It used to be very easy to calculate merchandise profitability. The metric was called demand per thousand pages circulated (DMPC). You identified an item, measured how much space it occupied in your catalog, quantified the total circulation of the catalog and calculated DMPC.
Assume that you circulated 1 million catalogs, and the item being analyzed occupied exactly one-quarter of a page. The item generated $7,500 demand. Here's how DMPC is calculated:
DMPC = ($7,500 / (1,000,000 * 0.25)) * (1,000) = $30.
This number ($30) would then be compared against other items in the catalog that generated other levels of demand. Take an item that occupied exactly one-third of a page, generating $8,000 demand.
DMPC = ($8,000) / (1,000,000 * 0.3333)) * (1,000) = $24.
The beauty of DMPC is that it allowed marketers to realize that the $7,500 item was truly “more productive” than the $8,000 item. Armed with this knowledge, marketing efforts were made far more efficient.
Today, we no longer have this level of clarity. Items are also available for sale online, and many different forces determine the level of online sales for an item. Email marketing, search marketing, social media, portal advertising, affiliate marketing and any of thousands of microchannels also play a role in the sales levels of items online.
This is why what I call the “large A/B test” has become so important.
Most marketers would never hold out a large percentage of their circulations for the sake of an “A/B” test. If you're mailing a million catalogs, or delivering a million email campaigns, you might consider, at most, a holdout group of 50,000. Holding out more than 50,000 would cause you to lose too much demand, right?
In 2010, give this a try: If you're mailing a million customers, try holding out 250,000.
Yes, 250,000.
First of all, you're not going to lose all of the demand. Cannibalization of marketing activities causes demand to spread out across catalogs and email campaigns.
Second, you're going to learn how online items sell. Sum up all online demand for 1 million mailed customers and for 250,000 in the holdout group. Gross up total demand in the 250,000 holdout group by a factor of four to equalize circulation levels.
Then subtract the difference between the mailed and holdout groups. Add this difference to your phone demand. Finally, calculate DMPC on this total.
This method for calculating DMPC yields the true impact of catalog and/or email marketing activities. Now you're accurately assessing the productivity of each item.
You'll obtain results that are contrary to what you believe you know. Some items drive demand to other items; some items cannibalize other items. Calculate DMPC via A/B split tests with large holdout groups, and learn how customers truly interact with your business.
Kevin Hillstrom is president of MineThatData, a database marketing consultancy. Kevin can be reached at kevinh@minethatdata.com.
- People:
- Kevin Hillstrom