March 16, 2016

But If I Cannot Attribute New Orders Properly, I'm Sunk, Right?

Wrong.

Attribution is a mirage. We've all done it wrong for decades and managed to grow businesses to ten percent pre-tax profit.

When I analyze an e-commerce-only business, I get to see unique and interesting trends.

For instance, it frequently takes a half-dozen visits to harvest a first purchase. Those visits tend to come from:
  • Social Media.
  • Facebook Advertising.
  • Google / Paid Search.
  • Email Marketing and/or Company Blog.
  • Direct Load.
  • Visit Resulting in Purchase.
Now, you give this problem to 10 attribution vendors, and you'll get 10 different answers. Every answer is right. Every answer is wrong. Worse, give this problem to Google, and they'll prove that advertising works and therefore they'll prove that you should spend more money, and then they'll ask you to use Google Analytics to prove that digital marketing works.

I'm not saying you shouldn't use attribution algorithms. If anything, I'm saying you should pay for three of them and have your in-house team execute a fourth routine and then average the results.

But what I'm saying is more important ... work every source you have ... work the low-cost or no-cost sources VERY HARD ... and seek to obtain new customers as a function of all activities, not as a function of a handful of optimized channels. When I worked at Nordstrom, we generated more than 10% pre-tax profit and acquired more than a million new buyers in stores each year and couldn't attribute almost any of them to any one marketing activity ... and because we didn't spend more than a percent or two of net sales on marketing, the new customers just kinda rolled in because we focused on the important stuff.

The same thing happens in your business ... many vendors don't want you to know this, of course, because they don't get paid. If you work your platform and low-cost / no-cost channels VERY HARD, you will get a ton of new customers independent of marketing activities. Focus on getting the details right, and the return on investment takes care of itself. Then you can have an argument about which attribution algorithm gets closest to your biased version of the truth.