I have been closely following the conversations the APMA is leading regarding the surge in Media Mix Modelling (MMM) within the channel. There has definitely been mixed commentary: I have seen guides explaining the huge advantages of MMM to the channel, alongside real-world examples of the risks posed when affiliate programmes are turned off. It is certainly a tricky and divisive subject.
Wearing my old agency hat, we saw a surge in brands turning to single source of truth platforms; where this might once have been GA4, it was increasingly third-party attribution software and MMMs. While I haven’t directly experienced affiliates being turned off due to MMMs, I have experienced the challenge of proving incrementality throughout my 15-plus years in the channel.
As recently as last year, I had a client turn off the channel in its entirety. After months of discussions on the value of affiliates, we were exhausted and battle-worn, and the only remaining outcome was to turn it off. There are only so many ways you can explain the data! The programme worked with Skimlinks as its only subnetwork, had no voucher sites or browser extensions, and operated with just one cashback partner, closed user groups, CSS, and content partners. It was extremely well monitored and showed strong year-on-year growth in GA4 session-based reporting, indicating it was actively initiating customer journeys.
None of this data was enough. The investment was small compared to other channels, and despite having plans, spend was pulled. The programme was closed on the assumption that they would get those sales anyway. You know what happened? They didn’t get those sales. The gap was affiliate partnerships, and they reopened a new programme less than two months later. It was the ultimate incrementality test, but also one that required a lot of effort and resulted in missed revenue.
The point of this example is that MMM is not a new risk; it is simply a new way of viewing the channel that enables decision-makers to make top-line calls. These types of decisions have been happening for years, but with more brands turning to MMM, and let's be honest, with some MMMs openly favouring other channels (Fospha and TikTok, for instance), we need to be ready.
If we continue to speak merely about 'affiliates' as an all-encompassing term, I fear many programmes won't cut through the noise in MMMs. The call for greater segmentation is 100% the right move; a big part of this must involve identifying the specific partners driving volume, giving them their own face and business identity. Affiliates can easily be dismissed as insignificant in an MMM report when grouped under a single label.
These publishers are major brands and established businesses in their own right. We are talking about cashback sites with millions of members actively choosing to earn cashback as part of their buying journey, closed user groups with huge reach among targeted audiences (audiences brands actively want to speak to, and which affiliates already own), and national newspapers (the very ones coveted via PR, yet easily turned off in affiliate reports). Let's say it again - affiliate partners are often also brands in their own right, and we must fight to ensure they aren't reduced to a faceless channel in a high-level report with zero context.
If you manage an affiliate programme, whether your business is currently looking at MMM or not, spend time getting to know your top 10 partners. Understand who their audience is and how it aligns with your own, know their user base, identify which major brands they work with, and learn how their audience shops.
MMM could potentially highlight great results for certain partner types, but others will need to step out of the shadows to educate, explain, and showcase the wealth of data they hold. The most powerful tool might be the offer of an incrementality test. CSS partners do this quite well (though it is not advised for long-term growth); used to constant questions about their impact on PPC, they will turn off activity to demonstrate value, usually resulting in them being turned back on pretty sharpish. As a publisher, advising on a suggested way to test incrementality could be key to preventing a total switch-off and also provide a way; we, as a channel, advise on the test, the parameters, and what makes it fair.
While this is not a new storm, it is in a new form and one that will likely gain traction alongside fragmented data sources and wider adoption; being ready might just mean being more informed than ever and taking a bold approach to showcasing the true value-driven.
