Almost nobody hires on first contact. Someone sees a van, searches the company name a week later, clicks an ad, reads the reviews, and calls from the Google listing. An attribution model is the rule that decides which of those moments gets the credit for the job.
Briefly explained
- How credit gets divided among the steps that led to a sale.
- Last click is the default, and it flatters the final touch.
- Every platform reports a different answer, and all of them are partial.
- The model matters less than using one consistently.
What are the models?
You will meet four of them in practice.
- Last click: All credit to the final touch before the call. Simple, and biased toward branded search.
- First click: All credit to whatever created the awareness. The mirror image, with the opposite bias.
- Linear or position-based: Credit shared across the path, either evenly or weighted toward the ends.
- Data-driven: Credit assigned from patterns observed in your own account. The most useful, and it needs volume to work.
Why does last click cause trouble?
Because it rewards the channel that was always going to be last.
Someone who already knows your name searches it and clicks. Last click hands the job to branded search, and the billboard, the article and the display ad that built the awareness show nothing. Cut those, and a few months later the branded searches thin out too, for reasons the report will not explain.
Why will the numbers never agree?
Each platform counts what it can see and ignores what it cannot.
Google Ads claims the Google touches. Meta claims the Meta touches. Both use their own windows, both count view-throughs differently, and neither knows what your CRM knows. Adding their conversions together double-counts the same job.
Pick one system as the book of record. Use the others to diagnose, not to decide.
How do you use this without overthinking it?
- Choose one model and keep it stable long enough to compare periods
- Treat the gaps between models as information rather than noise
- Tag everything that is not auto-tagged. See UTM parameters.
- Ask customers how they found you, and record the answer on the job
- Check the whole picture against the thing that cannot be argued with: total booked work against total spend over a season
What do the terms mean?
- Touchpoint: Any interaction on the way to a sale.
- Attribution window: How long a touch stays eligible for credit.
- View-through conversion: Credit claimed for an impression nobody acted on.
- Data-driven attribution: Credit worked out from what your own account has actually seen.
- Book of record: The one system you let settle arguments.
Owner questions
