Multi-touch attribution coverage is eroding now
Path-based attribution degrades quietly, because the model keeps reporting on whatever it can still see. Twenty minutes to measure how much of your own it has lost.
Three platforms, three answers, one budget to defend. Your GA4 export already holds the answer, and the first finding is free and never leaves your browser. Get your first finding free
By Joris van Huët, Founder & CEOUpdated 3 min read
Run the numbers for your store: the free customer journey credit calculator.
A path-based model does not announce that it has lost coverage, it keeps reporting confidently on whatever it can still observe. That is the whole problem: the degradation is invisible from inside the tool.
The twenty-minute measurement
| Pull | For this quarter | For the same quarter two years ago |
|---|---|---|
| Share of sessions with an identifiable source | Note it | Note it |
| Consent acceptance rate | Note it | Note it |
| Share of orders your analytics recorded vs your store | Note it | Note it |
| Unassigned or direct share of revenue | Note it | Note it |
Four numbers, two points in time.
Why the model does not tell you
A multi-touch attribution model distributes credit across the touches it can see. When touches become invisible, the model distributes across fewer of them and reports the same shape of output. Nothing in the interface says "this quarter I observed sixty percent of what I observed two years ago".
The consequence is a slow drift: credit concentrates on the channels that remain observable, which are typically the ones closest to the conversion and the ones running through platforms with their own identifiers. Budget follows credit, and the drift becomes a strategy nobody chose.
What the trend implies
Both the regulatory direction and browser policy have moved consistently in one direction for years. There is no scenario visible in which third-party observability improves. Planning on the assumption that coverage recovers is planning on a reversal nobody has forecast.
That does not mean path models are useless today. It means their useful life is finite and the erosion is not something you will be notified about, so the measurement above is worth putting on a quarterly calendar.
What holds up
Aggregate methods degrade far more slowly, because they compare counts across periods and regions rather than reconstructing individual sequences. Consent refusal reduces precision; it does not remove the comparison. The trade is wider intervals rather than a shrinking sample of observable people, which is covered in cookieless tracking for ecommerce.
The cheap parallel test
Run an aggregate read on a window your current model already reported, and compare.
Where the two disagree most is usually where observability has eroded most, which makes the comparison diagnostic as well as evaluative.
The honest urgency
Nothing breaks on a particular date. What happens is that each quarter your path model describes a slightly less representative slice of your customers, and no alarm fires. The urgency is only that the measurement is cheap and the drift compounds.
Related answers
Frequently asked questions
How do I measure my own attribution coverage erosion?
Pull four numbers for this quarter and the same quarter two years ago: share of sessions with an identifiable source, consent acceptance rate, orders recorded in analytics against orders in your store, and unassigned or direct share of revenue. The direction is the finding.Why does a path model not warn me about lost coverage?
Because it distributes credit across the touches it can see, and fewer visible touches produce the same shape of output. Nothing in the interface reports that this quarter observed less than the last one, so the degradation is invisible from inside the tool.Will attribution coverage recover?
There is no forecast that suggests it will. Regulatory direction and browser policy have moved consistently one way for years, so planning on a reversal is planning on something nobody has predicted. Aggregate methods degrade far more slowly.
Go deeper: Incrementality testing, explained.
Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.
Have an audience of ecommerce operators? Earn 20% recurring commission by recommending the €99 read.
Keep reading
Terms in this article
- AnalyticsAnalytics is the systematic computational analysis of data. It reveals customer behavior and measures campaign performance.
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- ConversionConversion is a specific, desired action a user takes in response to a marketing message, such as a purchase or a sign-up.
- CookieCookie is a small piece of data stored on a user's computer by a web browser, used for tracking behavior, personalizing content, and remembering preferences.
- Multi-Touch AttributionMulti-Touch Attribution assigns credit to multiple marketing touchpoints across the customer journey. It provides a complete view of channel impact on conversions.
- RevenueRevenue is the total income generated by the sale of goods or services related to a company's primary operations.
- SessionA Session is a group of user interactions with your website within a given timeframe. It can include multiple page views, events, and transactions.