What is an Attribution Model?
An attribution model is a rule set or algorithm that decides how to distribute credit for a conversion across the touchpoints of a user’s journey—e.g., first ad click, organic visit, email reopen, or direct return. In other words, it answers: Which interactions should get how much credit for the outcome? Platforms such as GA4, Matomo, Plausible, and Simple Analytics all support attribution, but they differ in which models they offer and how they calculate them.
How does an Attribution Model work?
Attribution modeling aggregates users’ paths (sequences of sessions and hits) and assigns fractional credit per touchpoint. Credit can be assigned to channels (e.g., Paid, Organic), to source/medium or campaign, or even to specific creatives via UTM tags. The calculation typically respects a lookback window (how far back from the conversion you’re willing to give credit) and scope (per user, per session, or cross-session).
Common models (rule-based & algorithmic)
- First Touch: 100% credit to the first recorded interaction. Useful for discovery KPIs; biased toward upper-funnel.
- Last Touch: 100% credit to the final interaction. Useful for short cycles; biased toward bottom-funnel and brand/direct.
- Linear: Equal credit to all touchpoints. Simple and fair-looking, but may over-credit long paths.
- Position-based (e.g., U-shape): More weight to first and last; remaining credit divided among middle touches. Good for both discovery and closing.
- Time-decay: More credit the closer the touchpoint is to the conversion. Reflects recency influence; can underweight early discovery.
- Data-driven/algorithmic: Uses statistical learning on your data (e.g., probabilistic removal effects) to estimate marginal contribution. Best when you have sufficient volume and stable tracking.
Minimal example
If a user journey is:
Ad → Email → Organic → Direct → Purchase
- Last Touch: Direct gets 100%.
- Linear: each of the four gets 25%.
- Position-based (40/20/40): Ad 40%, Direct 40%, Email+Organic split 20%.
Implementation notes for analysts
- Ensure consistent tagging and channel grouping; misclassified referrals or missing referrers will skew any model.
- Align the model to business questions: discovery growth vs. efficiency near conversion.
- Compare models, don’t pick once and forget. Re-evaluate when journey length, mix, or privacy settings change.
- Alternative tools: lightweight suites (Plausible, Simple Analytics) focus on clarity and basic touch attribution; Matomo and GA4 can support broader model comparisons when configured well.