Platt Retail Institute: Audience Measurement Technology Optimizes the In-Store Digital Media Opportunity at Whole Foods Markets
By David Haynes,
founder, pressDOOH
Through
a special arrangement, presented here for discussion, is a summary
of a current article from the Platt Retail Institute’s Journal
of Retail Analytics.
A network of
digital screens installed in Whole Foods Markets in North America is
providing all the medium’s stakeholders with some unique and valuable
insights on the shopping audience and its dynamics. Operating so far
in Whole Foods in Chicago and Toronto, with more under agreement, the
network uses digital screens to help vendors raise brand and product
awareness, educate shoppers and, most critically, drive sales.
The program
is particularly important in the nutritional grocer sector because many,
if not most, of the companies with products on the shelves do not have
the budgets to build brand awareness using mainstream media.
Each store
has as many as 11 “zones” – or marketing channels – defined and located
by product area such as bakery, produce or seafood. Those stations have
one or more large flat panel LCD screens positioned within easy view
of shoppers, with a small sensor fixed in the screen fascia. Biometric
software technology detects and calculates how many different faces look,
for how long and when, and also segments viewers by gender.
A regular report,
dubbed Pulse Analytics, provides details on overall audience viewership,
individual station viewership, and by request only, detailed campaign
viewership for the digital in-store program. The data so far shows some
commonality across the retail estate, and also some unique store-to-store
dynamics such as:
- Total audience:
Foot traffic in a typical store is about 100,000 monthly, and the impression
count suggests more than 75 percent viewership store by store. - Peak viewing
times: Peak time-of-day impressions vary considerably store to store.
Early to mid-afternoon is consistently a high traffic period, but while
impressions drop off steadily from then in a suburban location, they
stay high through the dinner hour at a more urban location and actually
peak mid-evening in the Chicago store. - Day of week:
Tuesday is the strongest weekday for audiences, while Saturday is the
highest day of the week in the Toronto stores and Sunday slightly higher
than Saturday in the Chicago store. - Most viewed
stations: Results vary by store, but grocery and supplements are high.
The stations where there are not, typically, brand-oriented messages
are the weakest – such as seafood, meat and produce. - Gender mix:
More than two-to-one female over male, but the data show in some locations
the male percentage is higher at certain stations, such as prepared
foods.
The data also
gives Whole Foods an
understanding of the actual viewing time, confirming this is a glance
or pass-by media in which shoppers steadily scan the surroundings. The
technology is showing shoppers are engaging with the screens for less
than three seconds on average, with some spots appearing to hold attention
longer than others. To counter that short engagement time, content is
being developed that is short and punchy and has a quick call to action
of some kind. Vendors are now using the data to get a far better grasp
of shopper demographics and traffic patterns across the entire store,
down to the individual stations.
Discussion Questions:
What true benefit do you expect retailers will gain from shopping metrics
tied to digital signage networks? Is the ROI there yet for retailers?
What factors will impact vendor adoption of the medium?
Most of the things you mentioned were things that any retailer worth his salt should already know.
Studies like this have been taking place through observation for many years. This appears to be far more detailed, but it all boils down to what this means and how stores can turn data into actionable information.
I assume it will take some time to gather enough data to do this. It still begs the question–what’s it going to accomplish? Is it going to help sell more goods? Make for more efficient assortments? Better store design? Or, is it another exercise in statistical self-satisfaction?
Given that there are still not meaningful metrics tied to television advertising in terms of ROI (i.e., where the return is based on sales lift, not some metric like reach, impressions, awareness or even purchase intent), there’s no reason that we should expect audience measurement in-store to advance ahead…with one glaring caveat: the medium is in-store!
There are ways to establish tests and controls to determine sale lift, both based on customer and product sales. To the extent that merchants employ proper testing techniques, it’s all measurable. It really depends on the degree to which that firm is oriented around quantitative tracking of marketing–and especially media–investments.
Today, there are in-store signage companies with rigorous testing capabilities and given the nascence of the industry, they are key to proving the concept. It might not be here today pervasively but it will be, and sooner rather than later.
Ok, we know what screens were looked at, when, for how long, and which ones were more popular by gender. My question is simple–what did it do for sales?
I am sure that Whole Foods didn’t enter into this program to provide their customers access to various messages from their suppliers. They saw this as a way to sell more stuff or more additional advertising money or probably both. Did they?
It would seem that the impact of the program should be measurable. If the in-store media industry wants to sell more retailers on the concept, then they need to provide research that documents the results.
The metrics discussed in this article are comparable to what out-of-home media used for billboards when there was no other way of measuring attention. It would seem that the purpose of the current measure is to validate the effectiveness of the screens in terms of their location and ability to convey marketing messages.
Importantly, the article reveals that “The technology is showing shoppers are engaging with the screens for less than three seconds on average, with some spots appearing to hold attention longer than others.”
This is another similarity with billboards, and marketers should take this brief viewing experience into account when executing their material. If the message can’t be conveyed in three seconds or less, rethink the design.
I’ve been testing billboards, signs and point-of-sale material for over 30 years. It is always a surprise to me when clients believe that shoppers are willing to stand still and ponder their messages, never mind pull over to the side of the highway to better understand a confusing visual on an outdoor board.
Another factor to take into account is the ability of the message to register on the minds of the shopper. Just because I look at something (for 3 seconds or less) doesn’t mean “I get it.”
Finally, the article doesn’t discus consumer reaction to the screens in Whole Foods. Are they glad the screens are there? Do they see them as added value to shopping in Whole Foods? Do they see the advertisers in a better light because they appear on the screens? Part of the message from the advertisers is the medium itself.
My comments may seem negative, but they’re intended to merely be cautionary. If you plan to use this medium, construct the message and its execution within the context the medium provides.
This might work better in different retailer channels than supermarkets, where everyone is in a hurry and stopping in the middle of an aisle to look at something is likely to get you bumped by a hurried shopper with “cart rage.” The “three seconds on average” comment would give me pause if I were spending the money for this research. There’s value down the line for this sort of thing, but the low-hanging fruit is not in the supermarket.
Am I missing something, or does this article ignore the single most important data source available–POS transaction logs (T-logs)?
At the risk of sounding harsh, who cares if shoppers spend 3 seconds vs. 4 seconds gazing at a plasma screen showing ads unless it actually changes shopper behavior? “Tuesday is the strongest weekday for audiences” is a purely academic statement unless it ends with “…and strong days for audiences directly correlate to strong days for sales.” Whole Foods is running a store, not a movie theater.
If no one is tying audience measures to actual sales results, then they are missing (or willfully ignoring) the point: in-store media is about driving sales. I am completely willing to believe that in-store ads work, but this article stops short. Here are some of the statistics I’d like to see:
– What is the relationship between # of shoppers who view a product’s ad and sales (holding constant for overall category sales, store traffic, etc.) of that product?
– What is the relationship between average time spent viewing an ad and number of units of the advertised product sold (again, holding all else constant)?
– Is there a measurable difference between sales in stores with advertising content shown on the screens versus, say, a screen-saver that cycles through glamor shots of food and produce? (i.e., does the mere presence of the screen and food imagery lift sales, or is it really the ads?)
These questions are critical to understanding the ROI of putting in one of these systems.
While I applaud the effort mentioned above, I believe that it is still a case of not fully understanding the true value of the retail environment. Success at retail is measured in sales; shoppers register their approval by opening their wallets…simple as that.
It seems that traditional marketing is looking for a place to validate its services and tries to apply the rules from outside retail into the retail arena. Intent to purchase and awareness are all trumped by conversion. As true marketers, we need to look for the clues that may be subconscious and hard to read, and develop strategies that are based on actual shopper actions.
There are significant differences between legacy push media such as digital signage, and emerging interactive digital media where engagement is 100% measurable. Audience demographics are similar and day-of-week, time-of-day are similar between the two channels since these appear to be correlated to traffic. What is different, based on our experiences, is the length of engagement and where the engagement takes place.
Interactive media engagement averages 67 seconds according to our internal data, versus the three seconds mentioned in this study for passive digital signage. Of particular interest is engagement by location, which the study found grocery and supplements stronger and seafood, meat, and produce weak due to lack of branded messaging. Our internal data suggests the opposite pattern, with the highest usage found in the meat aisle, followed by produce. Both of these departments have the highest shopper dwell time and consequently reflect the opportunity for delivering a brand message to an engaged shopper.
As other comments have suggested, of greater interest is the correlation between the 3 second “glance time” and sales. If 3 seconds results in measurable lift, then interactive is a bargain, particularly when the cost to advertise on the two channels is comparable.
It is good to see everyone picking up on the 3 second thing. I’m sure there are many more shoppers for whom the exposure is simply a glance. Anything moving in an otherwise static environment needs to be assessed by the shopper for filtering out–which is probably what most of the few exposures are accomplishing.
The real problem here is that there is a beguiling tendency to think of television and computer screens when thinking about digital media in-store. People have wasted hundreds of millions of dollars trying to “think TV” relative to in-store media.
The problem is so bad that I have resorted to saying “shoppers are NOT an audience!!!” The ephemeral, incidental exposures in the store are mostly an addition to the clutter already there. The exception is not the rule.
These comments are addressed particularly to CPG/FMCG stores, such as Whole Foods.
The article states, “Biometric software technology detects and calculates how many different faces look, for how long and when, and also segments viewers by gender…” this is one key fact that has not been brought out in many of the comments so far. Technology exists today (e.g. http://bit.ly/MSUwA) that can not only perform the functions mentioned, but can also actually determine the human reactions to the advertising messages. We are now at a point where we are creating computers that know how you feel. This is the source for generating new metrics that can drive true, tangible ROI for these kind of efforts.
Whole Foods is hopefully embarking upon this innovative journey to find real value in their marketing strategy based upon actual measurable habits that their customers are exhibiting. This is good stuff. More companies need to take advantage of what is now out there.
I’ll apologize in advance as my response on this has very little linkage to the questions posed because I think to some extent they miss the real point.
All that’s happening here is brands that can’t afford to pay to interrupt the consumer in their home via TV, radio or print are paying less to interrupt the consumer in the store via POS media. Regardless, it’s an interruptive approach to marketing and measuring who looks at it will give little insight into the degree to which they might resent it.
My core belief is that better interruption tactics won’t really move the needle in retail. Instead, I think retailers need to do a better job of creating information outposts that work on a “permission” basis with the consumer–providing information when and where they need it.
“Smart carts” are a clear step in this direction. With the right functionality they make the store visit more efficient and pleasant for the shopper. At the same time, their ability to cross promote products in an item-specific way grows the average basket size.
As usual, I side rather strongly with Herb on this issue. Exposure is not the same thing as sales impact. Measuring opportunities-to-see has scant value as a proxy for promotional lift.
Even if we measure audiences better and better with slick technology like face detection, dwell times, eye movement or other high-tech tricks, the collected data has limited value unless it is causally linked to current sales and enduring attitudes.
In order to gain any sense of the value of an exposure we need serious and sober-minded study in at least five areas:
How many see the message? In what retail context? How does it make them feel then? What do they do afterward? How do they feel about it now?
The answers to these questions must all be linked to sales and profit impact for the brands and the host retailers.
When we start to define the store as an audience aggregator like a television network, we seem to lose track of its initial purpose–which is to be a purveyor and seller of goods, at a profit.
In the supermarket environment, in-store digital signs make a pretty crummy advertising medium, but a somewhat more effective promotional medium. In a shopping mall corridor or transit hub, digital signs may work better at advertising. The thing to remember is that it’s a bad idea to try to conform the ancient television ad model to an in-store context.
Chant after me: “It’s not TV. It’s not TV. It’s NOT TV!”
In-store digital signage with aggregate measurement is just another form of mass media–no customization is possible based on customer knowledge. As another post noted, there is also no direct connection of these in-store ads to sales, and that is the most critical link of all.
In all the tests I have seen of in-store media, there is a short-term lift of around 10%–no matter what the media is. Then as the effort becomes more common, the results dwindle to nothing. I would expect the same results here.
In today’s data-rich world, the keys to the kingdom belong to marketers who can leverage customer insight to personalize and expedite the individual customer’s experience. Anything else is just “spray and pray,” even if it is only to store customers.
Active TVs in a retail environment are nothing more than POP noise. These continuous advertisements distract their audience from purchasing their products in the environment that Whole Foods has tried to establish and instead, set up a background noise that must be overcome on every aisle. Whole Foods and other retailers need to stick to customer service, eliminating out-of-stocks, better shelf presentation, and product management. Anything else is a distraction from their core competencies.
This will be just another data point, which the store will add to their coupon data points, register data points, etc; all of which are not properly data drilled or mined in order to maximize their information. Keep it simple and Whole Foods will continue with their success.