July 23, 2012

Email Marketers Miss This Fact, Often

I simply don't understand why email marketers are stuck analyzing opens/clicks/conversions.  An entire industry follows best practices that miss much of the value of an email marketing program.


Go spend a few minutes with your geeky statistician, and if you don't have one, hire me (click here) or ask your co-op or database provider to run an analysis for you.

  1. Run a logistic regression.  Freeze your file as of one year ago.  Create independent variables (recency, frequency, average order value), and include email subscriber status as a 1/0 variable (subscriber = 1, non-subscriber = 0).
  2. With your file frozen as of a year ago, run a ordinary least squares regression model ... similar variables as above, predicting how much a customer will spend if the customer purchases in the next year.
By performing this analysis, you control for customer quality, allowing you to isolate the impact of email marketing on the overall business.  This is not an analysis where you segment by presence of an email flag (you'll obtain terribly biased results if you do that, because email subscribers are either your best buyers or they never buy).

I recently performed this analysis for an email marketer.
  • Non-Email Subscribers (after controlling for other factors) had a 35% chance of buying again in the next year (across all channels).
  • Email Subscribers (after controlling for other factors) had a 42% chance of buying again in the next year (across all channels).
  • Non-Email Subscribers (after controlling for other factors) spent $200 in the next year, if they bought again (across all channels).
  • Email Subscribers (after controlling for other factors) spent $230 in the next year, if they bought again (across all channels).
What does this mean?  Well, we can identify the value of an email address, on an annual basis.
  • Non-Email Subscribers = 0.35 * 200 = $70.00 of future demand.
  • Email Subscribers = 0.42 * 230 = $96.60 of future demand.
An email subscriber is therefore worth $96.60 - $70.00 = $26.60 more per year, after controlling for all other factors.

Now, this business typically measures email via opens/clicks/conversions.  They have 100 email campaigns per year, and typically observe performance of $0.12 per email delivered.    On an annual basis, this yields 100 * 0.12 = $12.00 of value.

But this only accounts for opens/clicks/conversions.  It does not count the fact that an email campaign inspired a customer to drive to a store and buy something, or caused the customer to buy via an email campaign, become more loyal, and then buy from a subsequent display ad or catalog.

In other words, the email marketing program is more than twice as valuable as measured via opens/clicks/conversions.

If you were an email marketer, wouldn't you want to know that?