Linear attribution is a multi-touch attribution model that splits conversion credit equally across all touchpoints in a user’s journey. If four interactions helped close the deal (ad click → blog post → email → direct), each gets 25% of the conversion value. It’s the no-drama baseline: simple, transparent, and great for comparing channels without arguing which one “won.”
Short formula
If a path has N touchpoints, each touchpoint’s credit = 1 / N of the conversion (or revenue) value.
How it works (mini example)
A user converts for $200 after 4 touches:
| Step | Channel (touchpoint) | Credit (linear) |
|---|---|---|
| 1 | Paid Search | 25% → $50 |
| 2 | Content Blog | 25% → $50 |
| 3 | 25% → $50 | |
| 4 | Direct | 25% → $50 |
This equal-split view helps you quantify assisted conversions instead of over-rewarding a single click like last-click or first-click.
When to use
- You want a fair, channel-agnostic baseline before testing more sophisticated models.
- Journeys are short-to-medium and channels play comparable roles.
- You’re auditing channel mix and normalizing reporting across teams.
Limitations (read before you ship it)
- Role blindness: Linear ignores the function of a touch. An awareness view and a checkout retargeting click get the same weight.
- Path length sensitivity: Long paths dilute credit; micro-touches (e.g., multiple navigations) can add noise.
- No time decay: A click from 30 days ago counts the same as the final nudge. Consider position-based or time-decay models when recency clearly matters.
Implementation notes
- Define your attribution window (/wiki/conversion-window) and de-duplicate touches (session stitching, user IDs).
- Keep consistent channel grouping (/wiki/channel-grouping) and clean UTM parameters (/wiki/utm-parameters) to avoid splitting credit across mislabeled sources.
- Tie linear credit back to cost to get channel-level CPA (/wiki/cost-per-acquisition) or ROAS comparisons.
Why analysts keep it around: Linear attribution is the “control group” of attribution—predictable, explainable, and a solid yardstick before you move to rule-based or data-driven models in broader multi-touch attribution.