences with propensity matching, in which we

compared the difference in sales between a “treat-

ment city” (a city with a showroom) and a “control

city” (a city without a showroom), after adjusting

for the fact that the locations with showrooms were

deliberately chosen by the Warby Parker manage-

ment team, rather than simply selected at random.

Using the adjusted data, we then compared the dif-

ference between treatment- and control-location

sales prior to the introduction of a showroom to

the difference between treatment- and control-

location sales after the showroom was opened.16

Our regression analyses highlighted several bene-

fits of Warby Parker’s showrooms. First, and perhaps

not too surprising, total sales increased about 9% in

the locations within the trading area of the show-

rooms. (Remember, this increase is relative to other

locations that are “matched” in other ways but that

do not have a showroom.) The company was able to

expand its sales by providing information offline.

Next, we found that website sales that had their origin

in the showroom trading area (as measured by the

ZIP code) increased significantly, too, by about 3.5%.

An offline showroom thus appears to confer aware-

ness and brand-legitimacy benefits such that new

customers show up in the online channel.

For the sampling channel, or Home Try-On pro-

gram, we analyzed not only sales but also the number

of customers who tried the program. After the show-

room opened, sales through this channel decreased

by 5.5% and the total number of customers trying

the sampling program declined by 8%. So, while this

channel now generated fewer sales, it became more

efficient as the conversion from ordering try-ons to

actual purchases increased significantly.

Hence, an online-first retailer that starts providing

product information offline can see improvements in

both realized demand and operational efficiency. A

key efficiency for Warby Parker was higher conversion

in the sampling program, but there were other

efficiencies as well. In locations within the trading area

of a showroom, online returns declined, as did the

probability of individual customers placing multiple

Home Try-On orders. Thus, for an online retailer,

adding showrooms is more than a mechanism for ex-

panding awareness and total demand. It also allows

customers to “sort” into their preferred channel on the

basis of their need for prepurchase information.17

The same positive effects can be tracked in tempo-

rary or “pop-up” stores. As we saw positive demand

impacts for fixed showrooms, we expected to see sim-

ilar benefits from increasingly popular pop-up or

movable stores as well.18 Warby Parker, for example,

has a retrofitted school bus that traveled throughout

the United States for several months, making stops in

numerous cities and towns.19 In analyzing its effect,

we found that in locations where the bus stopped,

sales increased, both in total and through the website,

implying that pop-up stores boost both sales and

awareness. Short-term pop-up retail is evolving rap-

idly as new platforms emerge to facilitate it. An

intriguing example is thestorefront.com, a website

that “connects artists, designers, and retailers with

beautiful, local retail space.”20

The Importance of Information for Customers Product information gets delivered to potential

customers not just by a retailer but also by other

customers of that retailer. Managers need to recog-

nize the impor tance of variation in physical

geography and types of offline environments. Prior

research has shown that online sales of commodity

products vary significantly with variations in real-

world factors, such as population density and

access to stores.21 But we found an equally impor-

tant impact on products with nondigital attributes.

This was reinforced through work we did with

Bonobos.com, a men’s fashion brand with e-com-

merce as its core distribution channel.

Bonobos, like Warby Parker, began life in quadrant

4 as a pure-play online retailer and later developed so-

called offline “guideshops.” These relatively small

(typically about 1,200 square feet), high-service loca-

tions provide sufficient inventory for the customer to

try on, but not to buy at the store and take home im-

mediately. That is, they utilize online fulfillment in the

same way that Warby Parker showrooms do.22 This

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 51

quadrant 3 strategy has been highly effective for

Bonobos, and aggressive expansion is planned.23

Moreover, several Bonobos items are available in

Nordstrom department stores throughout the United

States, and this traditional, quadrant 1 retailing has

been important for Bonobos’ growth.

Nevertheless, we speculated that offline delivery

of information by customers might be a critical fac-

tor as well. We specifically focused on offline

variation in what sociologists call “social capital” —

the extent to which colocated individuals share

information with, interact with and trust each

other.24 We examined online sales data from the in-

ception of the Bonobos site in October 2007 and

matched this data with data from the Social Capital

Community Benchmark Survey, available through

the Roper Center for Public Opinion Research at

the University of Connecticut at Storrs.25

What we found was striking. First, our statistical

analysis of the geographical distribution of sales

implied that up to half of all first-time sales to new

customers of Bonobos.com were partly influenced

by “social learning.” Second, sales grew at a faster

rate in locations with greater levels of social capital.

This second finding is quite subtle. It is not the case

that higher levels of offline social capital, per se,

spur online sales. Higher levels of social capital in a

location make information transfer there more

efficient — that is, what gets said locally is more re-

liable and believable in these locations than in

locations with lower levels of social capital. Hence,

when what is said is positive (as it was in the case of

Bonobos), sales increase more quickly.26

Since it is not always possible or practical for

managers to get access to academic datasets like

SCCBS, we also examined the ability of readily

available proxy variables to capture the “offline

information effect.” As is turns out, the number of

bars and liquor stores per capita in a location is a

workable proxy for social capital among 25- to

45-year-old fashion-forward males (the Bonobos.

com target customer). Holding everything else

constant, online sales are higher in locations where

this measure is higher.27

Virtual Fitting Rooms Reduce Returns As noted above, physical locations with “better” offline

information transmission between existing and

potential customers have higher online sales. More-

over, in physical locations with offline showrooms,

retailers can benefit from lower rates of returned

products. These important informational benefits

translate to the online world, as we have also found

that online sites with better information delivery

can reduce product returns.

This is critical for sellers. While product returns

have always been part of the traditional retail land-

scape (quadrant 1), the Internet channel has taken

the challenge of returns to a completely new level.

Andy Dunn, cofounder and CEO of Bonobos, notes,

“Between gross revenue and net revenue, you typi-

cally have a meaningful returns line item.”28 In fact,

the impact of returns on the bottom line of all online

retailers is becoming more pronounced. The United

Parcel Service, for instance, expected returns to in-

crease by 15% in the 2013 holiday season, relative to

what they were the year before; by some estimates,

one-third of all Internet sales get returned.29

“Fitting rooms” placed on websites are one poten-

tial antidote to the returns problem. Specifically, these

are technologies that provide online customers with

accurate fit information and size recommendations in

advance of any buying decision. In an ongoing

research project with the virtual-fitting-room tech-

nology company Metail,30 we explored whether the

richer information delivered online with those type of

tools results in higher sales and lower returns.

To conduct our tests, we randomly assigned cus-

tomers from a large online apparel retailer to one of

two conditions: They either had access to the Metail

virtual fitting tool, or they did not. We found that cus-

tomers with access to the virtual fitting tool had higher

conversion rates and lower return rates than those cus-

tomers without access to it. Thus, higher-quality

personalized information, even if delivered through an

online channel, is a potentially powerful ally in one of

most important battles in omnichannel retailing —

namely, the fight to bring down product return rates.

Serving the Omnichannel Customer In a 2013 article in MIT Sloan Management Review,

researchers Erik Brynjolfsson, Yu Jeffrey Hu and

Mohammad S. Raman predicted: “As the retailing in-

dustry evolves toward a seamless ‘omnichannel

retailing’ experience, the distinctions between physi-

cal and online will vanish, turning the world into a

52 MIT SLOAN MANAGEMENT REVIEW FALL 2014 SLOANREVIEW.MIT.EDU

R E T A I L I N G

showroom without walls.”31 We concur and think

our article provides sellers with a framework for nav-

igating this landscape.

We began this article describing a shopping pro-

cess in which a customer undermined the efficiency

of a company’s existing information and fulfill-

ment methods in order to get exactly what she

wanted. The drive for a solution to such inefficien-

cies promoted the development of our information

and fulfillment matrix — a customer-focused

framework for ar ticulating how information

should be delivered (online or offline) and how

demand should be filled (pickup or delivery).

We showed that a traditional retailer such as

Crate & Barrel can realize considerable gains by im-

proving its online information about nondigital

attributes of products. This gives the customer a

reason and willingness to interact with the retailer

outside the store environment and to initiate and

partially complete a transaction online before

entering the store and finishing it. By providing

accurate price and inventory information online,

the retailer was able to drive economically signifi-

cant offline increases in traffic and sales.

Similarly, we demonstrated that an online-first

retailer like Warby Parker can experience substantial

benefits from an offline presence that simply

showcases inventory. Offline showrooms deliver

economically significant impacts on sales, returns,

awareness and sampling efficiency in locations within

the showrooms’ trading areas. Furthermore, custom-

ers are able to choose the channel that best fits their

needs, with those customers wanting to touch and feel

before buying most likely to visit the showroom.

Like it or not, customers are omnichannel in their

thinking and behavior. Sellers need to be as well.

Omnichannel features initially perceived as “nice

add-ons” are becoming “must-haves.” The question

for sellers is no longer whether to operate an om-

nichannel strategy, but how to implement it most

effectively. Our research underscores that the best

sellers will win the omnichannel revolution by work-

ing across the permeable boundaries of information

and fulfillment, offering the right combination of

experiences for the customers that demand them.

David R. Bell is the Xinmei Zhang and Yongge Dai Pro- fessor and professor of marketing at the Wharton School at the University of Pennsylvania in Philadelphia,

Pennsylvania. Santiago Gallino is an assistant profes- sor of business administration at the Tuck School of Business at Dartmouth College in Hanover, New Hamp- shire. Antonio Moreno is an assistant professor of managerial economics and decision sciences at the Kellogg School of Management at Northwestern Uni- versity in Evanston, Illinois. Comment on this article at http://sloanreview.mit.edu/x/56115 or contact the authors at [email protected]

REFERENCES

1. S. Banjo, “Rampant Returns Plague E-Retailers,” Wall Street Journal, Dec. 22, 2013.

2. As Andy Dunn, founder and CEO of men’s clothing brand Bonobos, has written, “At the end of the day, you’re not building an e-commerce company, you’re build- ing a brand that has e-commerce as its core distribution channel. The difference is subtle but momentous.” See A. Dunn, “E-Commerce Is a Bear,” May 20, 2013, http://medium.com.

3. “eCommerce Disruption: A Global Theme — Trans- forming Traditional Retail,” white paper, Morgan Stanley Research, New York, January 6, 2013.

4. “Alibaba: The World’s Greatest Bazaar,” Economist, March 23, 2013.

5. The important distinction between digital and nondigital attributes was introduced to the academic literature in R. Lal and M. Sarvary, “When and How Is the Internet Likely to Decrease Price Competition?,” Marketing Science 18, no. 4 (November 1999): 485-503.

6. See, for example, A.M. Degeratu, A. Rangaswamy and J. Wu, “Consumer Choice Behavior in Online and Tradi- tional Supermarkets: The Effects of Brand Name, Price and Other Search Attributes,” International Journal of Re- search in Marketing 17, no. 1 (March 2000): 55-78; and J.Y. Lee and D.R. Bell, “Neighborhood Social Capital and Social Learning for Experience Attributes of Products,” Marketing Science 32, no. 6 (November-December 2013): 960-976.

7. This practice has drawn the ire of retailers. See, for example, D. Coleman, “Showrooming Is the New Shoplifting,” May 24, 2013, www.retailers.com.

8. Showrooming, by some industry estimates, is thought to cost U.S. retailers more than $200 billion. See “Show- rooming: A $217 Billion Problem,” May 2013, www.360pi.com.

9. The Internet retailer as a purveyor of “infinite” product variety was probably first popularized in C. Anderson, “The Long Tail: Why the Future of Business Is Selling Less of More” (New York: Hyperion, 2006). Academic re- search has also shown that Internet shoppers are more likely to buy niche products; see E. Brynjolfsson, Y.J. Hu and M.S. Rahman, “Battle of the Retail Channels: How Product Selection and Geography Drive Cross-Channel Competition,” Management Science 55, no. 11 (Novem- ber 2009): 1755-1765.

10. For the former, see M. Fisher and R. Vaidyanathan, “Which Products Should You Stock?” Harvard Business Review 90, no. 11 (November 2012): 108-118. Aggregat- ing inventory decisions for n independent locations (or, more generally, demand streams) results in lower relative

SLOANREVIEW.MIT.EDU FALL 2014 MIT SLOAN MANAGEMENT REVIEW 53

uncertainty and lower inventory costs, as demonstrated in G.D. Eppen, “Note: Effects of Centralization on Ex- pected Costs in a Multi-Location Newsboy Problem,” Management Science 25, no. 5 (May 1979): 498-501.

11. The cost of holding inventory at a distribution center is lower than the cost of holding inventory at a store. Exclud- ing shipping costs, it is usually substantially less expensive to operate a system with centralized fulfillment than a system that holds inventory in each store.

12. For example, business researchers Antonio Moreno and Christian Terwiesch show that as automotive compa- nies increase the number of products they carry, the supply-demand mismatches increase, and negative effects of uncertainty become more important. When automotive manufacturers extend their product lines, they have to carry more inventories and offer higher discounts. See A. Moreno and C. Terwiesch, “The Effects of Product Line Breadth: Evidence from the Automotive Industry,” April 7, 2013, http://papers.ssrn.com.

13. More recently, some companies have incorporated similar programs in which the product bought online does not need to physically be in the store at the moment of purchase. It is shipped to the store after the customer completes the purchase. We call such programs “buy online, ship to store” (BOSS).

14. S. Gallino and A. Moreno, “Integration of Online and Offline Channels in Retail: The Impact of Sharing Reliable Inventory Availability Information,” Management Science 60, no. 6 (June 2014): 1434-1451.

15. This innovation earned WarbyParker.com the moniker “the Netflix of eyewear.” See D. Wong, “GQ Calls It the Netflix of Eyewear,” Nov. 29, 2010, www.huffingtonpost. com.

16. In experiments, the proper “matching” of treatment and control cities is very important, and in typical cases it is accomplished by random assignment of essentially identi- cal experimental units or subjects to either the treatment or the control condition. Clearly, however, Warby Parker man- agement does not open showrooms in random locations but in locations where they expect to see the greatest ben- efit. Any assessment of showroom impact needs to factor this endogeneity into the analysis. We account for it using propensity scoring. See F. Caro and C. Tang, “The 1st POMS Applied Research Challenge 2014 Awards,” Pro- duction and Operations Management, in press.

17. We also conjecture that when a company can effec- tively deliver customer experiences that lead to early purchases through an offline channel, for example by pro- viding outstanding service in a store, an inventory-only showroom or even a pop-up store, this can allow the com- pany to subsequently retain and service these customers via the online channel — essentially acquiring the customer through either quadrant 1 or quadrant 3 and retaining and maintaining the customer through quadrant 4. We thank Lawrence Lenihan for this observation. Authors’ communi- cation with Lawrence Lenihan, cofounder and managing director, FirstMark Capital, March 27, 2014.

18. Other successful online-first retail players are going into this retail format, including dress and accessory rental company Rent the Runway. See A. Jacobs, “Ready to Strut in Ready-to-Rent,” New York Times, January 22,

2014; and E. Brooke, “Rent The Runway Branches Further Offline, With A Showroom at Henri Bendel,” TechCrunch, October 17, 2013 (see http://techcrunch. com/2013/10/17/rent-the-runway-branches-further-of- fline-with-a-permanent-showroom-at-henri-bendel/).

19. For details and routes, see “Warby Parker Class Trip,” n.d., www.warbyparkerclasstrip.com.

20. See www.thestorefront.com; and J.D. Stein, “No Space Too Small, No Lease Too Short,” New York Times, December 20, 2013.

21. See, for example, J. Choi, D.R. Bell and L.M. Lodish, “Traditional and IS-Enabled Customer Acquisition on the Internet,” Management Science 58, no. 4 (April 2012): 754-769.

22. See M. Halkias, “E-commerce Retailers Open Physi- cal Locations in Dallas to Augment Online Stores,” Dallas News, June 4, 2014.

23. For more details, see S. Jacobs, “Bonobos, an Ecom- merce Darling, Finds an Edge in Brick and Mortar,” May 30, 2014, http://streetfightmag.com.

24. This was popularized in R. Putnam, “Bowling Alone: The Collapse and Revival of American Community” (New York: Simon and Schuster, 2000).

25. Documentation that comes with the SCCBS describes it as the “first attempt at widespread systematic measure- ment of social capital in the United States,” and it has been used extensively in economics. See M.B. Aguilera, “The Impact of Social Capital on Labor Force Participation: Evidence From the 2000 Social Capital Community Bench- mark Survey,” Social Science Quarterly 83, no. 3 (September 2002): 853-874; and C.A.L. Hilber, “New Housing Supply and the Dilution of Social Capital,” Journal of Urban Economics 67, no. 3 (May 2010): 419-437.

26. While not tested directly, the converse is implied by our research as well — that if customer experiences are negative, then online sales will slow down more rapidly in locations with more offline social capital.

27. For details, see Lee and Bell, “Neighborhood Social Capital and Social Learning,” 973.

28. Dunn, “E-Commerce Is a Bear.”

29. Banjo, “Rampant Returns Plague E-Retailers”; and “E-Commerce Returns Are Up,” Fox Business video, 03:21, December 23, 2013, http://video.foxbusiness.com.

30. Metail, headquartered in London, allows female shoppers to create a 3-D model of themselves and evaluate how products would fit them prior to making an online purchase. Other companies providing similar tools to apparel and footwear retailers include PhiSix Fashion Labs, which eBay acquired in February 2014, and Pittsburgh-based Shoefitr.

31. E. Byrnjolfsson, Y. Hu and M. Raman, “Competing in the Age of Omnichannel Retailing,” MIT Sloan Manage- ment Review 54, no. 4 (summer 2013): 23-29.

i. For more details, see D.R. Bell, J. Choi and L. Lodish, “What Matters Most in Internet Retailing,” MIT Sloan Management Review 54, no. 1 (fall 2012): 27-33.

Reprint 56115. Copyright © Massachusetts Institute of Technology, 2014. All rights reserved.

PDFs Reprints Permission to Copy Back Issues

Articles published in MIT Sloan Management Review are copyrighted by the Massachusetts Institute of Technology unless otherwise specified at the end of an article.

MIT Sloan Management Review articles, permissions, and back issues can be purchased on our Web site: sloanreview.mit.edu or you may order through our Business Service Center (9 a.m.-5 p.m. ET) at the phone numbers listed below. Paper reprints are available in quantities of 250 or more.

To reproduce or transmit one or more MIT Sloan Management Review articles by electronic or mechanical means (including photocopying or archiving in any information storage or retrieval system) requires written permission.

To request permission, use our Web site: sloanreview.mit.edu or E-mail: [email protected] Call (US and International):617-253-7170 Fax: 617-258-9739

Posting of full-text SMR articles on publicly accessible Internet sites is prohibited. To obtain permission to post articles on secure and/or password- protected intranet sites, e-mail your request to [email protected]

MITMIT SLSLOOAN MANAAN MANAGEMENGEMENT REVIEWT REVIEWhttp://sloanreview.mit.eduhttp://sloanreview.mit.edumailto:[email protected]mailto:[email protected]http://mitsmr.com/1f69NPF

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

  • 56115Wx.pdf
    • boilerplate.pdf
      • Global
      • Accelerated Innovation: The New Challenge From China
      • Accelerated Innovation: The New Challenge From China
        • The Push to Accelerate Innovation
          • About the Research
          • Industrializing the Innovation Process
          • Pushing the Boundaries of Simultaneous Engineering
          • Cycling Rapidly Through “Launch-Test-Improve”
          • Combining Vertical Hierarchy With Horizontal Flexibility
        • Implications for Global Competition
        • Responding to the New China Challenge
          • Reengineering Established Innovation Processes
          • Focusing R&D Activities on Leveraging Accelerated Innovation Capabilities
          • Exploiting the Potential of Alliances With Chinese Partners
            • About the Authors
            • References

AI ASSISTANTS WILL TRANSFORM HOW COMPANIES AND CUSTOMERS CONNECT. BY NIRAJ DAWAR AND NEIL BENDLE

ILLUSTRATION BY TOMASZ WALENTA

THE AUTONOMOUS CAR dropped Lori at her home and then left for its scheduled service at the dealership. It would be back in time to take her to the airport the next morning. On the way into her house, Lori gathered the drone deliveries from the drop box on her stoop. The familiar voice of Eve, a next-generation smart assistant like Alexa, greeted her in the foyer and gently reminded her of the travel plans for her upcoming conference in LA. Lori hadn’t bothered to learn the details, since Eve had taken care of finding the best flight, seat, and hotel room that her company’s expense policy would allow.

As she unpacked her grocery delivery, Lori saw that Eve had adjusted her weekly purchases, omitting perishables and adding travel-size toiletries and sunblock. Calculating that Lori was running low on detergent (and aware she’d be coming home with laundry to do), the bot had ordered more but switched to a new, less expensive brand that was getting good consumer reviews. And, knowing that Lori wouldn’t want to cook, it had arranged for her favorite takeout to be delivered upon her return.

Thank goodness for Eve, Lori thought to herself. In addition to managing her shopping and travel, the bot tracked her spending and kept her costs

down. Each quarter, for example, Eve checked all the telecommunications plans on the market and compared them against Lori’s projected data usage. Her current plan gave her the best price for her mostly evening and weekend usage, but with her brother’s 40th birthday approaching, Eve had anticipated a lot of data traffic among Lori’s friends and family and found a deal from an upstart firm that would save her money. That offer was instantly matched by Lori’s current provider, a company that had paid to be featured on Eve and to have the right to meet competitors’ prices. Lori relied on Eve for similar help with buying insurance, banking, and investment products, too. Sometimes she had to instruct her bot about her criteria and the trade-offs she was willing to make ( for example, to forgo higher returns for a greener investment portfolio), but more recently, Eve had begun figuring out what product attributes she was after—even aesthetic ones—without having to be told.

Lori didn’t know how she had ever coped without Eve. She had come to trust the bot not just for advice on complex purchases but also to make many of her routine decisions and to introduce her to new products and services she didn’t even know she wanted.

THE AGE OF ALEXA MARKETING IN

FEATURE MARKETING IN THE AGE OF ALEXA

80  HARVARD BUSINESS REVIEW MAY–JUNE 2018

MAY–JUNE 2018 HARVARD BUSINESS REVIEW 81 

DOES THIS SCENARIO SOUND FAR-FETCHED? It isn’t: All the technologies that Lori uses to interact with her world are either currently in development or already available—and being rapidly refined. Amazon, Google, Baidu, and other tech giants have launched ar- tificial intelligence platforms with increasingly skilled digital assistants. While none have yet attained Eve’s sweeping capabilities, that is clearly their goal—and it’s just a matter of time before they get there.

AI assistants are rapidly colonizing consumers’ homes. Analysts estimate that Amazon, for instance, has sold some 25 million Echo smart speakers, which people use to engage with its AI assistant, Alexa, and that number is expected to more than double by 2020. Once you take into account the millions of other devices that can already host Alexa through iOS or Android apps, Alexa’s market penetration looks even higher.

Google Assistant, accessed chiefly through Google Home cylinders and Pixel phones, is now available on 400 million devices. Earlier this year Apple launched a Siri-enabled HomePod, and Samsung has acquired Viv, an intelligent assistant company founded by Siri’s creators, to bolster its Bixby AI assistant plat- form. Microsoft and Tencent have platforms for their own AI assistants (Cortana and Xiaowei), and virtual assistants Chumenwenwen and Xiaoice (which is capable of uncannily human conversations and re- portedly has 40 million registered users) are already popular in China.

Over the next decade, as these firms and others fight to establish the preferred consumer AI platform, AI assistants will transform how companies connect with their customers. They’ll become the primary channel through which people get information, goods, and services, and marketing will turn into a battle for their attention.

AI assistants will help consumers navigate their increasingly overwhelming number of choices. Every year people buy from thousands of product catego- ries, deciding among dozens or hundreds of options in each. Even routine purchases can be time-consuming; nonroutine purchases often require sorting through the nuances of competing offers and are fraught with risk. While shopping for shoes may be fun, picking the right toothbrush from more than 200 products is pretty tedious. Choosing the wrong tennis racket can ruin your game, and buying an ill-considered cell phone plan or insurance policy can be costly.

AI assistants will not only minimize costs and risks for consumers but also offer them unprecedented convenience. They’ll ensure that routine purchases flow uninterrupted to households—just as water and electricity do now—and manage the complexity of more-involved shopping decisions by learning con- sumers’ criteria and optimizing whatever trade-offs people are willing to make (such as a higher price for more sustainability).

The effects on the business landscape will be far-reaching. Technologies that revolutionize the way consumers interact with a marketplace also tend to re- configure its dynamics and reshape the companies that sell into it. In the 1950s, for instance, the rise of super- markets made scale and mass media much more im- portant to marketers, triggering a wave of consolidation among consumer goods companies. AI platforms and assistants will likewise change the game for brands and retailers, altering the relative power of players in the value chain and the underlying basis of competition.

These predictions grow out of our ongoing re- search into the ways technology has been redefining relationships among customers, brands, and firms. In the course of it, we have reviewed hundreds of rele- vant academic, industry, and news articles, and held in-depth discussions and structured interviews with industry experts and executives at Google, L’Oréal, EURid, and other global businesses. (Ivey Business School graduate student assistants Gobind deep Singh and Vivek Astvansh helped us with the early literature reviews.) In this article we’ll outline in more detail the near-term changes we expect AI platforms to bring about and explain the implications they hold for marketing strategy.

MARKETING ON PLATFORMS Once the dust settles, we expect that just a handful of general-purpose AI platforms will be left standing. (See the sidebar “The Coming Platform Shakeout.”) Most consumers will use only one, whose assistant will be incorporated into their homes, cars, and mo- bile devices. The platform will gather and deliver in- formation, and the assistant will be the consumer’s interface with home systems, appliances, and other machines. The assistant will also be the portal to an infinite shopping mall of goods and services. The more consumers use a platform, the better it will un- derstand their habits and preferences, and the better it will meet their needs—increasing their satisfaction in a self-reinforcing cycle.

IN BRIEF

THE NEW ENVIRONMENT Over the next decade, smart assistants like Alexa will transform how companies sell to and satisfy consumers, and global firms will battle to establish the preferred artificial intelligence platform.

THE CHANGING BEHAVIOR AI assistants will become trusted advisers to consumers, anticipating and satisfying their needs, ensuring that routine purchases flow uninterrupted to their households like electricity, and guiding them through complex buying decisions.

THE STRATEGIC RESPONSE Brands will need to shift the focus of their marketing from consumers to AI platforms, seeking to influence platforms in order to get preferential positioning on AI assistants.

82  HARVARD BUSINESS REVIEW MAY–JUNE 2018

FEATURE MARKETING IN THE AGE OF ALEXA

WILL BRANDS MATTER? Thanks to AI platforms, the job of branded-goods companies is about to get much harder. Increasingly, AI assistants like Alexa will control access to those firms’ customers, and brand recognition will play less of a role in product selection than dynamic and idiosyncratic AI algorithms will. That doesn’t mean, though, that brands will no longer matter. They can respond in three ways:

First, they must invest aggressively in understanding the algorithms platforms use to recommend and choose products, including how they weight each brand for each consumer. In some categories and for some consumers, brands may be more important than price (Apple is an example). In others (say, toothbrushes), brands may be less relevant. AI algorithms will take such differences into consideration.

Second, brands should assess the value of maintaining direct ties with consumers. For some kinds of offerings, such as smart, connected consumer electronics, promoting brand awareness and loyalty outside AI platforms may be a good strategy. Smart products give companies a direct channel for communicating with customers and collecting data on them, which may make those companies less reliant on AI platforms. In such cases, ongoing investments in brand building will make sense.

Finally, while consumers are increasingly buying online, most purchases—currently about 90% of global retail sales—occur in brick-and-mortar stores. For the foreseeable future, consumers will continue to shop offline, where brands will remain influential. As consumers’ purchasing shifts to AI platforms, brands should regularly evaluate how important the physical retail channels remain (that will vary widely by category) and adjust their strategy accordingly. Brands will still be the experts in the product categories in which they operate, with deep knowledge about consumer behavior and product innovation.

Marketers’ current obsession with creating an omnichannel customer experience will fade as AI platforms become a powerful marketing medium, sales and distribution channel, and fulfillment and service center—all rolled into one. The concentration of those functions within a few platforms will give their owners enormous clout, and branded products will find themselves in a weaker position. Consumer companies that feel that large retailers like Walmart wield too much power today will be even more alarmed by the might of AI platforms. As a major—or even primary—means of communicating with con- sumers, and the repository of reams of data about their habits, preferences, and consumption, the platforms will have a lot of influence over prices and promotions and the consumer relationship itself.

Brands today owe their success to their ability to sig- nal quality and win buyers’ loyalty. But in a world of AI platforms, marketers may find that consumers like Lori shift their allegiance from trusted brands to a trusted AI assistant. The activities that help brands cement relationships with buyers over time—understanding and filling people’s needs, assuring quality, and consis- tently putting consumers’ interests at the center—will in many cases be performed better by AI. In fact, we predict that AI assistants will win consumers’ trust and loyalty better than any previous marketing technology. We therefore expect the focus of many brands to shift from reinforcing direct relationships with consumers to optimizing their positions on AI platforms. However, in selected cases it may still make strategic sense for brands to maintain strong ties with consumers outside the platforms. (See the sidebar “Will Brands Matter?”)

These changes will have an impact on companies at three critical levels: customer acquisition, satisfaction, and retention.

ACQUISITION With consumer data now being used to create finely targeted marketing, customer acquisition has become ever more efficient. Still, marketers’ aim is far from perfect. Ads continue to be directed at consumers who aren’t good prospects—and don’t reach everyone who may be interested in an offering. Even when an ad does find the right audience, its message is often blunted by consumers’ cognitive limitations: People might need to see the ad many times before it registers or may forget it entirely. They may remember only the parts that interest them (for example, the humor) but not the product’s name or distinctive promise.

Those problems will matter less in the coming years, when the main target of the billions in annual spend- ing on brand marketing will shift from forgetful, biased consumers to AI platforms that retain every last bit of information. Platforms will analyze that data, taking into account products’ pricing, characteristics, past performance, and reviews (weighted by authenticity and relevance) and the consumers’ preferences and past behavior. Customer acquisition will become even more of a science and will focus on a single channel— the platform—rather than on multiple channels.

In this universe, influencing the platforms’ algo- rithms will be the key to winning. It will be crucial for companies to understand the customized purchasing criteria that the AI applies on behalf of each consumer. Sellers will probably have to pay platforms to get that

CONSUMERS WILL SHIFT THEIR ALLEGIANCE FROM TRUSTED BRANDS TO A TRUSTED AI ASSISTANT.

MAY–JUNE 2018 HARVARD BUSINESS REVIEW 83 

information—and to be “listed” on them, in much the same way that brands now pay shelving fees to brick- and-mortar retailers. Marketers can also expect to bid on or otherwise pay extra for preferential positions, just as hotels today bid to appear at the top of the re- sults on Expedia, or marketers compete in Google’s AdWords auctions to come up first in searches. Though Amazon says it has no plans to add advertis- ing to Alexa, CNBC has reported that the company is talking with several consumer goods firms about pro- moting their products on the platform. Experiments “in the works,” said CNBC, would allow Alexa to make product recommendations based on a user’s previous inquiries (“How do I clean grass stains?”) or past shop- ping behavior. We believe that product placement and recommendations on AI platforms are inevitable and will, in time, be a major source of revenue.

In different ways, all these payments will be for access to the consumer. Companies will essentially reallocate to the platforms what they spend today on advertising, listing and slotting fees, and retail com- missions. Brands will shape their offers and innovation strategies around getting AI assistants to showcase their products.

The bustling ecosystem that now helps compa- nies woo customers, including ad agencies and me- dia buyers, will need to learn to market through AI platforms. Marketing services that target platforms will be even more accountable than media buyers are today and will need to show links to actual consumer behavior. Traditional market research may be sup- planted altogether by the intelligence about consum- ers’ actual behavior that brands will be able to buy directly from platforms.

SATISFACTION Customer satisfaction drives loyalty, word of mouth, market share, and profitability. No wonder marketers are fixated on monitoring it. Imagine, then, a world where reliable satisfaction data is easier to get from AI platforms than from consumers themselves.

A platform serves consumers by constantly antici- pating their needs. To do that it must collect granular data on their purchasing patterns and product use and try to understand their goals: Do they want food products to improve their health, energy products to minimize their environmental impact, and finan- cial products to increase their long-term returns? Or are their criteria taste, price, and short-term perfor- mance? Sophisticated AI platforms will go further and figure out the trade-offs consumers are willing to make: How much more will they pay for a more healthful product? How much room in a car will they

sacrifice to get better fuel efficiency? AI platforms will even know whether consumers are likely to adapt their requirements in different contexts—for exam- ple, if a person on a diet will make an exception for dessert when celebrating.

Because of all this, AI platforms will be able to pre- dict what combination of features, price, and perfor- mance is most appealing to someone at a given mo- ment. Ultimately, AI assistants may be able to satisfy customers’ needs better than the customers them- selves can. Relatively primitive recommendation en- gines are already moving in this direction, suggest- ing books, movies, and music that consumers didn’t know they would enjoy.

AI platforms will lead to more-efficient sorting and matching in the marketplace. Consumers who prefer the Four Seasons, for instance, will be unlikely to be of- fered reservations at a Trump hotel by their platforms. So brands will want to sharpen their positioning in ways that the platforms will register.

RETENTION Marketers assume that repeat purchases indicate customer satisfaction and are a sign of brand loy- alty. Yet many customers keep buying a product not because it delights them but because they can’t be bothered to explore alternatives if a brand is perform- ing adequately. Put simply, most people have better things to do than evaluate the ingredients of laun- dry detergents. An AI assistant, however, does not. It can regularly reassess all brands in any category, whether laptops or chewing gum, and recommend a new one that might serve the consumer better. Some consumers may even like to switch things up just for the sake of variety—so their assistants, being aware of that, will periodically recommend new products they might like.

That routine reevaluation of purchases will force incumbent brands to constantly justify their positions. But it will also open opportunities for challengers. Competition will get more intense.

Though incumbents will need to innovate to hang on to customers, they’ll be able to buy information from platforms that will help them inhibit brand switching. If a brand knows that a consumer is likely to defect (say, because she has indicated a desire for change to her assistant), it can compute retention metrics in real time to see whether she’s worth keep- ing. If she is, the brand can make her a customized offer that reflects exactly what she needs to stay put. If the consumer accepts it, both she and the brand win: The brand keeps the business and the consumer gets a better deal. The AI platform is in the middle,

CUSTOMER ACQUISITION WILL BECOME EVEN MORE OF A SCIENCE AND WILL FOCUS ON A SINGLE CHANNEL— THE AI PLATFORM.

84  HARVARD BUSINESS REVIEW MAY–JUNE 2018

FEATURE MARKETING IN THE AGE OF ALEXA

serving both in ways that create value for each while generating revenue for itself.

On their part, challenger brands c an use in- telligence from a platform to acquire customers. Promotions through AI assistants will be the tool of choice for upstarts. Of course, once a challenger breaks in, it will be subject to threats from the incum- bent and other rivals. The secret to competitive differ- entiation—and, hence, retention—will be constantly designing offers that meet a customer’s evolving cri- teria. For brands, this will become as much a focus of innovation as developing better products is.

THE IMPERATIVES FOR PLATFORMS AI platforms will succeed only if consumers have faith in them. As one platform leader at Google told us, “Building trust will be the most important thing we do.” To earn consumers’ confidence, platforms must ensure three things: accuracy, alignment, and privacy.

ACCURACY By continually learning each individual’s desires and requirements, the platform algorithms will hone their ability to please consumers. If a platform can recommend an alternative to a trusted brand that it thinks the consumer will like better, and the con- sumer is happier with the alternative, that platform will supplant the brand as the object of her trust.

ALIGNMENT There’s a built-in conflict of interest that platforms must manage carefully. On one hand, they must sin- gle-mindedly focus on meeting consumers’ needs; if they fall short, it will erode trust. On the other hand, they’ll have contractual arrangements to provide preferred placements and consumer data to brands. If people sense that an assistant is pushing a paying brand that isn’t aligned with their needs, that too will undermine trust. One solution might be for plat- forms to be transparent about their relationships with brands, just as Google is today when it identifies some search results as ads. Another may be to give paid and unpaid recommendations equal weight; if a consumer asks an AI assistant how best to remove grass stains, the response might include both a paying bleach and a comment that generic bleaches can be just as effective. The brand gets its plug, and the assistant demonstrates that it’s trustworthy.

THE COMING PLATFORM SHAKEOUT Today there are perhaps a dozen serious contenders in the nascent AI platform industry. But we expect that this field will eventually be narrowed down to only a few. What will drive this concentration, and how will winners be chosen?

For starters, the market has high barriers to entry. Large general-purpose AI platforms are extremely expensive to build and run. It took thousands of engineers several years to develop Amazon’s Alexa, for instance. Besides committing vast internal resources to development, each player must establish an extensive ecosystem of providers that supply data, services, skills, and apps. To succeed, platforms need a large installed base and a wide range of capabilities. Those that achieve scale and scope will have a natural advantage: The more a platform can do reliably and well, the more loyal users will be to it. Over time it will learn consumers’ preferences and habits, which will make it even better at anticipating and satisfying people’s needs, which will make consumers use it more. Those dynamics, combined with a lack of data portability across platforms, will make AI platforms sticky. Advantages will accrue mostly to just a few large platforms. While smaller platforms such as Uber’s or Expedia’s may coexist for a time, we expect they’ll ultimately be incorporated into the large general platforms as vendors or as specific AI assistant skills.

BRAND CONSUMER

AI PLATFORM

PLATFORM PROVIDES BRAND • Virtual shelf space • A single channel for

marketing, sales, and service • Data on consumer

preferences, purchases, and media exposure

• Payment for goods sold • Product fulfillment • A trust halo

BRAND PROVIDES PLATFORM • Listing and promotional fees • Product information • Innovations tailored to

consumers’ needs • Knowledge about

product category

PLATFORM PROVIDES CONSUMER • Customized

recommendations • Automated routine purchases • Convenience and savings • Reduced complexity • Continual scanning for

better deals

CONSUMER PROVIDES PLATFORM • Payment for goods • Information on product

preferences, purchases, and use

• Information on price sensitivity, risk tolerance, and privacy expectations

• Loyalty in exchange for good recommendations

HOW AI PLATFORMS CREATE VALUE

MAY–JUNE 2018 HARVARD BUSINESS REVIEW 85 

PRIVACY Platform owners, as well as marketers, will need to strike a careful balance between the use of personal information and AI performance. The more data gath- ered, the more accurate the platform—but the more exposed consumers may feel. A solution here could be to offer customized privacy settings, as Facebook now does, giving users control over what information they share and how widely. Another, less satisfactory solu- tion might be to argue, as Google sometimes does, that privacy is protected because consumer data is handled by machines without human intervention.

Consumers have long been willing to give up per- sonal information, and even privacy, for convenience. AI assistants will offer much greater convenience but be far more intuitive and intrusive than any software now in use, greatly magnifying the trade-offs.

ALL CONSUMER-FACING FIRMS can expect AI platforms to radically alter their relationships with customers. Their traditionally crucial assets, such as manufactur- ing capability and brands, will become less central as consumers’ attention shifts to AI assistants, and the value of consumer data and AI’s predictive ability soar. Push marketing (getting platforms to carry and pro- mote a product) will become more important, while pull marketing (persuading consumers to seek prod- ucts) becomes less so. The consumer will remain the target of brand-building efforts, but marketing that encourages trial and repeat purchases will be more ef- fective when aimed at AI. Though the marketplace will be more efficient, companies will encounter intense pressure to offer consumers the best deal—the one most closely aligned with the preferences identified by AI gatekeepers.

For a long time, consumer goods companies, used to maximizing economies of scale because of their large fixed investments in production and brands, have zeroed in on one strategic question: How much more of our product can we sell? AI platforms will pre sent a very different opportunity: to maximize the depth of the relationship with the consumer by offering a wide range of products—in other words, economies of scope. Investments in building trust with consumers and their AI assistants will be amortized by asking, What else does this buyer need? Superior marketing strategy will still matter—firms must acquire, satisfy, and retain con- sumers in the AI world—but what it involves is likely to change substantially. HBR Reprint R1803E

NIRAJ DAWAR (Twitter: @nirajdawar) is a professor of marketing at the Ivey Business School and the author of

Tilt: Shifting Your Strategy from Products to Customers (Harvard Business Review Press, 2013). NEIL BENDLE is an associate professor of marketing at the Ivey Business School and a coauthor of Marketing Metrics: The Manager’s Guide to Measuring Marketing Performance, 3rd edition (Pearson FT Press, 2015).

THREE QUESTIONS FOR BRANDS Whom is the platform working for? Before answering this question, let’s apply it to traditional platforms. Consider credit card companies and brick-and-mortar retailers: Both perform functions—providing convenience, efficiency, and risk reduction—for the buyers and sellers they connect. AI platforms likewise work for multiple stakeholders, including brands. But bear in mind that if they don’t serve the interests of the consumer, they won’t be adopted. And the more consumers trust and rely on them, the more effective they are as a source of data and a channel for marketers. As with any well- functioning platform, creating value for parties on either side generates value for the platform itself.

What do we want from the platform? The obvious but incomplete answer is, we want it to sell our products. However, at the outset marketers should not think of a platform principally as a sales channel; they should look at it as a source of information. For a price, AI platforms will offer a view of consumer behavior and motivations more detailed than anything that’s ever been available before. That nuanced understanding will allow companies to redesign every aspect of marketing— from segmentation to pricing to product features and promotional offers—to better meet consumers’ needs. Platforms in turn will promote the improved products—and become the superior sales channel marketers seek.

How can we make sure the platform chooses us? Here, brands will have two levers. One will be to pay for preferential positioning; the other, and likely the most powerful, will be to continually innovate their offerings so that they align with customers’ stated and implicit needs, drawing on data supplied by the platforms. This will require brands to sharpen their differentiation; hone their ability to compete on speed, quality, and cost; and recognize and respond to rapid or subtle shifts in consumer tastes.

3

2

1

86  HARVARD BUSINESS REVIEW MAY–JUNE 2018

FEATURE MARKETING IN THE AGE OF ALEXA

Copyright 2018 Harvard Business Publishing. All Rights Reserved. Additional restrictions may apply including the use of this content as assigned course material. Please consult your institution’s librarian about any restrictions that might apply under the license with your institution. For more information and teaching resources from Harvard Business Publishing including Harvard Business School Cases, eLearning products, and business simulations please visit hbsp.harvard.edu.

Order your essay today and save 10% with the discount code ESSAYHELP