I’ve been thinking about attribution recently — how your software decides which channel gets credit for conversion (see my article on campaign analytics). Attribution is really a question of credit and influence. Which promotion contributed to that sale?
I believe that previous ad clicks contribute to brand awareness and can drive future searches and response. But, they place a lesser degree in future sales. Yes, I’m more likely to come back and convert if I previously found your site another way. They deserve a some assistance credit, but it diminishes over time and with each subsequent click and channel.
Direct traffic is the biggest attribution hole. All of that traffic originated somewhere, we just can’t directly trace it back and it’s deflating other channel’s performance.
I think this is a religious argument. There is no obvious and easy answer. The best answer is the model that ultimately leads a business to invest in the channels that produce more profit, etc. That said, we are very far off from that state. That’s why I think that software should give people a choice, thus circumventing any nits they could pick with whatever one we forced them into.
What kind of attribution are you using today? Which kind would you like to be using?
{ 3 comments… read them below or add one }
I’m very glad to see this topic being raised because it gets ignored too often by both vendors and analysts. (Some) analysts don’t think about it, and (some) vendors just pretend it isn’t there.
I’ll go out on a limb and say that a one-size-fits-all attribution weighting scheme is about as bad as no attribution weighting scheme. And guessing about the weights is little better.
Academic marketing research has looked at the question of weights and how they can be derived from the data using multivariate methods. Once they’re derived from data, they can be applied going forward with a lot more confidence than arbitrary schemes can.
It’s unfortunate how we in web analytics, with our little tabulations, are ridiculously far behind statistical techniques that have been out there for decades!
Hey Chris,
That’s a great point. I don’t know about you, but most of the web analysts I know stumble into the field and don’t come from a statistical background. The same is true of consumers of web analytics data — marketers and brand managers.
I know that Google Analytics does last click attribution, but I’m not sure about the other tools off the top of my head.
Have you seen anyone tackle this problem effectively by any of the vendors? I’m really curious if anyone’s actually used the kind of statistical techniques you’ve discussed. It seems easier said than done.
-Alex
Hi Alex,
While I’m not a practitioner, I’ve conducted substantial research on the topic of attribution. My data shows that 31% of marketers are attempting to calculate for attribution in some way, but the extent to which this is happening is entirely unclear.
Stay tuned for a research report that I authored for Forrester on this topic as well as one additional white paper (that I know of) to be published soon. 2009 is the year for attribution adoption. I believe that within the next 18 months attribution will be a corporate mandate. Get started now.
Cheers,
John