Measurement
Unified marketing measurement: four ways to measure, one number to spend against.
Unified marketing measurement means running several measurement methods together rather than picking one. Attribution, marketing mix modeling and incrementality testing answer different questions — Dema runs them side by side and reconciles the answers into one number you can budget against.
Ad platform attribution
What Meta, Google and TikTok claim. Fast and granular, but each platform grades its own homework.
Multi-touch attribution
Credit split across recorded touchpoints. Useful for journey shape, blind to anything untracked.
Causal attribution
Weights that reconcile platform and MTA claims against experimental evidence, per channel and market.
MMM + incrementality
Cookieless. Aggregate modelling anchored by geo-based experiments that prove causal impact.
Every attribution method is wrong in a different direction
Ad platforms over-report, because each one counts conversions it can plausibly claim. Multi-touch attribution only sees touchpoints that were tracked, so it misses everything cookieless. Marketing mix modeling covers the whole mix but can't answer channel-level questions on demand. Incrementality testing is the only causal proof, but you can't test every channel every month. Unified measurement doesn't pick one — it reconciles them.
Ad platform attribution
Each platform reports the conversions it can plausibly claim, using its own rules and its own attribution window. Fast, granular, and structurally optimistic — Meta and Google will both count the same order. Useful for in-platform optimisation, unreliable as a source of truth.
Multi-touch attribution (MTA)
Divides credit across the touchpoints that were actually recorded. Good for understanding journey shape, but blind to anything untracked — and as consent rates fall and cookies disappear, the share it cannot see keeps growing.
Three methods, three jobs
Each of these answers a question the others cannot. Unified measurement is what makes them agree.
Causal factor attribution
Decides how much of each platform and MTA claim to believe, using weights derived from your incrementality tests. Set per channel, funnel stage and market, and reported on contribution margin.
Marketing mix modeling
Continuous, cookieless measurement of the whole mix from aggregate data. Response curves show where the next euro works hardest, and the optimiser runs on contribution margin rather than revenue.
Incrementality testing
Geo-based experiments that prove what spend actually caused. The only causal evidence in the stack, which is why its results calibrate both the model and the attribution weights.
Cookieless by design, not by patch
Marketing mix modeling and incrementality testing both work on aggregate data. No cookies, no user-level tracking, no consent dependency — so they keep working as tracking restrictions tighten, and they measure the channels attribution never saw in the first place.
Tests feed the model. The model guides spend. Both correct your attribution.
Incrementality results do two jobs: they anchor the marketing mix model with causal ground truth, and they set the weights causal factor attribution uses to discount platform and MTA claims. Run more tests and both get sharper. Causal attribution and MMM are where you end up — platform data and MTA are just inputs on the way.
Start with the fundamentals
Incrementality testing is the causal evidence the rest of the stack is calibrated against. Our complete guide covers treatment and control groups, the four test designs, reading significance, why results should be read on profit rather than revenue, and what incrementality testing cannot tell you.
Trusted by category leaders
Frequently asked questions
Unified marketing measurement means running several measurement methods side by side rather than picking one and defending it. Attribution divides credit among recorded touchpoints. Marketing mix modeling measures the whole mix from aggregate data. Incrementality testing proves causality by experiment. They answer different questions, so unified measurement reconciles their answers into one number — which means the result degrades gracefully as any single method loses coverage, instead of quietly going wrong.
It depends on which of the methods you actually need and who has to use the answer. Tools built around attribution are strongest on in-platform decisions; tools built around marketing mix modeling are strongest on budget allocation; and only some cover incrementality testing to calibrate either. The harder question is what the outcome is measured on — most platforms model incremental revenue, and revenue and contribution margin frequently disagree about which channel is working. We keep an honest comparison of the main options, including where competitors are better than us.
Multi-touch attribution divides credit between individual touchpoints it recorded, so it is granular but limited to tracked journeys and biased by whatever each platform reports. Marketing mix modeling works on aggregate spend and outcome data, so it covers the entire mix including channels with no click tracking at all, but it cannot answer a specific question like “did this campaign work last Tuesday”. They answer different questions; the mistake is reading one as though it answered the other.
Attribution divides credit among conversions that already happened, so it can only describe what it observed. Incrementality testing withholds or adds spend in some regions and measures the difference in actual outcomes, which answers whether the spend caused anything at all. It is the only one of the methods that produces causal proof, which is why its results are used to weight the others.
It is the layer that decides how much of each attribution claim to believe. Rather than accepting the ad platform's number or the MTA number, causal factor attribution applies a weighting — for example 30% of the platform's claim and 70% of MTA's — with the weights derived from your own incrementality tests, or from Dema's platform-wide benchmarks before you have run any. Weights are set per channel, funnel stage and market, because reporting bias is not uniform across them.
Yes. Marketing mix modeling and incrementality testing both run on aggregate data, with no user-level tracking or consent dependency. Ad platform data and multi-touch attribution are still used where available, but they are inputs to be weighted rather than the source of truth, so measurement does not collapse when tracking coverage drops.
They are stronger together. Incrementality tests answer specific causal questions but cannot cover every channel every month. MMM gives continuous full-mix measurement, and incrementality results calibrate the model so it stays accurate. In practice the tests are what keep the model honest.
Most tests run 4-6 weeks including a two-week post-treatment observation period. Setup takes minutes: pick a channel, country and campaigns, and Dema handles the geo-split, power analysis and statistical evaluation.
Connect your ad platforms (Google, Meta, TikTok), your e-commerce platform, POS or store systems, and your cost data. Dema builds the unified data model automatically. Most teams are set up within a day.