Abstract

Groups of shoppers can have a surprisingly large impact on key performance indicators like conversion rate and STARs. If you only focus on individual shopper counts, you may be getting a misleading view of performance. Metrics like conversion rate will shift significantly when groups are taken into account, but the real problem is that those metrics will shift differentially. In other words, it isn’t just that the conversion rate will change if you measure groups, the important part is that conversion rates will change differently by store, by section and by daytime part. That means that when you compare conversion rates without accounting for groups, you may just be getting the wrong answer. And since all measurement is ultimately about comparison, group impact on key metrics becomes very important.

Why Group Measurement Matters

Group measurement is pretty much just what it sounds like – the measurement of whether individual shoppers are part of a group. It’s an important measurement for a couple of reasons. The first and biggest is that it helps crisp up another metric – one that happens to be critical to measuring store and area performance.

If you measure 1000 shoppers entering a store and you had 400 transactions, your store conversion rate is 40%. 40% of the people who came in the store bought something. But as fundamental a measure as that is, it’s not quite right. As significant percentage of those people might have come as part of a group – a family, a couple or a group of friends. The store might have had only 750 groups enter – the additional 250 people were part of a group. That means the conversion rate was more like 53% not 40% – that’s a pretty big difference.

It’s the same at every level of the store. If you’re measuring conversion by section, measuring the number of shoppers and not the number of groups can be deceiving. That’s especially important to measure because not every section of the store will get the same percentage of group shoppers. Some gendered sections may get significantly less group shopping. Some store sections may get more or less kids. What that means is that your true conversion rate by section may be substantially off if you aren’t measuring groups at the section level. You can’t just count groups at door entry and think you’ve solved the problem.

Of course, the same dynamic can play out at the full store level. With different demographic populations, two stores with identical traffic may have fairly different conversion rates based on group percentage.

Nor is conversion rate the only metric impacted. If you’re using shopper counts to allocate Associates – whether at the store level or below, group percentages will have much the same kind of impact. You may over or under allocating Associates based on shopper traffic.

How much difference might grouping make? It’s obviously going to depend on the store. But in a recent analysis, the overall percentage of shoppers entering in a group occasionally topped 50% and was often in the 40s. What’s more, there were VERY significant differences by day of week, daytime part and by store section.

Group percentages sometimes dopped into the teens. That’s important, because it means that the impact of grouping is going to be much stronger for some day-times and store sections than for others. In the example above, a conversion rate for a section might be off by 4-5x if it didn’t take account of group impact. That’s a HUGE difference and it means measuring groups at the door-level is sub-optimal if you’re doing more detailed section level measurement.

Of course, group measurement isn’t just important to get your conversion rates correct. If you can accurately measure groups, you can start to figure out if your Associate strategies are optimal for both individuals and groups. By relating interactions to groups and then tracking conversion rates, you can begin to craft and TEST alternative engagement strategies.

Pretty cool stuff. Of course, none of this explains HOW we measure groups.

How to Measure Groups

So how we measure groups of shoppers? In one sense, it seems like it should be obvious. We measure shoppers who are shopping together. Since we’re tracking full journeys through the store, that seems like it ought to be easy. As with most real-world measurement, though, it’s harder than it looks – particularly when you try to account for all the edge cases.

First, shopper groups rarely stay together for their entire visit in the store. It’s not unusual for people to enter together to split up, rejoin, and then split up again. That means that while they are a group in terms of store conversion rate, they may or may not be a group when measuring statistics for any particular section. If a couple walk into the store together and head to footwear, then they split up with one going to Women’s and the other to Men’s running shoes, are they a group? And in what context? Their a group for the store but not for the section. On the other hand, what if they both spend 5 minutes in Men’s running shoes and then one of them goes to Women’s running shoes. Now they’re a group for Men’s running shoes and the store but not for Women’s running shoes.

It’s complicated!

To classify shoppers as belonging to a group, we rely on a few basic measurements. The most important is time in close proximity. Under normal circumstances, people who aren’t shopping together don’t spend much more than a few seconds in close proximity. So any two people with significant time in close proximity are probably together. But there are exceptions. Have you ever been seated next to someone in a shoe section? Or shopped the same aisle section as someone else for a minute or two. Probably. So in addition to measuring time in close proximity, we measure distance moved and sections in close proximity. People who are shopping together will nearly always have multiple sections in close proximity. If we’re trying to decide whether people in a single section shopped as group, we look at distance moved in close proximity. Finally, for both the store and section, we look at whether people entered and exited in close proximity. Entry and exit together are great indicators that the shopping was effectively done as a group.

By combining these metrics, we can be pretty confident that we’re getting an accurate read on the number of shopping groups at both the store and section levels. And getting shopper groupings right means being significantly more accurate in getting key metrics like shopper conversion rate and STARs correct as well.