Capture Rates: Find the Right Retail Data Properly
Back in the day, marketers didn’t have a lot data and, therefore, not a lot of computing power. But they did have a reasonable amount of certainty when analyzing performance.
The inclusion of retail, Web analytics, mobile, social media and whatever other data you want to include has been both a blessing and a curse. Now marketers can track more customer behavior than ever. But they can’t be certain what they're tracking is “right.”
Take e-mail, for instance. Say you send an e-mail campaign to a retail customer. The customer shops in the store a day later and pays using cash. The e-mail campaign worked. You don’t easily see that the campaign worked, however.
This is where the concept of capture rate comes into play. Your ability to capture retail sales transactions varies, depending on the customer segment you're analyzing.
You might be able to capture 80 percent of the transactions across best customer segments, for instance, missing a few cash transactions and new credit card numbers. Retailers analyze individual customers at different points in time. As new transactions are captured, the retailer knows the percentage of transactions that are actually captured at different points in time. This helps determine the “capture rate.”
Why is this important? Let’s go back to the e-mail campaign. A retailer sends a campaign to a customer segment and uses a mail/holdout test strategy to measure results. The mailed group spends $2 online and $2 in stores, for a total of $4. The holdout group spends $1.75 online and $1.60 in stores, for a total of $3.35. The difference between the mailed group and the holdout group is 65 cents.
But we know that in retail stores, among best customers, you’re only truly capturing 80 percent of sales. So you need to make an adjustment for the transactions you’re not actually capturing, but are actually happening. Here's a view of the results, after accounting for an 80 percent capture rate:
- The mailed group spends $2 online and $2.50 in stores, for a total of $4.50.
- The holdout group spends $1.75 online and $2 in stores, for a total of $3.75.
The difference between the mailed group and the holdout group is 75 cents. This is the secret to measuring return on investment with incomplete data. The campaign most likely outperformed your original evaluation.
The analysis is highly dependent on making reasonable assumptions, so try to be conservative. If you believe your capture rate at a segment level is 80 percent, for instance, use 90 percent as the assumption. You’ll underestimate actual performance, but you’ll protect yourself should the assumptions significantly change.
As retail channel integration continues to grow, you’ll be faced with many more data interpretation challenges. Your job is to try to estimate “what might have actually happened,” and then illustrate customer performance before and after applying assumptions.