MMM vs multi-touch attribution, which is better?
Usually neither on its own. Marketing mix modeling is better for splitting budget across channels if you have two or more years of weekly sales and spend. Multi-touch attribution is better for steering digital campaigns. Both estimate cause, so check either one with a holdout test.
By Joris van Huët, Founder & CEOUpdated 7 min read
Run the numbers for your store: the free customer journey credit calculator.
Usually neither on its own. Marketing mix modeling is better for splitting budget across channels, even ones nobody clicks, given two or more years of weekly sales and spend. Multi-touch attribution is better for steering digital campaigns week to week. Both estimate cause from history, so check either one with a holdout test.
The two methods read different files. A mix model reads weekly totals: sales, spend by channel, promotions and season. Meta's Robyn guide says it needs no user-level data and runs on aggregated data instead. Multi-touch attribution reads paths: the visits each buyer made before a purchase, with the credit split between them.
In GA4, multi-touch now means data-driven attribution. GA4 dropped the first click, linear, time decay and position-based models in November 2023. What is left is data-driven, paid and organic last click, and Google paid channels last click.
Each method suits a different decision. Google built its open-source mix model, Meridian, for channel questions: past return, response to spend, the next budget split. For anything below channel level, Meridian's own docs send you to data-driven multi-touch attribution on digital channels.
So the useful comparison is not which model is cleverer. It is which question you need answered, and whether your data can feed the method that answers it.
What one store's data shows
One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.
| What the export shows | Share of revenue | Source cell |
|---|---|---|
| Journeys with 1 touch (0.5 days to buy) | 79.5% | Journeys sheet, 1 touch row |
| Journeys with 10+ touches (16.0 days to buy) | 3.0% | Journeys sheet, 10+ touches row |
| Direct, in last click, first click and touched views | 57.7% | Channels sheet, Direct row |
| Touched view, all channels added up | 110.4% | Channels sheet, Touched column total |
This export is path data, the raw material of multi-touch attribution. So it shows how much work a multi-touch model would have at one store.
On the Journeys sheet, journeys with 1 touch hold 79.5% of revenue and took 0.5 days to buy. A one-touch path has nothing to share out. GA4's paths report gives that single touch all the credit, and so would any other model.
Journeys with 2 or more touches hold 12.2% + 5.4% + 3.0% = 20.6% of revenue on the Journeys sheet. That fifth is the only slice where the choice of model can move credit. The 10+ touch row alone holds 3.0% on the Journeys sheet and took 16.0 days.
Count paths instead of revenue and the picture flips. Multi-touch journeys are 3,656 of the 3,670 distinct path sequences in the export. They hold nearly all the variety and about a fifth of the revenue.
On the Channels sheet, Direct holds 57.7% of revenue in last click, first click and touched. A share that stays put across all three views gives a new model nothing to re-split. GA4's models also credit Direct only when a whole path is Direct.
The touched view sums to 110.4% on the Channels sheet, because a journey that touched two channels counts in both. Everything above one hundred percent is overlap, and splitting overlap is the job multi-touch attribution exists for.
For a mix model, the file is simply the wrong shape. It has shares of revenue by path and channel, but no spend column and no weekly totals to regress.
What the export cannot show is whether any channel caused its share. Paths record order, not effect, and a model that splits paths inherits that limit.
Why does the usual answer mislead?
The stock answer says use both: a mix model for strategy, attribution for tactics. It skips two checks. Can you feed both, and does either one measure cause?
Feeding a mix model is the harder half. Robyn's guide asks for at least two years of weekly data. Meridian's guide counts data points per model parameter. Its worked example ends at 156 weekly data points for 10 parameters, roughly 15 each. Meridian says that might give some directional information.
Feeding attribution looks easier, because the paths already sit in GA4. But a path only holds touches that led to a visit, plus some of Google's own ad interactions, such as YouTube engaged views. Robyn's guide calls multi-touch attribution very reliant on online signals. A TikTok or Meta video someone watched and never clicked leaves no step in the path.
Data-driven attribution also wants volume. Google Ads recommends at least 200 conversions and 2,000 ad interactions in a 30-day period for its data-driven model. It still runs with less, and Google says last click and data-driven can then give the same results.
Then the cause question. Both methods learn from what already happened. Meridian's FAQ says observational methods like MMM rely on assumptions to estimate incremental impact. Robyn's guide treats experiments as the ground truth to calibrate a mix model against.
What can neither one tell you?
- Whether a channel caused its sales. That takes a test with the ads switched off for a control group, such as a geo holdout.
- How channels work together. Meridian's FAQ says it does not support that kind of analysis. A path report shows the order of touches, which is not the same thing.
- Anything about a channel that never changed. A mix model learns from spend that moved. A path report learns only from touches it recorded.
What to do this week
- Measure the slice a multi-touch model can move. In GA4, click Advertising, then Attribution paths under Attribution. Under Path length, pick greater than or equal to, enter 2 and click Apply. Pass: those paths hold enough revenue to change next month's budget. Fail: they hold a sliver, so a better model will not change your decision.
- See what data-driven changes. Click Advertising, then Attribution models under Attribution. In the Attribution model (non-direct) columns, set Data-driven against Paid and organic last click and read % Change. Pass: some channels move, so multi-touch tells you something last click does not. Fail: nothing moves much, so spend the effort on tests.
- See if Google Ads can run a lift test. Open Lift measurement in the Goals menu and select the plus button. Choose Conversion Lift under Based on Geo. Pick campaigns and dates, then read the feasibility status before you save. Pass: High. Fail: Low, which Google advises against, so raise the budget or plan a manual geo holdout.
Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework
Sources, 1 October 2026: An Analyst's Guide to MMM (Meta); An introduction to Meridian (Google); Amount of data needed (Google); Meridian FAQs (Google); Get started with attribution (Google); Select attribution settings (Google); Key events attribution paths report (Google); Key event attribution models report (Google); About data-driven attribution (Google); Set up Conversion Lift based on geography (Google)
Related answers
Frequently asked questions
Is GA4's data-driven attribution a multi-touch model?
Yes. It splits the credit for each key event across the touches on the path, so reports can show fractional credit. Since November 2023 it is GA4's only multi-touch option, next to two last-click models.What does multi-touch attribution miss on TikTok or Meta?
Mostly views. A path only holds touches that led to a visit, plus some of Google's own ad interactions. Someone who watched a TikTok or Meta video and later typed your address shows up as Direct. A mix model can still catch that effect in weekly sales, if spend moved.Which should a small store set up first?
Usually neither. Tag your links cleanly and plan one holdout test on your biggest channel. A mix model needs two or more years of weekly data, and multi-touch attribution only re-splits journeys with several touches. A test answers the cause question both of them estimate.
Go deeper: Causal attribution, explained.
Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.
Keep reading
Terms in this article
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- Attribution ModelAn Attribution Model defines how credit for conversions is assigned to marketing touchpoints. It dictates how marketing channels receive credit for sales.
- Control GroupControl Group is a segment of an audience intentionally not exposed to a marketing campaign, used to measure the campaign's true causal impact.
- Holdout TestA holdout test is an experiment where a portion of the audience does not see a campaign. This measures the campaign's true incremental impact.
- Lift MeasurementLift Measurement: A method to determine the incremental impact of a marketing campaign by comparing exposed and control groups.
- Marketing MixThe marketing mix is the set of actions a company uses to promote its brand or product. It traditionally includes product, price, place, and promotion.
- Marketing Mix ModelingMarketing Mix Modeling (MMM) is a statistical analysis that estimates the impact of marketing and advertising campaigns on sales. It quantifies each channel's contribution to sales.
- Multi-Touch AttributionMulti-Touch Attribution assigns credit to multiple marketing touchpoints across the customer journey. It provides a comprehensive view of channel impact on conversions.