Making the Case for Trip Analysis

Springtime in Charleston is a sight to behold. The azaleas peaked a couple of weeks ago (great season with only a very late cold snap). The Lowcountry Strawberry Festival is running at Boone Hall Plantation. Flowertown just finished up in Summerville. Jasmine and wisteria are blooming all around. The trees are greening around the marshes and swamps. This turn into spring always makes me think about how our daily routines change. In areas where winter hits harder, warming temperatures bring people outside. Everyone seems to walk more. The kids start their spring sports and outdoor activities.

Seasonality is an important factor in our analysis of trends in nearly all transactional businesses.

Without an understanding of the variations that occur by day, week, or month, we would confuse periodic volume changes and longer-term directional shifts. Even now after decades of work, I'm still amazed that the math behind this kind of analysis is identical to what I learned as an electrical engineer in signal processing (though my amazement is misplaced – these are signals after all!)

This kind of analysis is usually done based on movement (shipments, quantities sold, etc.) With some work to control for other factors dependent on the industry (inventory availability, holidays, weather or unusual events), one can often tease out a signal showing the “oscillation” of demand over time periods. More frequently, there are multiple such signals running at different time scales. Folks in food service pay attention to day part analysis (“coffee cups by hour,” “item mix by time of day”). Retailers pay attention to annual peaks such as the Christmas season, Chinese New Year or the many spring holidays common around the world. Folks in ticketing model demand around approaching events to optimize pricing and decide on timing for new inventory. People buy more sun care products when it's hot and sunny outside.

These concepts all make sense intuitively and can be observed with real analysis based simply on shipments.

But there are still a lot of potential sources of error that could skew one’s forecast on a promotion or estimate of demand for an item at a given price. 

In a retail context, does higher volume indicate more customers responding to a promotion or loyal customers stocking up (“pantry loading”)? Is someone buying up inventory (possibly to trade later) or are we seeing new buyers who will actually use the product or service (taking it off market)? Would we have made the same sales without incentives?

A good next step in profitable insights is to leverage trip analysis. 

Knowing how many customers (shoppers, visitors, followers) we have associated with our sales gives us a far better view of how our marketing activities are succeeding. Knowing we sold 100 widgets yesterday (versus 80 on the same day in the previous year) is interesting. Knowing there were 22% more customers who bought those widgets tells us we have managed to interest more shoppers, which is usually one of our prime goals with promotions.  

Most of our retail clients are really interested in promotional analysis, given the importance of promotions and the (often overwhelming) array of options available from suppliers to incentivize shoppers. Which offers make sense? Which items did our shoppers buy with the items on sale? Did we manage to interest new customers for our items or did we needlessly discount to sell what we were going to sell anyway (missing an opportunity to make a different offer)?

While the scope of promotional analysis can involve many other data points, a good first step is to count trips (visits). 

Building from customer count and average ticket size, we can then extend to count transactions involving specific items on promotion, categories or combinations. We can leverage our CDP (Customer Data Platform) to compare responses based on our different customer segments or factors such as overall purchase frequency.

It can be overwhelming to consider all the data at hand in analyses such as these. With the changing season, consider this a pitch to add trip analysis as a relatively easy next step in profitable insights. Once your team has confidence adding this important metric, you can expand to counting metrics that track baskets “with promotions”, “with items” or “with categories.” From there, you can unlock the insights from market basket analysis to understand relationships between items and categories (affinity and cannibalization) to optimize “co-promotion” results and better estimate overall promotional impact. Promotional analysis should be a key part of most retailers’ toolkits. It starts small and simple with trip analysis.