What influences your customers to buy?
We watched last week as, again, the attribution debate come back to the surface. (You can read a five part series on the subject from April 2012.) First-click attribution to time-decay attribution to last-click attribution to algorithmic smoothing, OH MY!

All of them are techniques to assign value to something that happened, and, that has uses. All models have their uses, and all models, by definition, are too small to fit all the complexity. Understand the uses and use them to set expectations that make sense in that context.

That aside, a set of independent variables in the attribution debate that may be more predictive are the actual influences on customer purchase.

In your sector, in your domain, what influences your customers to buy?

t can’t just be display and search ads alone, can it? Display, especially when combined with video, is an important way to drive awareness and traffic into something. But they’re not the sole influence, are they?

Consider the tablet, saturation in smartphone penetration, e-readers, DVR’s, and the over-the-top-box revolution. The past three years have seen the last of laggards getting onto Facebook. Twitter has gone from sub 1% penetration to 8% penetration. Ratings and reviews (Amazon, Disqus, and so on), mobile search, and, the amplified effect of social, have all intensified.

You should expect it to become more fragmented. You should expect to contend with more screens and more experiences by 2019. You should expect the HTTP cookie to cease being the primary key you use to join things together. (It’s been faulty since at least 2000).

The debate about centered entirely around cookie-deletion attribution models was a debate for the previous decade. We could settle it. Science isn’t decided by committee, so I suspect that there would be many disappointed parties. But it wouldn’t address what’s happening now, and, what we expect to happen in this decade.

How to adapt?

Answer ‘what influences your customers to buy’ first, in your context, and then deploy the dollars. There are enough tools in your kit to predict future performance, optimize during, and assess probable effectiveness after the fact.


I’m Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca