How position-based attribution works
Position-based attribution is a multi-touch attribution model built on a simple belief: two moments in a customer's journey matter most. The first touch, when they discovered you, and the last touch, when they decided to buy. Everything in between helped, but less.
So the model combines first-touch and last-touch attribution. It gives each end 40% of the sale and lets the middle touches share the other 20%. Plot the credit per touch and you get a U, hence the other name.
The credit rule
first touch = 40% · last touch = 40% · each middle touch = 20% ÷ (n − 2)
Here n is the number of touchpoints in the attribution window. The edge cases need their own rules:
- One touch: it gets 100%, as in every model.
- Two touches: there is no middle. Splitting 50/50 between first and last is the natural choice; write down which rule you use.
- Three or more: 40/40 to the ends and 20% split across the middle, so each middle touch gets less the longer the journey.
Worked example
A customer pays $1,000 after four visits:
| Touch | Channel | Position | Share | Credit |
|---|---|---|---|---|
| 1 | Reddit post | First | 40% | $400 |
| 2 | Organic search | Middle | 10% | $100 |
| 3 | Newsletter | Middle | 10% | $100 |
| 4 | Google ad | Last | 40% | $400 |
Compare the same path under other models:
| Channel | Position-based | Linear | First touch | Last touch |
|---|---|---|---|---|
| Reddit post | $400 | $250 | $1,000 | $0 |
| Organic search | $100 | $250 | $0 | $0 |
| Newsletter | $100 | $250 | $0 | $0 |
| Google ad | $400 | $250 | $0 | $1,000 |
With eight touches instead of four, the two ends still get $400 each, and the six middle touches get about $33 each. Long journeys push the middle toward zero.
Pros and cons
| Pros | Cons |
|---|---|
| Credits both discovery and closing | The 40/20/40 split is a convention, not something measured from your data |
| Still gives the middle something, unlike single-touch models | Middle touches shrink as journeys get longer, however useful they were |
| Shares add up to real revenue, with no double counting | The last touch is often a direct return, so Direct takes 40% unless you exclude it |
| Easy to explain once people see the U | Needs complete, linked visit histories to know which touch was first |
When to use position-based attribution
- You care about acquisition and conversion equally, and don't want to pick one single-touch model.
- Journeys usually have three to five touches. Very long journeys starve the middle; very short ones make it identical to linear or single touch.
- You run both prospecting and retargeting campaigns and want each to get credit for its job.
If you think the touches near the sale matter most, time-decay attribution fits better. If you can't tell what matters, linear attribution is the neutral choice.
Where position-based attribution is available
Google Ads and Google Analytics used to offer a position-based model. Both have removed it, along with first click, linear and time decay, and now offer data-driven and last-click attribution. Conversion actions that used it were switched to data-driven. To use it today, you need full visit paths from your own analytics, or a spreadsheet.
How to compute it yourself
- Export each buyer's visits in order, with source, and the payment they led to.
- Keep visits inside your attribution window. Decide whether a direct visit can be the last touch.
- Give 40% of the payment to the first visit and 40% to the last. Split 20% across the rest, or 50/50 when there are only two.
- Sum by channel and campaign; the total must equal attributed revenue.
Common mistakes
- Losing the first touch. If visits on different days aren't linked to the buyer, the first touch you see is often a later visit, and 40% goes to the wrong channel.
- Letting Direct be the last touch. A bookmark visit before checkout gets 40%. Skip direct touches when another source exists.
- Changing the split to make a channel look better. Pick a rule and keep it.
- Reading the middle as unimportant. The middle gets 20% because of the model, not because the data showed those touches don't matter.
Does MetricRaven support position-based attribution?
No. MetricRaven credits each payment to the customer's first tracked visit: the 100% first-touch end of the U. It has no position-based, linear or time-decay setting, so a channel's revenue means the same in every report.
The other end of the U, and the middle, are visible as side views that don't change revenue totals:
- Exit page revenue credits the last page buyers viewed before paying.
- Assisted revenue in the Meta Ads and Google Ads reports counts payments from customers who clicked a campaign after first arriving elsewhere.
- Paths to payment lists the pages buyers saw across all their visits, from entry to first payment.
Linking visits to the buyer is what makes any of this work: pass the visitor ID through Stripe Checkout or identify users at signup.
Questions
- What is the 40-20-40 attribution model?
- Another name for position-based attribution: 40% of the credit to the first touch, 40% to the last touch, and 20% shared equally by the touches in between.
- What is the difference between U-shaped and W-shaped attribution?
- U-shaped gives most credit to the first and last touch. W-shaped adds a third heavy point in the middle, usually the touch where the visitor became a lead or signed up, which suits B2B funnels with a clear lead stage.
- How does position-based attribution handle two touchpoints?
- There is no middle, so the 20% has nowhere to go. Splitting the sale 50/50 between the first and last touch is the usual approach.
- Is position-based attribution available in Google Analytics 4?
- No. GA4 removed position-based, first click, linear and time decay models. It offers data-driven attribution and two last-click models.