Marketing mix modelling (MMM) is increasingly influencing how brands understand marketing effectiveness – and, crucially, where they put their budgets. For affiliate marketing, that presents both an opportunity and a risk.
This summer, the APMA surveyed 67 brands and advertisers to understand how widely MMM is being used, how affiliate is represented within those models and what impact the results are having on investment decisions.
The findings point to a measurement challenge the industry can no longer ignore.
Almost 60% of respondents are already using MMM, either alongside attribution (46%) or as their primary budgeting tool (13%). A further 13% are developing a model.
And these models carry weight. More than half (52%) of respondents described MMM’s influence over marketing budget decisions as either moderate or very high.
And whilst MMM is becoming increasingly important, our research suggests affiliate isn’t always being represented in a way that reflects the complexity of the channel.
That was the focus of a recent APMA and Acceleration Partners webinar, hosted by Helen Southgate (Acceleration Partners), featuring Amy Johnson (TUI), James Little (TopCashback) and Kevin O’Farrell (Analytic Partners), bringing together advertiser, publisher and measurement perspectives.
First: what exactly is MMM?
At its simplest, marketing mix modelling uses historical data to understand what’s driving changes in business performance.
Sales can rise or fall for hundreds of reasons: marketing activity, promotions, pricing, the economy, seasonality, weather or changes in consumer behaviour.
MMM uses statistical modelling to separate those different influences and estimate the incremental impact of individual activities.
That makes it a powerful tool for marketers. It can help businesses understand which activities are genuinely driving results, where they should invest more and where budgets might be better spent elsewhere.
But there’s an important caveat: the model is only as good as the data behind it.
And that’s where affiliate has a problem.
The good news: affiliate is visible. The problem is how it’s represented.
Our survey found that 70% of respondents measure affiliate as a distinct channel within their MMM. Just 8% said it wasn’t distinct, while 5% grouped it with other channels.
On the surface, that’s encouraging. Affiliate has a seat at the measurement table. But being visible doesn’t necessarily mean being understood.
The most common way of representing affiliate – reported by over a third of respondents – is as one single channel, with no split by publisher type or individual partner.
By comparison, 13% split affiliate by partner type, 10% by individual partners and 19% by both partner type and individual partner.
That’s a problem because ‘affiliate’ isn’t one thing.
A content publisher introducing a consumer to a brand operates very differently from a cashback platform, voucher site or loyalty programme. Even two publishers within the same broad category may have very different audiences and play very different roles in the customer journey.
Yet when all that activity enters a model under a single ‘affiliate’ line, those differences can disappear.
And brands aren’t convinced the answer is accurate
That lack of granularity matters when we look at what the models are actually saying about affiliate.
Just 12% of respondents said their latest MMM showed affiliate making a strong contribution – better than other digital channels.
More than a fifth (22%) said it showed a weak contribution, while 37% reported a moderate contribution comparable to other digital channels.
Perhaps more tellingly, the people using these models aren’t necessarily convinced they’re getting the complete picture.
When asked directly whether they believed their MMM measures affiliate accurately, only 9% said yes. Another 24% said ‘mostly’, but 28% said no and 18% were unsure.
So we’re in an unusual position: MMM is increasingly influential in marketing decision-making, while many of the people closest to affiliate aren’t convinced the channel is being measured accurately within it.
This isn’t theoretical – it’s already affecting budgets
This is where the measurement debate becomes a commercial one.
In the last 12 months, 18% of respondents said MMM findings had resulted in affiliate budgets being reduced, while one respondent said their affiliate programme had been paused or closed.
By comparison, 16% had seen affiliate budgets increase because of MMM. The largest group, 34%, reported no change.
And there is little expectation that the issue will disappear.
Looking ahead, 18% expect MMM findings to decrease affiliate investment, compared with 15% who expect them to increase it. More than a quarter (27%) said it was still too early to tell.
The point isn’t that MMM is inherently bad for affiliate.
It’s that measurement decisions are becoming investment decisions. If the inputs don’t properly reflect what affiliate does, the consequences can be significant.
Affiliate has a data problem
One of the clearest messages from both our research and the webinar was that better measurement starts with better data.
When we asked brands about their biggest affiliate measurement challenges, incrementality emerged as the number-one issue, selected by more than half of respondents.
But it wasn’t the only one.
Brands also highlighted difficulties measuring upper-funnel, assisted and cross-channel journeys; understanding affiliate’s role within AI search and emerging customer journeys; and reconciling conflicting results between MMM, attribution platforms and affiliate reporting.
Around a third said those different measurement systems can produce conflicting results. Others pointed to insufficient data, affiliate being undervalued within measurement systems and the channel being treated as a single entity rather than differentiated by publisher type.
The open-text responses reinforced the same point. Among the 34 respondents who explained their biggest challenge in more detail, problems with the underlying data were the most commonly identified theme.
To date, affiliate measurement has largely revolved around clicks, conversions and last-click sales. But that doesn’t necessarily capture the full influence publishers have on consumers.
A publisher might expose millions of consumers to a brand through onsite placements, emails, apps, social content or paid media without those interactions appearing in traditional affiliate reporting.
Historically, publishers haven’t necessarily been asked to provide that type of data. Perhaps now they need to be.
Better data – but also more consistent data
Simply collecting more data isn’t enough.
Affiliate is fragmented. Different networks use different technologies. Publishers operate different business models. Tracking implementations vary. Consent rules differ.
Even apparently straightforward metrics need common definitions.
If a consumer sees two different placements for the same advertiser on one publisher page, for example, is that one impression or two?
As James highlighted during the webinar, the industry needs to establish what should be counted and how that information can be provided consistently before simply pushing more data into measurement models.
Other marketing channels have established common currencies and accepted approaches to measurement. Affiliate has an opportunity to move towards something similar.
And our survey suggests brands want the industry to help.
Brands are asking for practical solutions
When we asked respondents what the industry could do to improve affiliate measurement, the strongest message was a desire for practical help rather than another measurement theory exercise.
Best-practice guidance for how affiliate should be represented within MMM and attribution was the most frequently mentioned area.
Respondents also asked for help with incrementality testing, better data provision from networks and publishers, and resources that affiliate teams can use internally – including education, case studies, benchmarks and industry norms.
This points towards an important opportunity for the APMA and the wider industry: establishing a clearer picture of what good affiliate measurement actually looks like.
That could mean common taxonomy. Consistent definitions. Guidance on the data that should be available. Agreed approaches to incrementality testing. And potentially an industry ‘gold standard’ against which brands can assess their own measurement.
MMM shouldn’t be the only lens
Better MMM doesn’t mean MMM should become the only answer.
At TUI, Amy Johnson explained that MMM sits alongside other forms of measurement, including internal attribution, impact reporting and individual partner testing.
That matters because every methodology has limitations.
MMM can provide an important strategic view across an organisation, while targeted experiments can help teams understand individual publishers, placements or customer behaviours in much greater detail.
The opportunity is to combine those different lenses to build a more complete picture.
As Amy explained, improving TUI’s approach has been a journey: working directly with its MMM provider, helping them understand the nuances of affiliate and exploring what additional data can be fed into the model. Each iteration has helped improve the output.
So, what should the industry do now?
The message from both our research and the webinar is clear: waiting for MMM to get better on its own isn’t an option.
Brands, publishers, networks and measurement providers all have a role to play in improving how affiliate is represented.
And with MMM already influencing budgets, this needs to happen before the next set of results lands – not after.
For brands: get affiliate into the measurement conversation
If you’re responsible for an affiliate programme, find out how your business is measuring it.
Don’t wait until an MMM output lands on your desk or a budget decision has already been made.
Affiliate teams should:
- Understand how affiliate enters the model. Is it one line? Is it split by publisher type? Are individual partners represented where there is enough data?
- Build a relationship with your MMM or analytics team. Help them understand how your programme actually works, rather than assuming they already know the nuances of the channel.
- Challenge the inputs, not simply the outputs. If a result doesn’t reflect what you’re seeing elsewhere, understand what data has gone into the model and what’s missing.
- Use multiple measurement lenses. MMM should sit alongside attribution, incrementality testing, partner-level analysis and other evidence – not replace them.
- Bring publishers into the conversation. Before reducing investment because a partner or publisher type appears weak, ask what additional data they can provide and whether the conclusion can be tested.
Amy’s experience at TUI shows why that involvement matters. By working directly with its MMM provider, explaining the nuances of the programme and continually improving the data being supplied, the affiliate team has been able to improve the model with each iteration.
The biggest risk for affiliate teams isn’t MMM itself. It’s allowing decisions about the channel to be made without their expertise.
For publishers: help to prove value beyond the click
Publishers also need to recognise that the data brands require is changing. Clicks and conversions remain important, but increasingly they won’t be enough on their own.
Publishers should start asking: what can we show an advertiser about our influence before the final click?
That could include exposure and impression data, onsite engagement, email activity, app interactions, paid social activity or other signals that demonstrate how consumers are interacting with brands through publisher environments.
Publishers should:
- Audit the data you already have beyond clicks and conversions.
- Identify the exposure and engagement metrics you could realistically share with advertisers and networks.
- Be prepared to support incrementality testing and other experiments that help demonstrate the additional value you create.
- Talk proactively to advertisers about measurement, rather than waiting until budgets are challenged.
- Work with networks and the wider industry on consistent definitions and data standards, so brands aren’t receiving completely different signals from every publisher.
As James argued during the webinar, publishers haven’t historically been asked to provide many of these signals. But if providing better data can help secure existing budget or unlock more investment, publishers have a very clear reason to engage.
For networks: help make richer data usable
Networks have an important role in bridging the gap between publishers and advertisers.
If every publisher measures and supplies exposure data differently – and every network has a different way of receiving it – the industry risks creating more complexity rather than solving it.
Networks should help establish common definitions and practical ways for publishers to pass richer data through the affiliate ecosystem, so that it can ultimately be used by brands and their measurement partners.
The goal shouldn’t simply be more data. It should be more useful, consistent and comparable data.
And as an industry: create a common measurement language
Neither brands nor publishers can solve this individually.
Affiliate’s fragmentation is part of what makes the channel powerful, but it also makes consistent measurement difficult.
We need greater agreement around what we measure, how we define it and how that data moves between publishers, networks, brands and measurement providers.
Our survey shows there is appetite for exactly that. Brands asked for best-practice guidance on representing affiliate within MMM and attribution, support with incrementality testing, better publisher and network data, and benchmarks and industry norms they can use internally.
That’s where the APMA has a role to play.
Our next step is to work across the industry to establish what good affiliate measurement looks like: clearer taxonomy, more consistent data definitions, practical guidance and frameworks that brands, publishers and measurement teams can actually use.
The industry has a measurement problem – but it’s one we can solve
Kevin’s final message summed up the discussion:
“MMM is not the enemy of affiliates… poor data is.”
Better inputs should produce better measurement. And better measurement should lead to fairer investment decisions.
So the immediate actions are clear: brands need to get affiliate teams into the MMM conversation; publishers need to provide better evidence of their influence beyond the final click; networks need to help make that data consistent and usable; and the industry needs common standards for what good affiliate measurement looks like.
The APMA will continue working with advertisers, publishers, networks and measurement providers to help establish that common ground.
View the charts and graphs from our brand survey here
