Nick Beck dives into how affiliates should focus on measuring incrementality and tracking whether the marketing actually changed behaviour.
The affiliate industry loves calling itself data-driven.
Every deck says it, every platform claims it and every report is packed with dashboards, attribution paths and conversion windows – enough graphs to make the whole thing feel scientific.
But here’s the uncomfortable truth: while affiliate channels often employ sophisticated tracking (including multi-touch attribution and server-side measurement) most still lack the incrementality testing needed to confirm whether the channel truly drove the outcome, or simply witnessed it.
There’s a difference between reporting activity and proving impact. Affiliate marketing has spent years mistaking one for the other.
Last-click attribution is data. Multi-touch attribution is data. Platform reporting is data. But data is not the same as truth.
Owning a dashboard doesn’t make you evidence-led any more than owning a thermometer makes you a climate scientist. You have numbers. That’s not the same as understanding what caused them.
A customer searches for a brand they already know, clicks a voucher code site at the final moment, and suddenly the affiliate gets credited with driving the sale. Did the affiliate create demand? Change behaviour in any meaningful way? Or did it simply appear at the finish line and collect commission for a customer who was converting anyway?
Most brands and agencies are aware of this dynamic – applying lower commission tiers to voucher sites, devaluing last-click attribution, or using contribution modelling to assign a nominal value to late-journey touchpoints. These are sensible mitigations. But they are workarounds for a measurement problem, not solutions to it – and they still fall short of answering the harder question: would that sale have happened without the affiliate at all?
That’s the question the industry has been avoiding. Because once you ask it properly, a lot of the comfortable reporting starts to wobble.
There’s also a structural reason this conversation keeps getting deferred. Platforms and networks have a direct commercial interest in attribution systems that look impressive. Nobody wins new business by showing a client that half their affiliate spend wasn’t moving the needle. So the incentive has always been to produce richer dashboards rather than harder questions.
Measurement theatre is more commercially comfortable than measurement truth.
And AI is about to make the problem significantly worse.
We know that consumers are already discovering, evaluating and deciding inside AI environments before they ever click a link. That content ecosystem is dominated by affiliate publishers, independent editorial and user-generated content. Nearly 60% of Google searches now end without a click to any external website – and that proportion is climbing.
Affiliate content is feeding AI search. But the click that would trigger a commission often never happens.
This means affiliate publishers may be doing more influencing than ever while getting credited for less of it. Influence is increasingly happening in places traditional tracking cannot see.
This is why the rise of the ‘AI search expert’ matters. These experts are focused specifically on how AI systems source, validate and weight content, and how brands and publishers can earn visibility inside these environments.
This isn’t SEO by another name. It requires understanding how large language models construct responses, how generative search engines establish authority, and which publisher relationships position brands well inside AI-driven discovery.
That capability gap is already separating programmes that will lead the next chapter from those still defending fragile attribution logic.
So what’s the honest answer to whether affiliate activity actually worked?
Test it. Not model it. Not infer it. Not rely on platform self-reporting.
Take two statistically similar groups of customers. Expose one to the affiliate activity. Hold the other back. Compare the outcomes. That’s incrementality. Despite how complicated the industry makes it sound, the principle is simple: what changed because the marketing existed?
The good news is that causal measurement methods now make this answerable, and modern approaches can establish this without holding back customers or disrupting live programmes at all. The industry has simply been slow to demand them.
Once you measure this way, uncomfortable things emerge quickly. Some partners drive genuine incremental growth and deserve far more investment than they get – their influence happens earlier in the journey and gets under-credited by traditional models. Others look efficient in dashboards while contributing almost nothing incremental at all.
The irony is affiliates should be leading this conversation. It sits closer to commercial outcomes than almost any other channel. It has rich behavioural data. Testing is entirely possible. But too much of the market still optimises for attribution visibility rather than business impact.
Affiliate marketing now faces a straightforward choice. Carry on defending attribution logic that is becoming less reliable by the month. Or evolve into a channel that can genuinely prove what it does – one that embraces incrementality testing and builds the expertise to operate in measurement environments that look nothing like the ones this industry was built for.
Optimisation is not the same as growth. And being data-driven means very little if the data never answers the only question that actually matters:
Did the marketing change behaviour?
