Tag Archives: in-store measurement

Creating a Measurement Language for the Store

Driving real value with analytics is much harder than people assume. Doing it well requires solving two separate, equally thorny problems. The first – fairly obvious problem – is being able to use data to deepen your understanding of important business questions. That’s what analytics is all about. The second problem is being able to use that understanding to drive business change. Affecting change is a political/operational problem that’s often every bit as difficult as doing the actual analysis. Most people have a hard time understanding what the data means and are reluctant to change without that understanding. So, giving analysts tools that help describe and contextualize the data in a way that’s easy to understand is a double-edged sword in the best of ways – it helps solves two problems. It helps the analyst use the data and it helps the analyst EXPLAIN the data to others more effectively. That’s why having a rich, powerful, UNDERSTANDABLE set of store metrics is critical to analytic success with in-store customer tracking.

Some kinds of data are very intuitive for most of us. We all understand basic demographic categories. We understand the difference between young and old. Between men and women. We live those data points on a daily basis. But behavioral data has always been more challenging. When I first started using web analytics data, the big challenge was how to make sense of a bunch of behaviors. What did it mean that someone viewed 7 pages or spent 4.5 minutes on a Website? Well, it turned out that it didn’t mean much at all. The interesting stuff in web analytics wasn’t how many pages a visitor had consumed – it was what those pages were about. It meant something to know that a visitor to a brokerage site clicked on a page about 529 accounts. It meant they had children. It meant they were interested in 529 accounts. And depending on what 529 information they chose to consume, it might indicate they were actively comparing plans or just doing early stage research. And the more content someone consumed, the more we knew about who they were and what they cared about.

Which was what we needed to optimize the experience. To personalize. To surface the right products. With the right messages. At the right time. Knowing more about the customer was the key to making analytics actionable and finding the right way to describe the behavior with data was the key to using analytics effectively.

So when it comes to in-store customer measurement, what kind of data is meaningful? What’s descriptive? What helps analysts understand? What helps drive action?

The answer, it turns out, isn’t all that different from what works in the digital realm. Just as the key to understanding a web visit turns out to be understanding the content a visitor selected and consumed, the key to understanding a store visit turns out to be understanding the store. You have to know what the shopper looked at. What was there when they stopped and lingered. What was along the corridor that they traversed but didn’t shop. You have to know the fitting room from the cash-wrap and an endcap from an aisle and you have to know what products were there. What’s more, you have to place the data in that context.

Here’s what the data from an in-store measurement collection system looks like in its raw form, frame by frame:

TimeXY
04:06.03560
06:50.0966
09:10.02374
11:02.01892
11:35.03398
13:15.02874
14:25.0781
16:16.04175
19:09.04962
21:03.04572
23:23.05583
23:58.05490
24:09.04086
25:05.01590
27:24.0779
27:45.04399
28:42.03797
29:25.04580
32:07.04775
33:05.01677
35:31.03765
36:08.03475
36:33.0973
39:16.03576
40:07.01397

That’s a visit to a store. A little challenging to make sense of, right?

It’s our job to translate that into a journey with the necessary context to make the data useful.

That starts by mapping the data onto the store:

store journey analytics

By overlaying the measurement frames, we can distinguish the path the user took through the store:

StoreFrame1

With simple analysis of the frames, we can figure out where and when a customer shifted from navigating the store to actually spending time. And that first place the shopper actually spends time, has special significance for understanding who they are.

In DM1, the first shopping point is marked as the DRAW. It’s where the shopper WENT FIRST in the store:storeFrame2

In this case, Customer Service was the Draw – indicating that this shopping visit is a return or in-store pickup. But the visit didn’t end there.

Following the journey, we can see what else the customer was exposed to and where else they actually spent time and shopped. In DM1, we capture each place the shopper spent time as a LINGER:

storeFrame3

Lingers tell us about opportunity and interest. These are the things the shopper cared about and might have purchased.

But not every linger is created equal. In some places, the shopper might spend significantly more time – indicating a higher level of engagement. In DM1, these locations are called out on the journey as CONSIDERS:

storeframe4

Having multiple levels of shopper engagement lets DM1 create a more detailed picture of the shopper and a better in-store funnel. Of course, one of the keys to understanding the in-store funnel is knowing when a shopper interacts with an Associate. That’s a huge sales driver (and a huge driver – positive or negative – to customer experience). In DM1, we track the places where a shopper talked with and Associate as INTERACTIONS. They’re a key part of the journey:

storeFrame5

Of course, you also want to know when/if a customer actually purchased. We track check-outs as CONVERSIONS – and have the ability to do that regardless of whether it’s a traditional cash-wrap or a distributed checkout environment:

storeFrame6

Since we have the whole journey, we can also track which areas a customer shopped prior to checkout and we’ve created two measures for that. One is the area shopped directly before checkout (which is called the CONVERSION DRIVER) and the other captures every area the customer lingered prior to checkout – called ATTRIBUTED CONVERSIONS.

StoreFrame8

To use measurement effectively, you have to be able to communicate what the numbers mean. For the in-store journey, there simply isn’t a standardized way of talking about what customers did. With DM1, we’ve not only captured that data, we’ve constructed a powerful, working language (much of it borrowed from the digital realm) that describes the entire in-store funnel.

From Visits (shopper entering store), to Lingers (spending time in an area), to Consideration (deeper engagement), to Investment (Fitting Rooms, etc.), to Interactions (Associate conversations) to Conversion (checkout) along with metrics to indicate the success of each stage along the way. We’ve even created the metric language for failure points. DM1 tracks where customers Lingered and then left the store without buying (Exits) and even visits where the shopper only lingered in one location before exiting (Bounces).

Having a rich set of metrics and a powerful language for describing the customer journey may seem like utter table-stakes to folks weaned on digital analytics. But it took years for digital analytics tools to offer a mature and standardized measurement language. In-store tracking hasn’t had anything remotely similar. Most existing solutions offer two basic metrics (Visits and Dwells). That’s not enough for good analytics and it’s not a rich enough vocabulary to even begin to describe the in-store journey.

DM1 goes a huge mile down the road to fixing that problem.

[BTW – if you want to see how DM1 Store Visualization actually works, check out these live videos of DM1 in Action]

An Overview of In-Store Tracking Technology

How does it work? Can you really do this? Is it legal? Those are the questions that I get asked the most about in-store customer journey tracking. The same kind of questions, to be honest, I used to get fifteen years ago in digital analytics. And when you have to answer questions like these, you know it’s still pretty raw out there. Collection technologies are a core part of measurement – whether it’s tags in digital analytics or PCAP files for in-store customer tracking. Technology matters. And with in-store tracking, the data collection technologies aren’t half-baked, but they aren’t well-cooked either.

Here’s what you need to know:

Collection Technologies

There are four (!) common approaches to in-store customer tracking: camera, wifi, passive network and mobile apps. Each has distinct characteristics and at least some advantages and disadvantages. Camera is pretty easy to understand. The cameras used for in-store measurement are video. Each camera has on-board processors that identify people, “blob” them, and then track them across their field of view. This yields a stream of data that is positionally very accurate and can also identify basic demographics around each visitor. The anonymized data is then passed to a central server where systems like ours can use it.

Your existing WiFi system can also be used to track customer journey data. This works whether or not people login to your access points. Phones regularly ping out looking for a network and those pings – anonymized – can be triangulated to figure out the position. Put those pings together, and you have a journey. One of the best things about WiFi tracking is that almost everybody already has the necessary hardware in place. That means there’s no new installation; and most of the top-tier providers of internet access points make it super easy to route the data directly to your cloud-based system. Often at no additional cost.

Passive network sniffers are small WiFi-like devices designed explicitly for in-store measurement. They work on principles similar to WiFi but they solve some problems that WiFi doesn’t. They track multiple bands, not just passive WiFi pings, and they can deliver better positional accuracy because they can be deployed in very large numbers quite cost-effectively.

Lastly, you can use code inside a mobile application to track the customer journey. Mobile apps can deliver a steady stream of positional data and have the unique benefit of being able to tie that data to the customer’s digital in-app experience.

So what’s not to like?

Well, each technology has some significant issues.

Cameras are expensive, installation is a challenge, each camera only covers a small zone, and camera systems do a remarkably poor job stitching together the customer journey. So as typically delivered, camera systems cost a lot and deliver limited measurement.

WiFi isn’t very accurate positionally – meaning it can’t be used effectively for much beyond door-counting in smaller and mid-size retail spaces. Worse, changes in MAC randomization in the IoS world have essentially eliminated the ability of WiFi systems to passively track customers with Apple devices. That means you either depend on users to connect to your WiFi (which does yield stable measurement) or you only measure your Android customers. Two bad solutions don’t add up to a good one.

Passive network sniffers improve on WiFi in terms of positional accuracy and their ability to fingerprint devices. But they don’t solve those problems perfectly and, of course, they don’t give you the no-installation, no hardware cost convenience that WiFi did.

Measurement using mobile apps? That’s great, just like everything with mobile apps provided you can get customers to actually download the app. Depending on customer app downloads for measurement is inherently a limiting factor.

Bottom line? There are places and times for every technology and there are ways to combine the technologies to yield better results (we do that). But this isn’t measurement nirvana. No solution is perfect and you’ll find plenty of things to hate in any direction you choose.

To get more detail on the ins-and-outs of in-store customer journey tracking technology (and it’s complicated), ping me. I’ll send you a DM whitepaper that gives you everything you need to know to choose wisely!

I’ll tackle practicality and legality next time!

Optimizing Omni-Channel with Analytics from the In-Store Customer Journey

I’m going to be co-hosting a webinar with my friend John Morrell at Datameer on Omni-Channel Analytics and using In-Store Customer Journey Data. It should be pretty cool stuff – and, of course, it’s free!

You can register here!