2. Shopper Behavior Measurement Methods
METHODOLOGY DESCRIPTION ADVANTAGES LIMITATIONS COMMENTS
PREVALENT Shoppers surrender basic n Historical
sales data, segmented n Retailer-owneddata means Retailer-to-retailer comparisons can be
IN-MARKET personal information by shopper type permission-only availability for difficult. Shoppers who frequent more than
DATA: to obtain a retailer- n Records some behaviors, brands one retailer aren’t tracked cross-channel.
issued card, which is including: purchase frequency, n Hard to compare data across It is unclear whether each retailer’s sample
Shopper Card scanned at checkout to represents unique shoppers; data doesn’t
Data units/trip, promotional response, retailers
unlock discounts and price response, brand switching, reveal who went across town to buy the item.
n No qualitative brand
promotional offers. cross-basket data, demographic Aggregated shopper card data gets
Sales and promotional information (equity measures around this by linking multiple data sets,
segmentation data, prior over time, campaign
responses recorded purchase behavior, ZIP code but can’t then be segmented – in other
longitudinally among communication diagnostics, etc.) words, marketers can get the information by
segmentation
shoppers. retailer or by shopper, but not by both. (The
exception is cases where retailers agree to
release data for the purpose of aggregating,
using a research vendor intermediary.)
PREVALENT Trained interviewers n Flexibility of survey questions n Costly
(depending on The questions that shopper marketers
IN-MARKET intercept shoppers, pre/ n Identifies the “why behind the buy” penetration level) need to ask are hard to answer. They
TESTING: post-shop and/or in-aisle, n Time-consuming often require both observation of shopper
n Cross-channel data
to observe behavior and behavior and auditing of program
Shopper Intercept conduct interviews. n Can
n Intrusive (could lead to social execution. (For example, “How shoppable
yield both behavior and
Interviews and In- brand equity measures bias) is the aisle?” or “What is on the shopper’s
Aisle Observation
2
n Married
n Aligning
purchase data can be list?”) One advantage to this technique is
with focus groups and
cumbersome that it can address a range of such questions.
in-home interviews, can provide a
However, it can be expensive to recruit a
limited-sample, full-path picture
large enough sample to read quantitatively.
PREVALENT Cameras installed n Passiveobservation doesn’t n Retailers,
if they invest in the The method reports conversion versus
IN-MARKET in the natural store interfere with the shopping trip video set-up, own the data opportunity. It gives an “apples to apples”
DATA: environment passively n Robust data: day-parts, cross-aisle, n Limitednumber of stores look across programs, but doesn’t account
record activity without etc. (currently) implies limited samples for creating predisposition through pre-
In-Aisle Video intruding on shoppers; shop activities.
Observation n Matched control stores add to and restricted geographies
an algorithm then parses The observation technique includes
traffic and conversion metrics n Limited availability across all day-parts, all kinds of shoppers. Less
data. n Algorithms save money and time retailers (currently) expensive than labor-intensive interviewing.
n Quick reporting available
COMMON Tracked shopper cart n Passive
observation technique n Expense There is a richness of data here that is
IN-MARKET movement throughout for shopping behavior n Time most often used to gather insights during
TESTING: store identifies potential n Hugedatabase of in-store traffic the program development phase. This
n Scope
buyers in-aisle and observations information effectively helps to refine in-
Shopping Cart conversion rates. store tactics by location, message and visual
Tracking Wearable devices record n Incorporates real-world stimuli vehicle.
and Eye-Tracking eye movement in real or n Can
measure stopping power, To supplement the quantitative data,
simulated stores to record awareness many studies include interview overlays to
awareness of and reaction gauge impact on brand equity and/or get
to marketing stimuli. the “why behind the buy.”
continued on page 3
3. Shopper Behavior Measurement Methods continued
METHODOLOGY DESCRIPTION ADVANTAGES LIMITATIONS COMMENTS
NEW IN-MARKET Qualified respondents n Real-time data gathering n Data quality issues This method is similar to the traditional
TESTING: are equipped with n Quick data output n Respondents are at least partly shopper intercept technique, but places
an interactive mobile responsible for quality assurance the onus of administering the survey
Self- n 360-degree experience in and
platform to record on the shopper. The advantages of
Administered, their own behavior, out of store n Frequent panel rotation means real-time reporting from a 360-degree
Smartphone- experiences and n Can limited data over time shopper experience make it an exciting
record steps all along the
Based Shopper attitudes, and respond to new methodology. However, relying
path to purchase n Limited sample size
Surveys survey questions and live, on respondents for quality control is an
n Cross-channel data
on-the-spot interviews inherent disadvantage.
NEW IN-MARKET Shoppers perform the n Sales
tied to promotional n Just
like shopper card data, the Puts shopper card data on steroids by
DATA: function of scanning response instantly and accurately information is retailer-owned. including in-store location and real-time
their purchases as they n Speedy reporting n Respondents are partly information.
Shopper-Operated
shop (by scanner or responsible for data quality This methodology is intriguing because
In-Store Scanners mobile app); may be n Questions may be asked during it opens the door to incorporating other
fed specific offers based the trip in a time-, location- or n Participating
shoppers may not behaviors outside of the store, such as
on location or purchase behavior-sensitive way be in the marketer’s target digital list-making and social networking.
history. n In-store
trip patterns can be n Currently few participating The shoppers who are currently sampled
Marries shopper card recorded and linked to sales retailers, making data limited may not represent the total population
data with location and and promotional response and n Like
well. If hand-held scanners (or more
shopper card data, the
time information, as shopper card history likely apps) become the “new normal” in
information is retailer-specific (or
3
well as some behavioral shopping, this data may become a critical
aggregated and blind)
reactions. tool for marketing success.
PRE-PROGRAM METHODS
PRE-LAUNCH Tests in a central location n Genuine stopping power and n Expense Used to best effect during the program
TESTING: using technology that awareness are measured with n Time development phase for insights, decision
records eye-movement accuracy support, message refinement and
Virtual Store n Scope
across stimuli, including n Awareness
tied to diagnostics on benchmarking.
Shopping a virtual in-store To supplement the quantitative data,
and Eye-Tracking communication and brand equity
environment. many studies include interview overlays for
n Measured behavior in response gauging impact on brand equity and to get
to stimuli the “why behind the buy.”
n Testvs. control cells readily
available
n Quick reporting
PRE-LAUNCH “Mocked up” aisles n Fasttesting window and fast n Limitedsample size is best suited This method usually is employed pre-
TESTING: simulate a retail results to qualitative studies execution to refine messaging and tactics.
environment. Subjects n Reasonablecost (depending on n Not “real world” Trained moderators typically interview the
Mock Store
are interviewed depth and scope) shoppers while marketers observe. Not
Environment individually pre/post- a tool for predicting full-scale, in-market
Observation and shop and observed while n Allowsfor team members to program performance.
Interviews performing shopping observe.
tasks.
4. SPECIAL REPORT
Softness Worth Sharing
Measurement bucket Results Measurement tools
Sales n Reversed a sales decline and sparked a 3% increase Scan data
Behavior n Activated more than 360,000 brand advocates: Google Analytics
r 361,800 boxes of tissues shared
r 720,000 samples distributed
n Engaged more than 72,000 Facebook users
B2B relationship n Reached shoppers through retailer email lists Internal reporting
n Brand co-hosted in-store clinics with retailers
Source: Shopper Marketing 5.0: Creating Value with Shopper Solutions” (Booz & Co./GMA, 2011)
integrated program, with social media, ex- floor graphics, packaging, FSIs, digital, televi- always the darling of brand managers. In a
periential marketing and in-store elements, sion, print, and out-of-home. All these ve- spring cleaning event, for example, many
among others,” McCallum explains. hicles work together to push shoppers along household cleaner brands may be promoted
The campaign worked neatly as “shopper the path to purchase, influencing behavior simultaneously. Communication and activa-
marketing” because it activated Kleenex’s and attitudes along the way. tion emphasize the group of products, rather
brand equity – the emotional benefit of com- Shoppers accept the Challenge by sub- than any individual brand. So brand man-
fort and caring – by incenting shoppers to stituting a bowl of Special K for two meals agers wonder if they get their “fair share”
“share” a box of tissues with a loved one and every day for two weeks. The promise is of the event, and also what it does for their
thereby earn one for themselves. Shoppers that, by doing so, they’ll drop one jeans individual performance – especially when it
visited Kleenex.com or the brand’s Facebook size. A paramount business objective for comes to shopper behavior.
page to send a free box as a gift as Kleenex the brand is to create demand for meal oc- While these questions are commonplace,
used a sampling program to activate brand casions beyond breakfast. The global cam- scale events are a business necessity. Market-
advocates. paign also seeks to change behavior both ers taking a shopper-centric approach will
The program didn’t just stop a sales de- in-store (stock-ups on Special K to complete often need to develop broader programs
cline, it grew the brand by more than 3% the Challenge) and beyond. that provide holistic solutions. These events
nationally and increased category sales as also benefit the entire company when brand
well. One retail customer blitzed its shoppers Measuring Scale Programs objectives are developed in the context of
with emails and in-store clinics, which result- In addition to measuring their own cam- portfolio growth. Moreover, retailers are fond
ed in sales and returns surpassing objectives paigns, brands need to measure the effect of scale events because they deliver higher
by double digits, according to a case study of “scale” programs, which involve multiple basket rings, pulling more shoppers into more
in Shopper Marketing 5.0: Creating Value brands in a company’s portfolio and aren’t categories across the store.
with Shopper Solutions
(Booz & Co./GMA, 2011).
Another prime example
of building brand equity to
drive sales is the “Special
K Challenge” that Kellogg “ e need to measure the
W
Co. originally launched in
2004. The “Challenge” ac- effectiveness of campaigns in
tivates shoppers by mak-
ing the brand promise of totality, rather than the impact
weight control a reality
through a two-week diet. of an individual element.”
Communications typically
Scott McCallum, OgilvyAction
encompass various me-
dia, including shelf talkers,
4
5. SPECIAL REPORT
‘Cars 2’
Measurement bucket Results Measurement tools
Sales All brands increased share in Q2 vs. the previous quarter Nielsen scan data
(average gain was 7%, compared with a 5% goal)
Behavior Shoppers who were aware of the promotion increased n Pre/post-shop interviews
their K-C share of basket by 40% among baby/child care
n Shopper card data, includ-
brands and more than 100% among family care brands
ing cross-basket information
B2B relationship 25% better display support and 38% better feature Nielsen feature and display
support versus prior solo events. reports
Source: Kimberly-Clark
Interestingly, shoppers were activated in program is finalized.
various ways, boosting performance for each There is an important difference in examin-
brand uniquely. ing the impact on shopper behavior before
n For Kleenex, the event acted as reminder or after execution. Using a methodology to
advertising, driving impulse purchases. assess behavior prior to in-market execution
n For paper brands (Cottonelle, Scott, Viva can identify optimal tactics. However, un-
and Kleenex), the event acted as a trial derstanding a campaign’s impact while in
driver. market can require a different methodology.
n The event was most effective as a market- Real-world conditions are messier and less
ing platform for the baby brands (Huggies, predictable than controlled testing environ-
Pull-Ups), with the Disney connection pro- ments. The fundamental goal of post-execu-
viding an emotional driver for moms. tion measurement is to discover what really
happened, not how to refine a message. (An
Measurement Approaches exception here is digital marketing, where in-
While the Kimberly-Clark program used pre/ market test cells can be employed to change
post-shop interviews and shopper card data tactics “on the fly” and maximize results.)
to gauge behavior, there are numerous other
techniques available. The collective wisdom Choosing Methodologies
Rachael Norton, vice president of shop- of the industry advises that the primary key Certain attributes make a methodology at-
per marketing at ConAgra Foods, explains to success is gaining consensus on the proper tractive and practical for measuring the per-
further: “There are times when brands need metrics up front. formance of an in-market program. Inter-
to serve occasion-based, in-store platforms, Methodologies exist for measuring shop- views with industry executives yielded these
such as holiday, tailgating, and so forth. Each per behavior all along the path to purchase. four common considerations to use as criteria
brand is called to support the portfolio and Some of the more noteworthy methodolo- for choosing the right method:
the retailer. This is ‘scale’ multi-brand pro- gies are reviewed in the accompanying chart Expense. Obviously, measurement costs
gramming, but supporting the retailer very (see pages 2-3). need to be in line with program profitability.
specifically.” (Note: Measuring the effective- While many of these measurement tech- The cheapest ways to track shopper behavior
ness of retailer-driven scale programs will be niques yield rich results, their processes can are also the ones closest to simple sales trans-
covered in article five.) be time-consuming and expensive. In the action data: shopper card information. When
Driving shopper behavior through a scale words of one blunt, brand-side shopper mar- looked at longitudinally, as if it were a behav-
marketing event is a challenge, but it can be keter, “I don’t have the time to use these, and ioral panel, certain metrics can be gleaned
done. I don’t have the money either.” Therefore, such as cross-category purchasing, switch-
In spring 2011, Kimberly-Clark launched a most are not feasible for smaller programs. ing or promotional response among certain
multi-brand tie-in to Walt Disney’s “Cars 2” As a practical matter, the more elaborate shopper segments (like men or moms). This
theatrical release. Beyond sales increases, the methodologies are better suited for larger ef- data can answer behavioral questions like
objectives included increasing K-C’s share of forts during the development phase to glean “What else did they buy?” “Were men mo-
basket. The campaign successfully drove sales insights (Where are the shoppers’ eyes look- tivated to repeat purchase?” or “Were new
and shopper behavior, as well as improved ing?) or for decision support (Which is more users brought into the category?”
business-to-business results (see chart, above). effective, version A or version B?) before the But even for this relatively simple, passive
5
6. SPECIAL REPORT
data, there is a hurdle: retailer consent. There ing” could alter the reality of what was be- stimuli specifically helped drive behavior.
also is a constraint: uneven availability across ing observed.) So in-aisle interviews, which Also, the effectiveness of in-store marketing
the market. Despite these potential drawbacks, can be terrific for diagnostics, aren’t always efforts can be accurately judged, because
the behavioral objectives of many simpler pro- the best for recording behavior. Unobtrusive store-level execution elements are recorded
grams can be addressed cost-effectively with technological observation is the more accu- alongside the purchase behaviors.
shopper card data. rate approach, but is expensive and not yet Single-source methodology would let mar-
Timeliness. Shopper tracking must occur widespread. However, either human or tech- keters track the impact of their efforts on
at intervals that coincide with the program. nological observation trump the “recalled” shopper behavior, on a retailer-specific basis,
The Nielsen Panel, for example, has robust, behavior provided by shoppers asked to re- over any length of time. While this will not
market-wide behavioral data, but does not member what they did. eliminate the need for financial analysis or
capture it frequently enough to measure the Grappling with these issues is challenging; qualitative pre/post-shop interviews, it cer-
impact of faster-cycle programs. (Also, the the issues and priorities shift according to pro- tainly would supply a fact-based grounding
data can’t be read at the retailer-specific level.) gram objectives and size. This can make the upon which a solid understanding of shopper
The gold standard in terms of ideal expec- process seem haphazard, compared to the behavior can be built.
tations is “real time” collection and reporting. relatively routine business of media buying,
Today, programs that come closest to real financial planning or even measuring digital
time include in-store video monitoring and behavior.
mobile-enabled shopper panels. For the mo- But there is hope on the horizon: “single About the Author
ment, however, such tools have limited reach source” data, which refers to a thread of re-
and sample sizes, which leads to the next corded behavior tied to a specific shopper.
consideration. Single source data is emerging as a promising
Sample Size. Scope can be an issue. It new approach, because it marries purchase
would be ideal to get a complete census of all behavior to media consumption. It works
shopper behavior. Unfortunately, this doesn’t roughly this way: purchase behavior is tracked
exist as of now. Marketers therefore rely on using scanner data derived from credit/debit
samples, which need to be projectable. While card transactions. Each shopper is assigned
aggregated shopper card data is available, it an anonymous tag, which is then matched
can’t be broken out by retailer. As a practical to online activity by a company that connects
matter, pre/post-shop interviews are still one a “surfer” to a “shopper” to identify a “sin-
of the fastest, cheapest ways to get informa- gle source.” In a final step, certain volunteer
tion by retailer, by program. households have their TV viewing monitored.
Liz Crawford has more than 20 years
Accuracy. When researchers interfere too The tags are then married to these house-
of brand management and consulting
actively, it can change results. (Theoretical holds, completing a single-source, in-store/
experience with a concentration in
physicist Warner Heisenberg realized as far out-of-store tracking system.
strategic innovation. Over the last
back as 1927 that the very act of “observ- Data obtained through other sources (such
few years, Crawford has focused
as exposure to in-store messaging via audits)
on developing integrated shopper
can be incorporated into these behavioral
marketing strategies for Fortune 500
Series Schedule findings, too. Smartphones soon may also
clients. Currently, Crawford is an
feed the data stream, delivering both Internet
analyst and contributing writer for the
or app usage and geographical location in
Part 1: Rationalizing the Path to Purchase Institute. McGraw-
and out of the store. For instance, Hamp-
Investment shire, England-based Path Intelligence’s new
Hill released her book, “The Shopper
Economy,” in March.
Footpath technology eliminates the need for
Part 2: Measurement of
shoppers to formally join a panel or even in-
Shopper Behavior stall anything on their phones. Instead, the
system is installed by a retailer to detect sig- JWT/OgilvyAction Inc., conducting
Part 3: Measurement of Brand nals emitted from mobile phones and track business under the OgilvyAction and
Impact movement during the trip. JWT Action brands, is a fully integrated,
While companies involved in single-source end-to-end shopper marketing and ex-
Part 4: Effective Integration tracking maintain stringent privacy policies, periential marketing agency with main
Practices there still may be some nervousness associ- offices in New York, Chicago and Akron,
ated with this new, deeper level of shopper Ohio. It is part of the WPP Group.
Part 5: Retail Collaboration traceability. Even so, the advantages for mar-
keters are many. Since all behavior, includ-
Part 6: Directions for the ing media consumption and purchasing, are
Future recorded in real time, it is possible to discern
whether exposure to particular marketing
6