First touch
Assumes: The visit that introduced the buyer did the work.
Where it goes wrong: Everything after it, including the visit right before the conversion, gets no credit.
Free tool
Attribution model calculator: one journey, seven models.
Enter one deal and the visits that came before it. See how first touch, last touch, linear, time decay, U-shaped, W-shaped and AI platforms attribution would split it, using the same rules SourceTrack applies to your own data. No signup; the calculation runs in your browser.
This is a journey simulator, not live attribution: it splits the one journey you enter. It gives no verdict on which model is right and no figures from anyone else's data.
Seven models simulated for $1,200.00 across 4 visits. First touch: Paid Social. Last touch: Email.
How each model splits the deal
- Paid Social
- Organic Search
- AI Search
First touchLargest: Paid Social
Last touchLargest: Email
LinearLargest: Paid Social, Organic Search, AI Search, Email
Time decayLargest: Email
U-shapedLargest: Paid Social, Email
W-shapedLargest: Paid Social, Organic Search, Email
AI platformsLargest: AI Search
Credit by channel under each model
| Channel | First touch | Last touch | Linear | Time decay | U-shaped | W-shaped | AI platforms |
|---|---|---|---|---|---|---|---|
| 1. Paid Social | $1,200.00 | $0.00 | $300.00 | $71.34 | $480.00 | $360.00 | $0.00 |
| 2. Organic Search | $0.00 | $0.00 | $300.00 | $173.92 | $120.00 | $360.00 | $0.00 |
| 3. AI Search | $0.00 | $0.00 | $300.00 | $384.05 | $120.00 | $120.00 | $1,200.00 |
| 4. Email | $0.00 | $1,200.00 | $300.00 | $570.69 | $480.00 | $360.00 | $0.00 |
| Total credited | $1,200.00 | $1,200.00 | $1,200.00 | $1,200.00 | $1,200.00 | $1,200.00 | $1,200.00 |
| Largest share | Paid Social | Paid Social, Organic Search, AI Search, Email | Paid Social, Email | Paid Social, Organic Search, Email | AI Search |
Show the math
Visits are listed oldest first. Credit is split to the cent. Each share is cut down to whole cents, and the cents left over go one each to the visits with the largest cut-off remainder (on a tie, the later visit), so the credits add up to the deal value.
First touch
The earliest visit inside the attribution window gets the whole deal.
- Visit 1 (Paid Social, 21 days before) is the earliest visit in the window, so it gets the whole $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $1,200.00 | 100.00% |
| 2 | Organic Search | 12 days | $0.00 | 0.00% |
| 3 | AI Search | 4 days | $0.00 | 0.00% |
| 4 | 0 days | $0.00 | 0.00% |
Last touch
The latest visit inside the attribution window gets the whole deal.
- Visit 4 (Email, 0 days before) is the latest visit in the window, so it gets the whole $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $0.00 | 0.00% |
| 2 | Organic Search | 12 days | $0.00 | 0.00% |
| 3 | AI Search | 4 days | $0.00 | 0.00% |
| 4 | 0 days | $1,200.00 | 100.00% |
Linear
Every visit gets an equal share.
- $1,200.00 ÷ 4 visits = $300.00 each.
- Every share is a whole number of cents. The credits add up to $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $300.00 | 25.00% |
| 2 | Organic Search | 12 days | $300.00 | 25.00% |
| 3 | AI Search | 4 days | $300.00 | 25.00% |
| 4 | 0 days | $300.00 | 25.00% |
Time decay
Each visit is weighted 0.5 ^ (days before the conversion ÷ 7), a 7-day half-life: a visit 7 days out counts half as much as one at the moment of conversion, and 14 days out a quarter. A visit’s share is its weight divided by the total weight.
- Total weight of the 4 visits: 2.1027.
- Visit 1 (Paid Social, 21 days before): weight 0.5 ^ (21 ÷ 7) = 0.1250, share 0.1250 ÷ 2.1027 = 5.95%.
- Visit 2 (Organic Search, 12 days before): weight 0.5 ^ (12 ÷ 7) = 0.3048, share 0.3048 ÷ 2.1027 = 14.49%.
- Visit 3 (AI Search, 4 days before): weight 0.5 ^ (4 ÷ 7) = 0.6730, share 0.6730 ÷ 2.1027 = 32.00%.
- Visit 4 (Email, 0 days before): weight 0.5 ^ (0 ÷ 7) = 1.0000, share 1.0000 ÷ 2.1027 = 47.56%.
- Rounded to the cent: each share is cut down to whole cents and the 2 cents left over go one each to the visits with the largest cut-off remainder (on a tie, the later visit). The credits add up to $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $71.34 | 5.95% |
| 2 | Organic Search | 12 days | $173.92 | 14.49% |
| 3 | AI Search | 4 days | $384.05 | 32.00% |
| 4 | 0 days | $570.69 | 47.56% |
U-shaped
40% to the first visit, 40% to the last, and the other 20% split equally across the visits in between. Two visits split 50/50; one visit takes the whole deal.
- 40% of $1,200.00 = $480.00 each to the first and the last visit: visit 1 (Paid Social, 21 days before) and visit 4 (Email, 0 days before).
- The other 20% ($240.00) is split across the 2 middle visits: $120.00 each.
- Every share is a whole number of cents. The credits add up to $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $480.00 | 40.00% |
| 2 | Organic Search | 12 days | $120.00 | 10.00% |
| 3 | AI Search | 4 days | $120.00 | 10.00% |
| 4 | 0 days | $480.00 | 40.00% |
W-shaped
30% each to the first visit, the middle visit and the last visit, and the other 10% split equally across the rest. The middle visit is picked by position (visit ⌊(n − 1) ÷ 2⌋ + 1 of n), not by what happened on it. Three visits split in equal thirds, two split 50/50, one takes the whole deal.
- 30% of $1,200.00 = $360.00 each to the first, the middle and the last visit: visit 1 (Paid Social, 21 days before), visit 2 (Organic Search, 12 days before) and visit 4 (Email, 0 days before).
- The other 10% ($120.00) goes to the one other visit.
- Every share is a whole number of cents. The credits add up to $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $360.00 | 30.00% |
| 2 | Organic Search | 12 days | $360.00 | 30.00% |
| 3 | AI Search | 4 days | $120.00 | 10.00% |
| 4 | 0 days | $360.00 | 30.00% |
AI platforms
The most recent AI Search visit inside the window gets the whole deal. A paid AI ad click (Paid AI) does not count. With no AI Search visit, this model credits no one for the deal.
- Visit 3 (AI Search, 4 days before) is the most recent AI Search visit, so it gets the whole $1,200.00.
| Visit | Channel | Before conversion | Credit | Share |
|---|---|---|---|---|
| 1 | Paid Social | 21 days | $0.00 | 0.00% |
| 2 | Organic Search | 12 days | $0.00 | 0.00% |
| 3 | AI Search | 4 days | $1,200.00 | 100.00% |
| 4 | 0 days | $0.00 | 0.00% |
Not simulated here: needs your data
- First touch (non-direct). It skips every visit that has no utm_source tag, and a channel name does not say whether a visit was tagged: an untagged visit from a Google search is Organic Search, yet this model treats it as direct.
- Last touch (non-direct). It skips every visit that has no utm_source tag, and a channel name does not say whether a visit was tagged: an untagged visit from a Google search is Organic Search, yet this model treats it as direct.
- Linear (Paid Only). It decides whether a visit is paid from that visit’s own UTM medium, source and ad click IDs, and a channel name does not settle that: a visit tagged only utm_medium=sem is Paid Search, yet this model does not count it as paid.
How it works
Same deal, same visits, different rules.
Visits and revenue stay fixed. Only the rule for sharing the credit changes.
- Step 1
Enter one deal
The revenue from a single conversion, and the attribution window your reports use.
- Step 2
Add the visits before it
Up to ten, each with its channel and how many days before the conversion it happened.
- Step 3
Compare the splits
Seven models split the same deal side by side, to the cent, with the arithmetic shown for each.
The models
What each model assumes, and where it goes wrong.
Every model is a rule, not a measurement. Each one leaves something out by design.
Last touch
Assumes: The final visit did the work.
Where it goes wrong: Whatever introduced the buyer gets no credit, so channels that start journeys get nothing under it.
Linear
Assumes: Every visit mattered equally.
Where it goes wrong: A quick return visit counts as much as the one that introduced the buyer, and a long journey spreads the deal thin.
Time decay
Assumes: Recent visits mattered more.
Where it goes wrong: With a 7-day half-life, a visit 30 days before the conversion carries about 5% of the weight of a same-day visit, so early visits fade fast.
U-shaped
Assumes: The first and the last visit matter most.
Where it goes wrong: The visits in between share 20% however many there are, so each one gets less as the journey grows.
W-shaped
Assumes: One middle visit matters as much as the first and the last.
Where it goes wrong: SourceTrack picks that visit by its position, so it marks no particular event, and the rest share 10%.
AI platforms
Assumes: The question is whether an AI assistant sent this buyer.
Where it goes wrong: It credits no one when the window holds no AI Search visit, so its total covers only AI-referred deals, not your revenue.
Not simulated here: First touch (non-direct), Last touch (non-direct), Linear (Paid Only). They need facts about each visit that a channel name does not carry, which the calculator section above explains. SourceTrack runs them on your own tracked visits. Read more about all ten attribution models.
Limits
What this does not tell you.
- It splits one journey you entered. SourceTrack applies the same rules to every conversion and visit it records on your site, so your reports depend on that data, not on this example.
- Each row here is one visit. SourceTrack starts a new visit after 30 minutes without activity, so several pages viewed in one sitting count once.
- Rows with the same number of days keep the order you listed them. In your data, visits are ordered by the time they were recorded.
- In your reports, AI platforms names the assistant (ChatGPT, Perplexity and others) and also checks the conversion event itself for an AI referral. Here it is the AI Search channel.
- It does not say which model suits your business. Each model answers a different question about the same journey.
FAQ
Questions about the models.
What does this attribution model calculator do?
It takes one customer journey you enter (a deal value, and each visit’s channel and how many days before the conversion it happened) and splits the deal under seven of SourceTrack’s ten attribution models, to the cent, with the rule each model follows. The calculation runs in your browser.
Which visits count toward a model?
Only visits inside the attribution window: those that happened no more days before the conversion than the window allows, the boundary day included. In SourceTrack you set the window per site, from 1 to 365 days. Visits outside it get no credit under any model.
How does the time decay model weight each visit?
Each visit is weighted 0.5 raised to the power of its days before the conversion divided by 7, a 7-day half-life. A visit 7 days out weighs half as much as one at the moment of conversion, and one 14 days out a quarter. A visit’s share is its weight divided by the total weight of the journey.
How are U-shaped and W-shaped attribution different?
U-shaped gives 40% to the first visit, 40% to the last and splits the other 20% across the visits in between. W-shaped gives 30% each to the first, the middle and the last visit and splits the other 10% across the rest. SourceTrack picks the W-shaped middle visit by its position in the journey, not by a CRM stage.
Why do the credits add up to exactly the deal value?
Each share is cut down to whole cents, and the cents left over go one each to the visits with the largest cut-off remainder, the later visit first on a tie. SourceTrack splits a conversion’s value across visits with the same rule, in the smallest unit of the conversion’s currency.
Why are three of the ten models not simulated?
First touch (non-direct) and last touch (non-direct) skip visits without a utm_source tag, and Linear (Paid Only) decides whether a visit is paid from its own UTM tags and ad click IDs. A channel name does not carry either fact, so this page cannot reproduce those models exactly and does not guess.
Related tools
More free tools.
ROAS calculator
Return on ad spend, profit on ad spend and the break-even ROAS your margin implies.
Open the roas calculatorCAC calculator
Blended and paid CAC, LTV, the LTV:CAC ratio and the months it takes to repay.
Open the cac calculatorTracking audit
Scan a site for its tags, consent tool and first-party setup in one request.
Open the tracking auditThis splits one journey. Attribution splits all of them.
SourceTrack applies these rules to the visits and conversions it records on your site, so you can switch models on your own revenue. One script tag, 28 days free.
No credit card required. The Free plan includes first-touch and last-touch attribution; every paid plan includes all ten models.
Which source earned it?Example
- Google Ads
- ChatGPT
- Google Search
- Direct
Sources ranked by the revenue they brought in.