Photo: NRF
The measured store, version 2.0
Through a special arrangement, presented here for discussion is a summary of a current article from the blog of Nikki Baird, VP of retail innovation at Aptos. The article first appeared on Forbes.com.
At the NRF Big Show, a number of startups in the Innovation Zone were dedicated to providing heat maps and other computer vision-based solutions for understanding consumer behavior in stores. Their presence again offered hope that retailers may one day understand in-store behavior a bit closer to the level of granularity with which they understand online consumer behavior.
Unfortunately, this is a problem that solution providers have been trying to tackle for a long time. Veterans like Sensormatic’s ShopperTrak have provided multi-level solutions that range from simple infrared portal traffic counters to camera-based footpath tracking.
The challenge has always been to demonstrate enough value to justify the capital outlay required to outfit an entire chain of stores. “Sample stores” can be justified, but that typically doesn’t provide the revenue that these startups need. Some solution providers are trying to tackle this problem by developing solutions that sit atop existing loss prevention cameras (typically lower resolution). Others are trying to up the “revenue” side of the ROI calculation by focusing on identifying insights that are easier to obtain and faster to respond to.
I’m optimistic that this rush of new companies — and the VC money willing to fund them — speaks to an ability to overcome those challenges that have historically slowed adoption of solutions for gaining better in-store customer behavior insights. Having seen the shipwrecks littering the shores of this problem over the last two decades, however, I fear my optimism exceeds reality. That said, the bottom line here is that the gap between online insights and store insights is large and growing. The shortfall needs to be addressed for stores to regain their position as a growth engine in retail. Thus, my optimism remains.
BrainTrust
Bob Amster
Principal, Retail Technology Group
Liz Crawford
VP Planning, TPN Retail
James Tenser
Retail Tech Marketing Strategist | B2B Expert Storytelling™ Guru | President, VSN Media LLC
Discussion Questions
DISCUSSION QUESTIONS: What are the main obstacles preventing retailers from tapping technologies to better understand in-store customer behavior insights? Do you agree that the gap between understanding online behavior insights and store insights is widening, and is that a problem?
Retailers face a litany of challenges, and the biggest is in extracting insights and applying them. Terrific technologies are great, but retailers need help in using the tools — it’s not just a matter of buying more technology. In fact, I’d argue that retailers have so many options and choices when it comes to technology that just selecting a solution can be a challenge. One executive told me that his company gets pitched literally hundreds of solutions a year, and even at their large scale (1,000+ stores), they can only practically do a handful of meaningful projects a year. I believe there will be blind spots between online and the physical store which will continue for years to come, but the gap is closing. Technology is part of the answer, but the bigger part is in retailers using and applying insights.
I could not agree more! It is one thing to choose from all that is out there but actually using the tools correctly and the ability to apply changes has always been the challenge. My thoughts on this have been to focus on the top 5%, choose options that will improve/enhance/make fun their journeys and then apply to all. For my 2 cents.
We’ve been playing this game since the mid-’80s, when IRI enabled tracking via its ill-fated Videocart experiment. The problem is not technology, it’s what you do with the data. Mostly, we don’t know what to do with the data. There is no “path” through the store that most shoppers follow – plot 10 shoppers and you get a plate of multi-colored spaghetti. If nobody goes to a section of the store, is it because it’s not appealing or because you’ve put all the low incidence items there? And could you have figured this out (I’m talking to you, Tony Orlando) by just watching your store for a bit?
As a baseline, in-store traffic has been beneficial for over twenty years for optimizing labor spend. More recent technologies have added additional layers to this data, but the privacy and regulatory concerns are still an unknown entity as customer in-store behavior is married with marketing and transactional data. So many retailers at various segments have no form of in-store traffic tech, which no doubt means the category will continue to grow. The question becomes for the modern store: are your front-line team members skilled to relate and understand to customers at various stages of their journey to maximize opportunity?
Sure online provides far better quantitative tracking of consumer behavior, but having tracked retail clickstream behavior for years, I can tell you that understanding attitudes and motivations behind those behaviors is a huge gap for online. A brick-and-mortar retailer that is well staffed with good employees that listen to customers, and that funnel their collected feedback up the organization, can have a richer understanding of their shoppers’ behaviors, intents, complaints and compliments than online retailers.
Most retailers face having to invest real money in these technologies without the certainty that they will experience a healthy ROI. Solution providers would do well to participate financially in conducting a proof of concept with measured outcomes in order to make retailers comfortable that there will be (or there will not be) a noticeable ROI. It would seem to me that online behavior is easier to measure and analyze by virtue of the fact that every customer move online can be detected, recorded, and analyzed to provide insights. The in-store equivalent, by contrast, is comprised of too many variables that the retailer cannot control, and this condition makes measurement and justification, more difficult.
I agree Bob. In my experience, you can deliver a real ROI with in-store tracking technology, but it’s more difficult to do since the solution itself doesn’t deliver the ROI, it’s the actions that are taken as a result of the insights that create the value which forms the basis of the ROI. It’s this requirement that the retailer act on the insights that’s the really hard part since retailers are often looking for immediate results. For example, some retailers install traffic counters in their stores and then immediately say, when can I expect my conversion rates to improve? My answer: that’s up to you and here’s what you need to do.
I guess we agree that there is a state called data overload.
Great point. It’s new information, new perspective, new insight. Hopefully propelling some new behavior that begets some new, better outcomes.
Nikki nailed the biggest obstacle: ROI. Plain and simple. That has stalled these technologies for decades, as well as RFID and others. I do see AI gaining strength based on some very low cost-of-entry opportunities currently.
I interpret all this as spending a lot of time and money on reviewing the shopping process but still providing little insight into the “why” of the buying decision. We can map a customer’s journey in the store and the dwell time at a location, but did that dwell time translate into a purchase, and why or why not? What if the same time, money and energy was invested in simply talking to store associates? Do we think that’s happening anyway? I really doubt it. There is no doubt that technology can and will be a great tool, but before spending a nickel on tech I would challenge the existing “listening and learning” process that the retailer currently employs. I keep reading that Zara is driven by straightforward best seller info and store input (and short lead times and air freight and…). Forget “fast” fashion. I much prefer “informed” fashion.
The ROI for in-store shopper tracking technology is elusive for a couple of reasons:
You can watch behavior, track it, and measure it. You may even get some insight from doing so. But none of those activities help us to plumb the depths of motivation, emotion, and individual concerns.
Cost vs. benefit is the obstacle. Nothing tells the truth like the register in a physical store and after all, store behavior can be obvious through simple observation, which should be ongoing anyway. I’ve never been a fan of this type of thing as I’m just not sure what the value would be compared to the expense, especially if you have more than 1,000 stores (and retailers with this store count would be the brands that would benefit the most). How does that even play out?
In the online world, there is little to no financial risk to monitoring shopper behavior and even conducting A/B testing. In the brick and mortar world, these activities cost. And as long as there is a cost attached, there will be lag and lack of knowledge.
People tracking has been around for decades. Behavior is difficult to track online or off. But for the store the challenge is identity. We know pathing online because we have identified a single customer and follow their specific path, plus the general path to checkout has limited focus – e.g. category, product pages and conversion. The store doesn’t allow that kind of flexibility. In-store heat maps are excellent at finding entrances, exits, and usually the restroom. The zones in a store are weakly designed to capture individual people tracking. Add to that the complexities of privacy and it can become chaotic. Actually, the gap isn’t widening – it’s just wide because we don’t know who the customer is or if the reason they’re in front of the toy section is to appease their kids while they walk around evaluating a product in their hand. The online experience gives us a controlled environment with an almost captive audience whom we at least partially know. Data will first be applied to efficiency and operations like store inventory to prevent stock outs before it is applied to customer behavior.
Just to add a thought — the right abilities here can transform the industry, especially if we can begin to apply strong business logic and AI to customer behavior. More than demographics, the behavior in the moment matters.
It’s hard to argue about acquiring more data, but it needs to be actionable and those actions need to produce the ROI for the efforts. So in some respects this is a “ditto” to others on this thread. I think there’s an argument though that says there’s a need to envision the next generation of the store; a multi-purpose distribution point that supports a variety of shopping needs, not just the traditional walk in, browse and buy model. Getting to Store 2.0 will require experimentation, and experimentation needs good feedback loops. Perhaps the role of these kinds of technologies is to study, experiment, review and experiment some more. That does not require a chain-wide roll out, it requires a properly orchestrated “test and learn” program. These technologies are a good fit and value for that.
This use case is a great example of why retailers struggle with almost anything data or analytics related. Most retailers lack an overarching vision and a use case road map for how analytics supports the business. Progressive retailers recognize this and are able to self fund more complex use cases by taking a crawl, walk, run approach to improving their analytics maturity. There is no doubt that “consumer insights” is an analytics use case, but if it renders in isolation with a high cost and an uncertain payback, it’s never going to happen. Test/learn and risk is inherent in any analytics use case. So you need to be confident and have a track record of success to a certain degree before you chew off use cases like this one.
Inferred behavior even at scale is still guesswork. Tracking shoppers in a store creates data. Data processed by algorithms designed to solve for patterns yielding insights. What is the value of an inferred insight? A prediction predicated upon a host of subjective nuances fed back to a retailer requiring action or response on behalf of the retailer to implement the insight, seeking a return on their investment. Online behavior, unlike the physical world, tracks hundreds of thousands of shoppers at once, depending on the size of the retailer. It’s a physical world issue, impossible to scale for relevancy in-store. Watch a shopper make the decision to purchase in the physical world? Good luck.
A challenge for gaining insight into customer behaviors in a physical store is equipping frontline employees — the people having direct interactions with those customers — with the knowledge and skills to pose the right questions and then, capture those responses for analysis. What better way to delve into the wants/needs, attitudes and more of an in-store shopper than to have the sales associate assisting the customer ask those sorts of questions. Of course, the reply is something like “our salespeople don’t have time to learn how to do that.” And I say they can because they must. Consider implementing a microlearning capability that offers bite-sized learning that can be easily absorbed and applied. Add to that a real-time performance capability to see how these employees are doing in this regard and really any other KPI. Then wrap it up with gamification to encourage engagement and the appropriate motivation and reward.
I rank in-store sensing as one of the most significant opportunity areas for retail in the coming half-decade. The numerous point solution vendors exhibiting at NRF is evidence that others see the potential too.
But “path-tracking” and video shopper behavior analytics have been in the hopper for years without gaining much traction. The explanation is well stated by several of our colleagues here: The business benefits cases have not been perfected.
Shopper observation (in its several forms) is just one important category of sensing with potential to illuminate physical retail by capturing relevant data at the store level. Others — demand signal, inventory/display status, message/promotion delivery, employee merchandising actions — all could fit into a larger matrix that has yet to be conceived.
That adds up to a long list of tech investment decisions that retailers must balance against other imperatives that seem to be more urgent. So far, most are in a wilderness without a trail map.
Kudos, Nikki, for focusing us on this topic.
There is a tremendous opportunity here for traditional retailers to consider how they can also leverage (read: monetize) in-store data and consumer insights enabled by new tech with their trading partners.
In fact, traditional retailers can become digital disruptors themselves, creating entirely new business models, and teams focused not only on driving internal ROI (via business analytics for their end users), but also offsetting some of related costs by re-selling this in-store data and consumer insights into “at shelf impacts” that has tremendous value for their brand partners and suppliers. Without this capability, retailers will continue to lose trade funds to their competitors anyway.
Further, considering the millions of dollars spent (and collected by) third-party providers (brokers, syndicated data providers, etc.), whose business model is essentially reselling retail execution data to measure in-store execution of displays, promotions, pricing, etc. in sample stores and often weeks after the fact, there is a natural new revenue model here for retailers capable of providing real-time and holistic data and insights by working directly with their trading partners in a more collaborative manner.
Rather than focusing on barriers, retail leaders are already leaning in, considering the bigger picture (e.g. data monetization), and the competitive implications of sitting on the sidelines.
In-store technologies, like sensors, electronic shelf labels, smart carts, tunnel scanners, etc., have been around for decades. What’s prevented retailers from benefiting from them is 1) their resistance to change 2) their ability to take action (in scale) from insights gathered.