Why the model choice matters
Every attribution model takes the same input (the touchpoint log) and produces a different revenue split. A campaign that looks profitable under last-click can look unprofitable under data-driven. The model isn’t a technical detail, it’s a strategic choice that decides which channels get budget.
Common attribution models
| Model | How it works | When to use |
|---|---|---|
| Last-click | 100% credit to the last touchpoint before conversion | Heavy-bottom-funnel businesses. Quick directional reads |
| First-click | 100% credit to the first touchpoint | Top-of-funnel-driven brands. Awareness measurement |
| Linear | Equal credit across all touchpoints | Simple, balanced default. Weak when journey is long |
| Time-decay | Recent touchpoints weighted more heavily | Most short-cycle e-commerce |
| U-shaped (position-based) | 40% first, 40% last, 20% middle | Brands where discovery and conversion are the two big moments |
| Data-driven | Algorithm learns weights from conversion data (typically Shapley values) | High-volume businesses with clean tracking |
| Custom rules | Hand-tuned splits with business logic (e.g. always discount the brand-search channel) | Mature teams with strong opinions about specific channels |
In-platform vs independent
A critical distinction: every ad platform applies its OWN attribution model and reports the result. Meta’s “7-day click + 1-day view” model attributes generously to Meta touchpoints. Google’s “data-driven” model attributes generously to Google touchpoints. The two never reconcile.
An independent attribution layer, one that ingests data from all sources and applies one consistent model, is the only way to get a comparable cross-channel view. Without it, “ROAS” doesn’t mean the same thing from one report to the next.
How to choose
Three questions to ask:
- What does your sales motion actually look like? Long, awareness-heavy journeys favour first-click or U-shaped. Direct-response funnels favour last-click or time-decay.
- How much data do you have? Data-driven models need volume, typically 10K+ conversions per month per modelled segment.
- What decisions does the number drive? A budget-allocation model can be different from a campaign-evaluation model. There is no rule against running both.
Common mistakes
- “Switching models” without versioning. Comparing yesterday’s last-click ROAS to today’s data-driven ROAS is comparing apples to oranges. Track model version per report.
- Blaming the model for tracking gaps. If your touchpoints are missing, no model will save you.
- Treating the model as universal. What works for prospecting campaigns may not work for retention.
FAQ about Attribution Model
What is the most common attribution model?
Last-click is the most common and the most biased, it gives 100% of credit to the final touchpoint, usually branded search or direct. Multi-touch models distribute credit more fairly.
Which attribution model should I use?
The right model matches how your sales motion works. Long awareness-driven journeys favour first-click or U-shaped. Direct-response funnels favour last-click or time-decay. Mature teams compare two or three in parallel.
What is data-driven attribution?
Data-driven attribution learns credit weights from your conversion data, typically via Shapley values. It requires sufficient volume (~10K+ conversions per modelled segment per month) to converge.