One of the biggest challenges in the type in-store shopper behavioral measurement that we do at Digital Mortar is making the information accessible. It’s a common problem in analytics. The most interesting and rewarding uses of behavioral measurement – digital or store – come from tackling specific analytic questions. But analysis isn’t everyone’s cup of tea. It’s time consuming and seemingly endless. People are often looking for a few straightforward performance indicators – which is the allure of KPIs.

 

The challenge is that when it comes to store performance (or website performance for that matter), the most used and probably the most useful KPIs aren’t behavioral. You naturally want to know how many visits you had to the store and how much you sold. Those come from door-counting (which is only kinda/sorta behavioral) and Point-of-Sale data.

 

But let’s say you’re looking to go a little deeper and have KPIs that reflect how well various aspects of the store are working. If you were building a dashboard of store performance designed to go beyond traffic and conversion, what would it look like?

 

That next level dashboard should be designed to help you understand the top-level numbers – why store performance is the way it is. You may also want KPIs that lend context to those numbers – things like costs, margins and satisfaction.

 

A good place to start that next-level dashboard is thinking about what drives performance. Three factors are worth focusing on: quality of customer, merchandising performance, and operational performance. In other words, if you get better customers in the door, your top-level metrics will improve even if everything else stays the same. Similarly, keep your customer mix constant, if you can improve your merchandising performance, you’re top line should improve. And ditto for operational performance – you should either be able to increase sales, improve satisfaction or increase margin by reducing cost or friction.

 

Measuring Shopper Traffic Quality

When it comes to dashboarding, nobody wants a huge pile of metrics to sort through. But figuring out the most salient metrics isn’t easy. To paraphrase Pascal, “If I had more time, it would have been shorter.” Short and salient is work. Unfortunately, measuring shopper quality is particularly hard – probably the most challenging metric to capture behaviorally.

 

What makes shopper quality hard to measure? It’s relatively easy to measure whether a shopper visit was successful – at least in terms of ending up at a cashwrap and making a purchase. But if a shopper doesn’t buy anything, was it because of the shopper or the store? How do you distinguish between a customer who had intent to buy but was lost to (for example) an out-of-stock condition – and a customer who visited the store with no intention to buy at all?

 

This problem is especially pressing for non-destination stores. If you’re a mall or high-street store, a fair amount of your traffic is probably not well qualified. And if your store is a common entry or exit point for a mall – especially one anchored by movie theatres or entertainment, then a majority of your traffic might not even be shopping.

 

While there might not be a way to do this with absolute certainty, there are behavioral journey metrics that can help. First, consider time in store. If you measure total journey time in store, distinguishing between pass-by traffic and actual shoppers is easy. This will also give you a fairly solid look at overall shopper quality. It’s useful to break this metric up into four or five segments, essentially classifying shoppers in terms of very short, short, average, long and very long. You can do this by thresholding (especially for pass-by traffic levels) or by using standard deviation ranges.

 

In and of itself, though, time in store isn’t enough. You can crisp it up by measuring user journeys based on the time they spend in functional groupings of the store. There are several different ways to create a functional grouping, but typically this would include main walkways, merchandising areas, operational areas, and miscellaneous areas. Some stores will have additional groupings for things like experiences. By measuring the percentage of time a shopper spends in each of these areas, we can classify the shopper into buckets based on their purchasing proclivity. People who spend most of their time in merchandising areas are much better customers than people who spend all their time in the walkway.

 

Bucketing people based on their primary time allocation- operations, pass-by, and merchandising and  cross-tabulating these buckets with the total time in store metric creates a beautiful little matrix of shopper quality distribution. Depending on the number of your buckets, this will have between 9 and 16 cells and provides a great, compact visual representation of shopper quality and how it’s changing.

 

Merchandising Effectiveness

The second major functional area that ought to exist in a behavioral dashboard is understanding merchandising effectiveness. Merchandising is the heart of the store – getting shoppers in front of product is the point.  Fortunately, measuring merchandising performance is the kind of thing that shopper journey measurement does best. Time-based metrics are natural indicators of merchandising engagement, but rather than use avg time metrics, it’s better to use threshold metrics.

 

Our DM1 platform provides two settable threshold metrics of engagement called Lingers and Considers. The idea of a threshold is that every shopper who exceeds a certain amount of time in a merchandising area is shopping. And by having multiple levels, you can quickly establish rates for the percentage of shopper traffic hitting each level of engagement.

 

In addition to calculating this for all merchandising areas, it makes sense to calculate average linger and consideration engagement rates per section, area and even display. Part of the detail of any good shopper behavioral reporting is going to be these linger rates by individual areas, sections and displays. In a dashboard, that’s too much detail. But if you have 20 sections, you can measure the linger rate for each and then calculate the average of those 20 rates. This gives you a single metric – average Section Linger Rate. A similar metric at the display level may average the linger rate for hundreds of individual displays.

 

What’s nice about this method is that it gets dashboard consumers used to thinking about the average Section or Display rate in the store – not only is that a very comparable cross store metric, but it provides contextual knowledge about whether a section or display is above or below average. If someone says that a sections linger rate is 61%, people know whether they should “ooh” in satisfaction or “oof” in dismay.

 

Providing that kind of context is what a good dashboard metric should do.

 

One nice additional technique is to extend the shopper focus segmentation to refine measurements of merchandising performance. By measuring the linger rates only for merchandising focused shoppers, you can clean up bias introduced by, for example, a section being in a common pass-through corridor.

 

It can even make sense to measure merchandising effectiveness for shoppers in the other two groups. You might, for instance, consider lowering the engagement threshold and measuring low-linger rates in various sections. That can help you identify merchandise that’s potentially more appealing to pass-through traffic. After all, even low quality traffic has potential value if you can find the right product/merchandising setup to create at least some engagement.

 

Operational Reporting

It might be surprising to think of journey analytics generating operational measurements for a dashboard. But Associates are a huge part of a store’s operational efficiency and cost. They are also, without doubt, the most variable component of performance.

 

One big driver of store performance is the number of Associates you have on the floor. More Associates should (but may not) drive more sales. More Associates will (with 100% certainty) drive higher cost. So a good store dashboard should illuminate whether staffing levels have changed relative to the shopper volume and whether those staffing levels are driving improved conversion.

 

The STARs clock visualization is an effective way to capture BOTH these things – associate ratios relative to traffic volume and the impact on conversion. It’s a great viz for a store dashboard.

 

One of the bigger challenges of managing store, though, is knowing whether adding Labor Hours actually made a difference in terms of time on floor and customer interactions. It can be frustrating to add Labor Hours, see no uptick in sales, and have no idea why.

 

We recommend tracking the ratio of allocated labor hours to two key people-measurement metrics: Associate Time on Floor and Associate Interactions. Associate Time on Floor measures the amount of time Associates spend out on the floor – potentially engaging shoppers. If you’ve added Labor Hours, but that time isn’t reflected in increased time on floor, that’s a problem. Similarly, if you cut or lost labor hours to churn or absenteeism, the time on floor metric will help reveal how much this matters. Time on floor selling is often the marginal casualty when staff gets reduced or is thin on the ground. Measuring interactions and interaction rates can be even more pointed. With people measurement you can measure how many interactions there are between shoppers and associates and what percentage of shoppers had interactions.

 

The first metric should track very closely to the marginal addition of labor hours. The second metric can be used to measure the potential for additional expansion and, by tying to conversion rate, it can help give you a measure of opportunity.

 

Not every shopper wants or needs an interaction; it’s important to measure the relationship between interaction rate and conversion rate (and this going to be time and day sensitive too). But once you have a sense for what interaction rate is optimal, then you’ll know that if you’re below that rate on a given volume of traffic, there is opportunity to be had with additional staffing. If you aren’t below that rate, then no matter how many labor hours you allocate, you probably won’t see any benefit.

 

A people-measurement dashboard that measures Intraday STARS to conversion rates, Associate Time on Floor vs. Labor Hours Allocated and Interaction Rates vs. Labor Hours provides extraordinarily powerful context around one of the key levers for store success – the staffing of the store.

 

 

Put these three sets of metrics together and you have a powerful, compact store dashboard that provides real intelligence about how the store is performing and how to make it better.