The Future of Marketing Attribution: Combining MMM and MTA
User-level tracking has shrunk and traditional MMM is too slow on its own. Why the future of attribution pairs mix modeling's rigor with attribution's speed.
There was a time when marketers could track nearly every ad impression and click, following people across websites with considerable precision. That time is over.
Privacy laws, Apple's App Tracking Transparency and the tracking protections built into Safari and Firefox have sharply limited how much of a customer's journey can be followed person by person. Brands rely more and more on the first-party data they collect directly. Google chose to keep third-party cookies in Chrome, but that hasn't brought the old visibility back.
The result is a set of blind spots:
- It's hard to see which ads build awareness.
- It's hard to see the full-funnel effect of a campaign.
Why user-level attribution struggles
For years, multi-touch attribution (MTA) was the standard answer. It tracked each user's interactions and split credit among them. With so much of the journey now untrackable, MTA works from a shrinking and skewed sample, and its answers are less reliable. Our guide to multi-touch attribution covers where it breaks.
The comeback of marketing mix modeling
As tracking faded, the industry turned back to marketing mix modeling (MMM), a long-established method that measures marketing using aggregated data rather than individual tracking.
MMM is privacy-safe and statistically sound. But the traditional version has real drawbacks:
- It's slow. Models took months to build and were often refreshed once a year.
- It isn't built for optimization. Results arrive too late to adjust campaigns while they're running.
- It lacks detail. Without user-level data, traditional models stayed at the broad channel level.
Marketers need something that balances accuracy, speed and the ability to act.
Combining the strengths of MMM and MTA
The answer is to bring the rigor of MMM together with the speed and detail of MTA, without depending on restricted user data. That is the approach Provalytics takes:
- Measurement on aggregated daily data. Daily data across every channel gives the model far more to learn from than annual or weekly snapshots, and it doesn't rely on tracking individuals, so it holds up as privacy rules change.
- Actionable, frequent results. Unlike a once-a-year mix model, results update often enough to guide decisions while campaigns are still in flight.
- Detail below the channel level. Results go down to campaigns and tactics, closer to the granularity marketers expect from attribution.
The future is adaptation
The days of tracking every click and impression aren't coming back. The future of attribution isn't about collecting more personal data. It's about measuring well with the data you can use.
Marketers who stop chasing disappearing signals and focus instead on measuring incremental impact across all their channels will make better, faster budget decisions, whatever the next privacy change turns out to be.
Frequently asked questions
Is multi-touch attribution dead?
Not entirely, but user-level MTA sees much less than it used to. Tracking limits mean many touchpoints, especially on iOS and in browsers that block third-party cookies, never reach the model. MTA is most useful today when combined with methods that don't depend on tracking individuals.
Why is marketing mix modeling making a comeback?
MMM measures marketing with aggregated sales and media data, so it doesn't need cookies or device IDs. As user-level tracking became less reliable, that independence made it attractive again.
What are the drawbacks of traditional MMM?
Traditional models were rebuilt rarely, often once a year, took months to deliver and reported at a broad channel level. That is too slow and too coarse to guide decisions while campaigns are running.
Did Google remove third-party cookies from Chrome?
No. Google reversed its plan to phase them out. But Safari and Firefox block them by default and Apple limits app tracking, so a large share of user journeys still can't be tracked person to person.