What Direct Attribution Models Miss

Series of events 1: Someone searches using a keyword. They see your adword ad. They click on it. They click ‘order it’ on the landing page. They convert for a net profit of 75 dollars.

Series of events 2: Someone searches using a keyword. They see your adword ad. They click on it. They click ‘about us’ on the landing page. They visit the home page after. They don’t convert right away.

What was the ROI of the first series of events?

Well then, what do you mean by ROI? If we’re talking about the ROI that is directly attributable to the adword ad, then the ROI was 75 dollars. And the second series of events…using the same attribution model, the ROI was 0.

But then, what’s the ROI of awareness then? In the first series of events, somebody goes from awareness to purchase in a short period of time. In the second series of events…what if somebody comes back and purchases? What if they bookmarked your site to return later, and then you wouldn’t be able to attribute it to anything. (Unless of course, you were to use a cookie, but even then, you’d be into an indirect attribution model.)

What if somebody goes in store to purchase? We know that more than 2/3 of North Americans say that they research products before going in store to buy them. To what extend does not having a competent website that lets users achieve their goals, actually hurt the company in question? What of the mitigated opportunity costs of not having a useful website?

Towards a more comprehensive ROI Measure

It’s important that marketers and web analytics practitioners try to understand the direct attribution models as perhaps a starting point in an incrementalist approach. You have to start somewhere, and why not start off with what is most easiest to measure? Aim for the quick hit. At the same time, it’s important to couch such ROI numbers as “ROI that is Directly Attributable to an Online Event”.

Once you do some research into how the website influences your customers, both existing and returning, do you start building out that ROI model. For instance, if you want to improve customer retention, you might want to start off by asking what’s the present value of the customer base. Then, you’ll want to find out the churn rate. Then, you’ll want to calculate what’s the actual cost to the company of that churn rate. Once you know that, you can calculate how much each percentage point of churn reduction will bring in. Knowing that figure will be critical in informing how much you want to spend, what kinds of goals and expectations to set around an initiative.

Direct Attribution Models miss a lot. However, they’re great starts.