The linear attribution model splits credit for a conversion evenly across every touch a customer had before it. Four tracked interactions before a closed deal means each one carries 25%. That is the whole rule, and nearly every argument about linear attribution comes from how you define a touch, not from the division itself.
Teams reach for it after noticing that last-touch reports keep crediting the same closing channel while the work that created demand stays invisible. Linear is the simplest correction: instead of handing the win to the final click, it spreads credit across the path. This guide covers the split, a worked example, and when first or last touch is the better call.
How linear attribution splits credit
With n touches in the path, each touch receives 1/n of the conversion value. Nothing is weighted by recency, position or channel cost.
The definition of a touch is where implementations drift. A touch is one trackable interaction tied to a person or an account inside an attribution window, and teams choose differently: session start, first visit per channel per day, or a named action such as a pricing view or a demo request. Each choice changes n, and changing n changes every number in the report. Write the definition down before you compute anything and keep it stable, or month-over-month comparisons stop meaning anything.
A worked example: four touches, one closed deal
A SaaS company closes a $12,000 annual contract. The buyer’s tracked path:

| # | Touch | Channel | Linear | First touch | Last touch |
|---|---|---|---|---|---|
| 1 | Click on a paid search ad | Paid search | $3,000 | $12,000 | $0 |
| 2 | Read a blog post on workflows | Organic | $3,000 | $0 | $0 |
| 3 | Pricing page after a comparison read | Direct | $3,000 | $0 | $0 |
| 4 | Demo request form | Organic | $3,000 | $0 | $12,000 |
Three models, three stories, one deal. Last touch says the demo form produced the revenue; first touch says the ad did; linear says the path did. The demo request would not exist without the pricing page, and the pricing page would not exist without the article. If a five-person buying committee touched this deal, split per contact and then sum. Mixing person-level and account-level touches in one calculation overweights whoever clicked most.
When linear beats first touch or last touch
First touch answers a different question
First touch tells you which channel introduces new names. That is worth knowing, but it is a poor basis for funding the middle of the funnel, because it treats an early ad click as the cause of revenue that took four more steps to arrive. If you are deciding whether to invest in comparison content or a demo-request page, linear shows those stages participating in closed deals.
Last touch collapses the path
Last touch is reasonable for short, single-session purchases and for demand capture, where the final click really does reflect intent. It breaks on long cycles. If most of your deals end with a demo request or a checkout, last touch reports that nearly all revenue came from demos or checkout: true, and useless, because it tells you where deals finish, not what built them.
Linear is not automatically right either. When n is one or two it collapses toward first or last touch and adds nothing. When a commission or a spend decision hinges on one specific action, attributing to that action is honest about what you are optimizing.
Make the numbers traceable
- Store every touch as an event with a stable visitor ID and, once known, a user and account ID.
- Fix the attribution window and the dedupe rule before analysis, not after the numbers look wrong.
- Exclude bots and internal traffic. A crawler that walks ten pages adds ten touches that never had a chance to convert and dilutes every share; our bot detection handles that at the source.
- Compute the split at conversion time and persist it, so the report does not recompute differently next week.
- Report linear alongside first and last. One model is one opinion; three show the range.
- Check whether a difference between converters and non-converters is meaningful before you move budget.

Prynt Revenue AI reports first-touch, last-touch and linear attribution from your own first-party events, with converters and non-converters compared under significance tests. It sits in the suite described on our platform page, and definitions are in the glossary.
FAQ

Is linear the same as even-credit attribution?
Yes. Linear weights every touch equally. U-shaped attribution is the one that weights first and last touches more heavily.
Does linear work with a long B2B sales cycle?
It works when you can tie touches to an account and set a window that covers the cycle. Beyond a quarter, the split starts measuring how many people visited your site rather than what influenced the decision, so keep the window and the touch definition conservative.
Can I run it without third-party cookies?
Yes, on your own domain. First-party events tied to a visitor ID and a known user ID are enough to compute the split; the model needs an ordered path, not a cross-site identity.
Linear attribution is a reporting decision, not a measurement of truth. Pick the model that matches the question, write the touch definition down, and compare before you move budget. You can see first-touch, last-touch and linear attribution together in Prynt Revenue AI, $99/month with a 14-day trial — start at pryntid.com.
Try it free
Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.