June 21, 2007

When Capture Rate Bites You In The Belly

A market researcher will conduct four focus groups of nine individuals each, recommending brand-changing strategies on the basis of thirty-six customers.

Your market intelligence analyst will measure market share by DMA, knowing full-well that the error rate on the estimates is plus or minus seventy percent. Your executive team oooohs and aaaahs at her findings.

Then you report on actual customer purchase behavior, and remind folks that your information technology team is only able to match sixty-five percent of your retail purchase transactions to a name and address (this is called "capture rate"). Your executive team badgers you about "the other thirty-five percent", wondering if those customers behave "differently". Of course they do, they paid using cash or check or an un-trackable credit card! You share that information with your executive team. They boot you from the "C-Level" table, while the market research analyst offers a smug smile, secure in the knowledge provided by thirty-six customers.

Such is the life of the database marketer.

Market researchers and market intelligence analysts get free passes from the "C-Level" table that so many of you crave to present to. You will get a free pass back to the cubicle farm. Why?

In a word ... "consistency".

In other words, market researchers and market intelligence analysts frequently analyze metrics that behave in a "consistent" manner. The numbers don't change significantly from year to year. Therefore, your beloved "C-Level" team "trust" the numbers, even if highly inaccurate.

Your numbers are "inconsistent". They don't know what to trust. Take a look at the following table:

Actual Performance, Measured Via The Customer Database





Overall

12 Month Repurchase Spend/ Average Retail Cap-

Buyers Rate Buyer Value ture Rate
2007 Results 10,000 20.60% $235.00 $48.41 65.00%
2006 Results 8,000 24.30% $230.00 $55.89 75.00%

When you analyze these results, you really cannot tell whether customers repurchased at lower rates, or if the ability of the "information technology" folks to assign name/address to transactions got worse.

More often than not, when capture rate changes, the number of customers who purchase change. Most often, retail customers are using the same credit card, or are using two credit cards. As a result, one might consider adjusting the repurchase rate for the fact that the capture rate is lower. The analyst might adjust the 20.60% repurchase rate by a factor of (75.00% / 65.00%), resulting in a 23.77% repurchase rate.

Performance now looks like this:

Actual Performance, Measured Via The Customer Database





Overall

12 Month Repurchase Spend/ Average Retail Cap-

Buyers Rate (Adj).
Buyer Value ture Rate
2007 Results 10,000 23.77% $235.00 $55.86 65.00%
2006 Results 8,000 24.30% $230.00 $55.89 75.00%

In this table, customer performance is essentially equal, year over year.

Another step to validate this assumption is to compare the results to comp store sales performance. For instance, if comps are flat, the table above may be reasonably accurate. If comps are up ten percent, you may need to further adjust the repurchase rate.

Once you've made this adjustment, you're best off footnoting it. Your "C-Level" buddies are not big fans of "cooking the books". And yet, if you don't adjust your findings, they will not be big fans of "inconsistent data".

For Database Marketers who have to report customer behavior across time periods, capture rate is probably the biggest challenge to database marketing credibility. Make sound adjustments, speak with confidence, and your credibility will improve.

2 comments:

  1. Anonymous7:43 AM

    I understand your pain Kevin, but in my experience, MR and MI people can easily run into the same problems. Quant MR data is not exactly the model of consistency yoy, things can swing wildly as most companies never invest enough money in good sample bases and error rates of +/- 7-15% are common.

    The larger problem imo is that not many C-lvls have a good understanding or appreciation of the 'fog of war' that surrounds ANY customer data.

    Paul

    ReplyDelete
  2. Good comments Paul, thanks!

    ReplyDelete