Every attribution model looks at the exact same customer data and tells a slightly different story. None of them is "correct" in an absolute sense — each one optimizes for a different question. The mistake isn't picking the wrong model; it's not knowing which question you're actually trying to answer.
| Model | How credit is split | Question it answers |
|---|---|---|
| First-touch | 100% to the first interaction | What creates awareness / brings in new people? |
| Last-touch | 100% to the final interaction before conversion | What's there the moment someone finally decides to buy? |
| Linear | Equal split across every touchpoint | A simple, defensible baseline when you don't want to argue about weighting |
| Time-decay | More credit to touchpoints closer to conversion | Which recent touches mattered most, when recency itself is meaningful |
| Position-based (U-shaped) | Large fixed share to first + last (e.g. 40/40), rest split across the middle | Balancing 'what found them' against 'what closed them' |
| Data-driven | Calculated per-account from actual conversion patterns | The most accurate picture, given enough data to calculate it from |
First-touch and last-touch are covered in full depth, with a worked example, in first-touch vs. last-touch attribution. Linear, time-decay, and position-based are covered together, alongside data-driven, in multi-touch attribution. This article is the decision layer on top of both — how to actually pick, given your specific situation.
It's worth noting that Google Ads itself moved away from offering first-click, linear, time-decay, and position-based as selectable options — they're no longer supported, with accounts automatically upgraded to data-driven attribution as the default. That's a meaningful signal about where the industry has landed: fixed-rule models are a reasonable starting point, but a data-driven approach — where the weighting comes from your actual conversion patterns instead of an arbitrary formula — tends to outperform them once there's enough data to support it.
If you're just starting to look past last-click reporting:
If your sales cycle is very short and mostly single-session:
If you have meaningful, ongoing conversion volume:
If your real conversion happens outside your website entirely (B2B):
Every model on this list is trying to summarize the same underlying thing: the actual sequence of touchpoints a customer went through. If you can already see that full sequence — every channel, every campaign, every gap in between — the exact model you apply on top of it becomes a much lower-stakes decision, because you're never blind to what's actually happening. That visibility, more than any specific model, is what separates useful attribution from a report nobody trusts. If you haven't already, start with what marketing attribution actually is and work outward from there.
See the complete journey behind every conversion — every touchpoint, every campaign, every channel that actually contributed.
Conversion tracking and attribution consultant. Spends most days inside Google Tag Manager, GA4, and server-side tracking setups, helping ecommerce and B2B teams figure out where their customers actually came from. More about Abdul →