Findings on Conversion Attribution

It’s going to be a short, hopefully informative post today.

First, some background. US online ad spending increased to $21.2 billion last year; but within the ontology of online advertising, paid search is faring best. Indeed, many are projecting that paid search advertising growth might truly be recession-proof, whereas the revenue for online publishers selling display advertising against their content may shrink considerably if the economy continues to turn sour.

At the same time, Atlas and others are trying to push the conversion attribution needle further back along the brand encounter, using technologies such as engagement mapping and attribution management to cut down on the amount of credit accruing to search. In essence, they are granting that Google/Yahoo/Live search will be the final gateway (the “last-click”) through which the prospective purchaser will pass, but they are arguing that the moment of persuasion has already occurred as a result of exposure some other media vehicle (be it display, TV ad, Facebook app), which preceded the search.

I’m not going to take a side on this issue just yet, but I will share some data. I looked at two questions in a study we are running for a major multi-channel US OEM, which does a supersized volume of business online. I looked at visitors who purchased during their sessions and who had never made a previous online purchase at this company's site; I did this to neutralize the impact of brand loyalty as much as I could. I looked at the share of conversions by referring site, and then I looked at the share of conversions by greatest purchasing influencer.

If we attribute on the basis of last-click, search is getting 14% of the credit. If we attribute on the basis of greatest influencer, however, the credit assigned to search falls by over 60%. At the same time, credible human referrals rise to the top in terms of importance, and good old fashioned vehicles like direct mail and print advertising start looking pretty effective. From question to question, there is no change in the attribution share for display ads, though.

Is there a case to be made for multi-channel engagement mapping?

Comments