Incrementality tests revealed profitable headroom on Google.
Attribution checked against your own tests.
Dema corrects what ad platforms report with your incrementality tests, so each channel gets credit for the sales it actually caused.
Platform-reported revenue
€2.84M
Causal revenue
€1.71M60% of reported
Tests behind the calibration
74 geo, 3 holdout
Causal ROAS
2.1×3.5× reported
Reported and causal revenue, weekly
Reported Causal
Calibration by channel, ROAS reported → causal
- Meta3.4× → 2.0××0.60Geo test, Nordics, March
- Google Ads4.8× → 1.9××0.40Geo test, Germany, May
- TikTok1.6× → 1.5××0.92Holdout test, UK, August
- Email6.1× → 3.7××0.61Holdout test, all markets, June
From reporting to reality
The sale is real.
The credit needs work.
An ad platform records a conversion. That doesn’t tell you whether the customer would have bought anyway. MTA sees touchpoints, but can miss the influence it couldn’t track.
Calibration uses incrementality evidence to adjust that credit. Your total revenue stays the same. The share attributed to each channel changes.
Same business. Same €100,000.
Before calibration€100,000
€50k
€50k
After calibration · ×0.60€100,000
€30k
€70k
Meta · beforeMeta · calibratedOther selected channels
Grounded in experiments
A factor you can
actually explain.
Start with Dema’s experiment benchmark. Bring in tests from your own business. See the evidence behind the estimate, including the uncertainty.
Then choose your calibration factor. It stays explicit, with a separate setting for each channel, funnel and market that needs one.
How incrementality testing worksEvidence behind the factor
Meta · Prospecting · Sweden0.60×
A range of evidence.
One explicit setting.
Evidence distributionYour experimentsChosen factor
Your calibration, in plain sight
Different channels.
Different corrections.
Choose the reporting source. Review the evidence. Set the multiplier. Explore the example below to see exactly how a channel’s attributed revenue changes.
Calibration settings
Each channel, funnel and market can have its own calibration.
Meta / ProspectingSweden
Calibration base
Reported revenue€50,000
×Factor0.60
=Calibrated revenue€30,000
0.60×
0.10×1.50×
Retain 60% of the reported revenue for this channel. A different base needs its own calibration.
Below 1× adjusts credit down. Above 1× adjusts it up.In Dema, review the experiment evidence before choosing a factor.
Bring it back to profit
Same revenue.
A different return.
Revenue is only part of the decision. Product margins, fulfilment and returns change what you keep. Read calibrated performance alongside contribution after advertising.
Full-price purchases
Higher margins. Fewer returns.
Discount-led purchases
Lower margins. More returns.
Both examples have €30,000 in calibrated revenue. Contribution is after product, fulfilment and return costs, before ad spend.
Plan the wider mix with MMMTrusted by category leaders.
Questions
A little more clarity.
Causal attribution uses evidence of incremental impact to calibrate recorded attribution. In Dema, causal factor attribution applies an explicit multiplier to a chosen baseline: ad-platform reporting or multi-touch attribution. Experiments help you assess that multiplier. The result is an evidence-informed allocation of channel credit, rather than treating a recorded touchpoint as proof that an ad caused a sale.
Multi-touch attribution divides credit among the touchpoints it can observe. It does not, on its own, establish what would have happened without the advertising. Dema can use MTA as the baseline, then calibrate its attributed value using incrementality evidence. You can also choose the ad platform as the baseline. The appropriate multiplier depends on which base you choose.
Dema shows a distribution informed by finalized incrementality experiments, together with matching experiments from your own business. Its Bayesian model combines benchmark evidence with your results. You review the distribution and experiment evidence, then choose an explicit multiplier for the relevant channel, funnel and market. A benchmark is a starting point; it is not the same as a test of your own business.
It means the calibrated channel value is higher than the selected baseline. For example, €20,000 multiplied by 1.20 becomes €24,000. This can be appropriate when the evidence suggests that the baseline under-credits a channel. The multiplier is relative to the chosen base, so a factor for ad-platform reporting cannot simply be reused for MTA.
No. It changes how revenue is attributed. When calibration reduces paid-channel credit, the difference is redistributed to the channel groups selected in your settings, such as Direct, Backfilled or Search Organic. Total revenue remains balanced. The €50,000 → €30,000 example changes channel credit; it does not remove €20,000 from the business.
The multiplier applies to the selected attributed value. When you calibrate conversion counts and keep spend fixed, CPA moves in the opposite direction: 50% of the conversions means twice the CPA. For profit decisions, look beyond revenue to contribution after product costs, fulfilment, returns and advertising. Equal attributed revenue does not imply equal profit.
Incrementality tests estimate what changed because of advertising and help inform calibration. Causal factor attribution applies that evidence to channel reporting. Marketing mix modeling takes a broader view of the mix and supports budget planning across channels. They work together, rather than being interchangeable measurement methods.
Bring one commercial job