For more than a decade, attribution has served as digital marketing’s north star. It told us where to spend, what was working, and who deserved the credit. Clicks became conversions. Conversions became KPIs. And for a while, that was enough.
But somewhere along the way, something shifted. Campaigns started performing better on paper than they did in terms of actual business outcomes. Performance marketers have been underserved, forced to use systems that measure what's easy, instead of what actually matters.
Attribution has long been the default for measuring marketing impact. But most models still rely almost entirely on clicks, ignoring view-throughs, post-purchase behavior, or upper-funnel influence. That means brands end up optimizing for the small percentage of users who actually click, while overlooking the broader drivers of demand. As tracking restrictions tighten and session data becomes increasingly unreliable, it’s getting harder to trust what those models are really telling us.
Most attribution frameworks today offer an oversimplified, often skewed version of the story based on incomplete signals. In a reality where budgets are under pressure and every channel, now more than ever, has the incentive to prove results, that’s no longer enough. To understand what actually drives value, brands need more than an oversimplification of interactions. They need a way to isolate effect. That’s where incrementality comes in.
The Limits of Attribution Logic
Attribution systems are designed to track interactions and assign credit. That makes sense in theory, and for a long time, it worked well enough. But as customer journeys became more fragmented, and as platforms began to report their own versions of the truth, the cracks started to show.
If a customer saw three ads, opened an email, visited your site twice, and finally purchased through a branded search, which touchpoint(s) should get credit for it? Ask your attribution model, and it will answer. Ask your finance team, and they’ll likely raise an eyebrow.
Because attribution doesn’t prove causality. It doesn’t tell you whether that ad actually changed someone’s behavior, or if the conversion would have happened anyway. It merely assumes a relationship based on sequence rather than substance. That’s not measurement. It’s storytelling.
And as every performance marketer eventually learns, narrative isn’t margin.
What Incrementality Testing Actually Does
Incrementality testing offers something attribution can’t: a causal framework. Rather than tracking paths and assigning credit, incrementality isolates the true lift of a marketing activity. It does this by comparing a group of users exposed to a campaign with a statistically similar group that wasn’t. The difference in outcomes, whether that be conversions, revenue, or the number of new customers, is the incremental effect.
This method moves beyond correlation-based storytelling and into causal insights. It doesn’t just ask “did they convert?” It asks, “would they have converted without seeing this ad?” That single shift in thinking reframes how brands understand value, and forces a level of accountability that attribution tends to gloss over.
For a long time, running these kinds of tests meant heavy lifting: manual holdouts, bespoke setups, or waiting for platforms like Meta to run lift studies on your behalf. But modern commerce intelligence platforms now offer built-in incrementality testing, meaning any brand, no matter its size, can apply this logic to its own campaigns on its own terms.
And they should. Because once you see what’s actually driving performance, not just what’s present when performance happens, it’s hard to unsee.
Incrementality Isn’t a Metric. It’s a Shift in Thinking.
At its core, incrementality is about more than measurement. It’s about measuring what truly matters. In a world where customer acquisition costs are rising and where creatives are fragmented across platforms and formats, precision is no longer a competitive advantage. It’s a way to survive.
This matters because media performance today is rarely bad, but it’s often misunderstood. A campaign can show a positive ROAS and still lose money. A top-performing channel might be cannibalising organic traffic. A retargeting ad may look like it’s closing sales, but only because it’s following people who were going to convert anyway.
Without incrementality, all of this gets buried under good-looking reports. With it, you can start to see which campaigns are truly pulling their weight, and which ones are just coasting on momentum.
Rethinking the Role of Marketing Measurement
It’s tempting to think of incrementality testing as a feature, something you toggle on or off. But the reality is that it requires a fundamental shift in thinking. Brands that adopt it successfully often start by acknowledging something hard: that many of their most trusted metrics aren’t telling the full story. This isn’t about distrust. It’s about context.
Attribution will always have a role to play. So will blended metrics and modeled insights. But when you treat incrementality as your north star, those other tools become more useful, not less. They gain relevance because they’re being contextualized against a causal foundation.
And when that happens, marketing starts working differently. Budgeting gets tighter. Campaigns get more focused. Teams shift from defensive storytelling to confident decision-making.
Start with What’s Costing You — and Let That Define Your Measurement Tool
The strongest measurement stacks aren't built by picking the most sophisticated option. They're built by being honest about what you actually need and what your business can act on. The question brands ask most is whether they're big enough to take measurement seriously. That's the wrong question. The right one is whether bad decisions are already costing you. Measurement isn't about scale; it's about the cost of being wrong.
That said, there are hard reasons to rule each method out before going any further. No consistent data across channels? MMM will produce confident answers built on shaky foundations. Spend too thin in a given channel to hold any back? Incrementality testing isn't viable. No tracking baseline at all? MTA has nothing to work with.
If none of those apply, the question is what problem you're trying to solve. Struggling with campaign-level visibility? That's an MTA problem. Confident in your campaigns but unsure how to split the budget across channels? That's where MMM earns its place. Have both covered, but can't tell whether your channels are driving demand or just claiming credit for it? That's a causality problem, and incrementality is the only method that answers it directly.
The strongest stacks use all three: MTA for day-to-day execution, MMM for budget planning, and incrementality as the layer that keeps the other two honest. But most brands don't need all three from day one. Start with whichever problem is actually costing you right now.
The Takeaway
Attribution was built for credit. Incrementality is built for confidence. The question isn’t whether your marketing is working. The question is whether it’s working better than doing nothing. That’s the bar. And if your measurement strategy can’t answer that, it might be time to rethink what you’re optimising for.
At Dema, we believe that clarity beats complexity. That’s why we’ve integrated incrementality testing directly into our commerce intelligence platform, so brands can stop guessing and start growing, with full visibility into what’s really driving results. Attribution says yes. Incrementality says prove it.
.png)
