Quick answer
Position-based attribution is a multi-touch model that gives 40% of a conversion's credit to the first touch, 40% to the last touch and splits 20% among the touches in between. It is also called U-shaped attribution. With one touch it gives 100%; with two, 50% each.
Formula: First = 40%, last = 40%, each middle touch = 20% ÷ (n − 2) for n ≥ 3; n = 2: 50% / 50%; n = 1: 100%
What is position-based attribution?
In marketing analytics, position-based attribution is a multi-touch attribution model that gives the first and last touches on a conversion path 40% of the credit each and splits the remaining 20% equally among the touches in between. Because a chart of the weights looks like a U, it is also called U-shaped attribution or 40-20-40 attribution.
The model is a pricing decision on an assumption: that the touch which introduced a person and the touch which closed the conversion matter more than everything in the middle. That assumption is reasonable for long, considered journeys such as B2B leads or expensive purchases, where discovery and decision are separate moments. It is pointless for short ones. On a path with a single touch, position-based gives that touch 100%, exactly like first-touch and last-touch; on a path with two touches it gives 50% each, exactly like linear. The model only says something new when paths have three or more touches.
How the credit is split
The 40-20-40 split needs rules for short paths. TapCub uses these, which are also the common convention:
| Touches on the path | First touch | Each middle touch | Last touch |
|---|---|---|---|
| 1 | 100% | – | (same touch) |
| 2 | 50% | – | 50% |
| 3 | 40% | 20% | 40% |
| 4 | 40% | 10% | 40% |
| 5 | 40% | 6.7% | 40% |
| n ≥ 3 | 40% | 20% ÷ (n − 2) | 40% |
Two consequences follow from the table. First, the middle touches share a fixed 20% no matter how many there are, so each one is worth less on longer paths: at 6 touches each middle touch gets 5%. Second, credit is assigned per touch, then summed per channel, so a channel that is both first and last on a path collects 80% from that path.
A "touch" here is whatever unit your tool records. In TapCub it is a session inside the lookback window, labelled by channel (or another touch dimension you choose).
U-shaped vs W-shaped attribution
| U-shaped (position-based) | W-shaped | |
|---|---|---|
| Heavily weighted points | First touch and last touch | First touch, lead creation and opportunity creation |
| Typical weights | 40 / 20 / 40 | A common convention is 30 / 30 / 30, with 10% shared by the rest |
| What it needs | Sessions or clicks before a conversion | A CRM that records when a lead and an opportunity were created |
| Good for | Lead-gen and long e-commerce journeys with one conversion event | B2B funnels with defined lead and pipeline stages |
| Blind spot | Undervalues nurturing touches in the middle | Weights depend on CRM stage definitions, which differ by company |
Weights for W-shaped and other custom models are not standardised; vendors and teams choose their own. If you need the middle milestone in a web-only setup, a practical substitute is to run position-based twice: once with sign-up as the goal and once with purchase as the goal, keeping only touches after sign-up in the second run.
Worked example: 4 paths, 10 conversions
Sample data: one month of conversions grouped into four paths.
| Path | Touches in order | Conversions | Position-based credit per conversion |
|---|---|---|---|
| A | Organic search → Email → Paid search | 4 | 40% / 20% / 40% |
| B | Paid social → Organic search → Email → Direct | 3 | 40% / 10% / 10% / 40% |
| C | Referral → Paid search | 2 | 50% / 50% |
| D | 1 | 100% |
Multiply each share by the path's conversions and add up by channel:
- Paid search: 4 × 0.4 (A, last) + 2 × 0.5 (C, last) = 1.6 + 1.0 = 2.6
- Email: 4 × 0.2 (A, middle) + 3 × 0.1 (B, middle) + 1 × 1.0 (D) = 0.8 + 0.3 + 1.0 = 2.1
- Organic search: 4 × 0.4 (A, first) + 3 × 0.1 (B, middle) = 1.6 + 0.3 = 1.9
- Paid social: 3 × 0.4 (B, first) = 1.2
- Direct: 3 × 0.4 (B, last) = 1.2
- Referral: 2 × 0.5 (C, first) = 1.0
The six totals add up to 10.0, one per conversion. Now the same paths under three other models:
| Channel | Position-based | Linear | First touch | Last touch |
|---|---|---|---|---|
| Paid search | 2.6 | 2.33 | 0 | 6 |
| 2.1 | 3.08 | 1 | 1 | |
| Organic search | 1.9 | 2.08 | 4 | 0 |
| Paid social | 1.2 | 0.75 | 3 | 0 |
| Direct | 1.2 | 0.75 | 0 | 3 |
| Referral | 1.0 | 1.0 | 2 | 0 |
Linear gives each touch on a path an equal share (path A: one third each; path B: one quarter each). Reading the columns together:
- Email drops from 3.08 under linear to 2.1 under position-based, because on long paths it sits in the middle. Position-based is telling you it nurtures rather than opens or closes.
- Paid social and Direct gain (0.75 → 1.2) because they are the edges of path B.
- Paths C and D (3 of the 10 conversions) get exactly the same credit under position-based and linear, and path D gets the same under every model. For those conversions the model choice changes nothing.
That last point is the test for whether position-based is worth reading: if most of your conversions have one or two touches, it adds no information beyond first, last and linear.
When to use position-based attribution
- If the average number of touches per attributed conversion is three or more, and the conversion is a considered decision (a demo request, a quote, a high-value order), then position-based is a sensible default for budget discussions: it rewards channels that open journeys and channels that close them.
- If most conversions have one or two touches, then read first and last touch instead; position-based will mirror them or linear.
- If your question is which channel caused the conversion, then no rules-based model answers it; compare models for patterns and test causation with an incrementality experiment or a media mix model.
Common mistakes
- Treating 40-20-40 as measured. The weights are an assumption you chose, not something the data estimated.
- Reading one model alone. A channel's position-based number means more next to its first-touch, last-touch and linear numbers; the gaps are the insight.
- Ignoring the lookback window. A 7-day and a 90-day window produce different first touches, and position-based gives the first touch 40%.
- Counting repeated sessions as new channels. Three direct sessions in a row are three touches for the model, so a channel people return through often can take both the edges.
- Assuming an ad platform's numbers will match. Ad platforms count their own clicks and impressions; on-site attribution counts sessions. Google Ads and Google Analytics no longer offer position-based as a selectable model, so for this comparison you need a tool that still calculates it.
Position-based attribution in TapCub
The four-path example above is the kind of question TapCub's attribution view is built for: one goal, several models, one column each.
- Set the goal and touch. Open All-platform analytics › Behavior › Behavior analysis, switch to attribution and choose the goal (one event or page, such as
purchaseor a thank-you page). The touch defaults to channel, falling back to referrer channel when empty; you can also choose referrer channel, UTM source, UTM campaign, referrer domain or landing page. - Pick the window. The lookback window defaults to 30 days (1–90) and includes the session of the conversion itself. Use a longer window for lead-gen journeys so the real first touch is inside it.
- Tick position together with the others. Select first, last, linear, position and time decay at once. TapCub's position model follows the table above: one touch gets everything, two split in half, and with more the first and last get 40% each while the middle shares 20%. Each model becomes a column of credited conversions per channel.
- Check whether position-based matters for you. Read the average touches number at the top. If it is close to 1 or 2, the position column will repeat first, last or linear, as paths C and D did. If it is 3 or more, compare position with linear to find middle-of-journey channels like Email in the example.
- Narrow to a later stage. The optional earlier event keeps only touches after, say,
signup, which gives you a W-shaped-style second view without a CRM. How to compare attribution models walks through each setting, and multi-touch attribution compares all the models on one dataset.
Sources
- Google Ads Help, About attribution models.
- Google Analytics Help, Get started with attribution.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Product Analyst
Focuses on event analysis, conversion funnels, and retention, using clear metric definitions and structured analysis to explain product performance.