Earlier this year I was asked by Julia Stent to read through a white paper on Payment on Influence. The research, carried out by Vodafone, looks at how affiliates could be rewarded for their part in the sales process as an alternative to the traditional last click wins the sale that’s currently in place.

You can read the white paper at affiliates4u along with a brief overview of the payment on influence work. It’s well worth reading!

The research and information presented is quite interesting and could pave the way for changes in the attribution model. In all honesty, having been concerned by a presentation customer journey cookie attribution at a4uexpo, I wanted to find holes galore in the analysis, but all the questions I thought of as I read through are well covered.

Few stones are left unturned.

However, my main area of concern remains how cashback sites would embrace such a model. Under the attribution model (which I’ve dubbed the Blockbuster Attribution Model because of the image below), the final click affiliate can earn between 60% to 100% of the commission available. This means a cashback site can develop their models to account for this (i.e. you will earn at least £x based at 60%).

But, cashback users are often encouraged to clear cookies and it only takes a message on a money saving board that “if you do this you get 60% cashback, but do the other and you get 100%” to put the reasoning behind the model, i.e. to reward those involved in influencing journey, in some jeopardy. It’s at this juncture that caution would need to be exercised under this model if was to be introduced.

Also the percentage of cashback users that may clear cookies before completing a purchase isn’t known, but in the research it’s shown that only 14% of sales from loyalty sites have a sole referrer interaction, but 35% are a sole referrer transaction. The reasoning given is “a customer either visits a loyalty site and purchases after a single click” or they click back and forth. The other or, they delete cookies and therefore wipe out the journey, isn’t mentioned!

Of course, the above also hinges on whether I’ve understood things correctly!

But for me the question now isn’t do cashback users delete cookies before making a transaction (imo some will do), but what effect does this have on the overall data analysis? I’m loathe to say it, but my gut feeling is it’s probably won’t make much difference to the overall basis and reasoning of the model.

So will this replace Last Click Wins?

In my opinion affiliates shouldn’t be starting to worry that BAM is going to become the norm across the board. The data only covers one merchant, in one sector and one quarter sales, so there’s more number crunching to be done! It may well be that it’s good for one kind of merchant (mobiles) but not for another (gifts).

The data presented seems good news for both affiliates and merchant, but what happens at different sale periods? Data needs to be presented for a whole year to show the variation between final click, dedupe and BAM (sorry Payment on Influence). The data shown suggests if Vodafone moved to a dedupe model it would knock 13% off affiliate earnings, whilst the influence model reduces that to 0.9%. How constant would that be across a full year?

I’d also like to see an effect on income of actual affiliates. I don’t know whether that is feasible to do, but I think a move to an attribution model could be better demonstrated if it showed real life events – i.e. the percent variation in revenue depending which scheme was used. How will 100% cashback site fair? Does Mr Codes get more or less? And is it good or bad news for the poor duck people?

Overall the study in multi attribution is a well put together piece of research. It’s a very good platform from which to explore the paid on influence model and it will no doubt open up the debate about moving affiliate marketing away from the last click wins scenario.

Payment on Influence – A Study in Multi Attribution