Analysing the true value of each affiliate that contributed towards a sale is not a simple task. Great effort has been put in by a team from Vodafone, Top10 and Affiliate Window, lead by Julia Stent, to analyse the Vodafone affiliate program in an effort to develop a better system compared to the existing last click model.
Attribution models for affiliate marketing is not a new thing – as is highlighted in the report, it has been the talk of the industry for at least the past three years. The complexity of the analysis, tracking, and commitment needed to implement has made it a mammoth of a undertaking. While the report doesn’t aim to provide a one-size fits all solution, it’s a good read that can provide a framework for developing models specific to individual affiliate programs.
Without looking to dissect the numbers or the analysis, a couple of things are left untouched in the report that I think are worth discussing:
1) Arbitragers and spammers will love the attribution model
Since affiliates will get remunerated by simply being part of the buying cycle, they will be increasingly incentivised to purely cookie drop. With today’s last click model, cookie dropping affiliates that are not adding value by even being the last click are effectively being overwritten (…but so are the one’s that add real value…). While the networks and merchants can look at their tracking systems to analyse referring traffic quality, by using at a five-step attribution model, there’s five times the amount of data to trawl through compared to today’s model.
2) The model doesn’t consider the cost associated with each touch point
As the report highlights, the last click is still considered the most important one, and is therefore up-weight to always be remunerated with at least 60% of the commission. However, not all clicks cost the same. If an in-house online marketing department would optimise its campaigns across the same channels as via affiliate (treated as one channel), it would also have to closely look at the cost per conversion by channel and optimise with this in mind to maximise their marketing budget (e.g. generic search CPC compared to branded terms, email, display click etc…). By treating all affiliate types the same, this optimisation is lost in the process of dividing the commission purely based on where in the path they contributed to the sale. With varying costs, the affiliates will need different levels of commission to maximise their impact.
Saying that, I understand that the report aims to provide a simple framework and avoid to over-complicating things. I think it’s clear that an attribution model required to further improve the affiliate potential, and this is a very good effort at making a real difference to the industry. Hopefully the model will come to fruition as I’m looking forward to seeing what it would result in for the Vodafone affiliate program.
>> Download the full report (pdf) or read Julia’s post at a4u.
What’s your experince with multi-touch attribution models?
Related posts:
