By Mehdi HassanWhat is an attribution model?
An attribution model is the set of rules a marketing team uses to assign credit for a conversion to the channels and touchpoints that came before it. A B2B buyer might read a blog post, see a LinkedIn ad, search your brand and then book a demo. The attribution model decides which of those four gets credit, and how much.
It matters because every channel report you read is built on one. Change the model and the same data tells a different story about what's working. It's one of the measurement terms in our SEO and AEO glossary.
What are the main attribution models in marketing?
The main attribution models fall into single-touch models, which credit one interaction, and multi-touch models, which split credit across several. Here they are side by side.
Model | How it splits credit | Good for | Blind spot |
|---|---|---|---|
First-touch | 100% to the first interaction | Seeing what creates awareness | Ignores what closed the deal |
Last-click | 100% to the final click | Seeing what closes | Ignores everything that built trust |
Linear | Equal share to every touchpoint | A simple, fair overview | Treats a footer click like a demo |
Time-decay | More credit to recent touches | Short sales cycles | Undervalues early research |
Position-based (U-shaped) | Most to first and last, rest spread | B2B with a clear start and finish | The weights are arbitrary |
Data-driven | Statistical model of your data | Sites with lots of conversions | A black box, needs volume |
In GA4, Google removed first-click, linear, time-decay and position-based models in 2023. GA4 now offers data-driven attribution and last-click variants, so the rule-based models live mainly in CRMs and dedicated attribution tools.
What does attribution look like on a real B2B site?
On a real B2B site, attribution rarely sees the whole journey. For example, suppose a founder reads a studio's article, sees the studio named in a ChatGPT answer, asks a friend, and then searches the studio's name to book a call. Last-click credits branded search. First-touch credits the article. Neither sees the AI answer or the friend.
That gap is growing. SparkToro found 68% of US Google searches ended with no click in early 2026, though it switched panel data that year, so treat the figure as directional. Research that happens in zero-click search and in chatbots leaves no trail for any model to credit. Our zero-click search strategy covers how to win those buyers anyway.
Hypothetical example: one deal, three models
For example, suppose a deal follows four touches: a blog post, a LinkedIn ad, a webinar and a branded search before the demo request.
First-touch gives the blog post 100%. Last-click gives branded search 100%. A position-based model gives the blog post and branded search 40% each, and splits the remaining 20% between the ad and the webinar. Same deal, three different budget arguments. That's why the model should be chosen on purpose, not left at the default.
Which attribution model should a B2B startup use?
Most B2B startups should use last-click for day-to-day channel reports and pair it with a self-reported "How did you hear about us?" field on every form. That combination is cheap, honest about its limits and catches the AI answers and referrals that tracking misses.
Full multi-touch tooling usually pays off only with long cycles, many channels and enough deals to analyze, which our piece on multi-touch attribution for B2B works through. Keep tracking intact when you rebuild, too: moving from Next.js to Framer or any other platform is when UTM handling and form events break quietly.
