Guide
Marketing attribution models, explained
How the six common models share credit for a sale, what each one is good at, and why comparing them beats picking one.
Short answer: A marketing attribution model is a rule for sharing the credit for a sale across the ads, posts, searches and emails that led to it. The six common models are first touch, last click, linear, time decay, U-shaped and W-shaped. None is objectively correct; each answers a different question, so the most useful approach is to compare several on the same data.
Updated 25 September 2026 · 7 min read
What is marketing attribution?
Marketing attribution is the practice of working out which marketing activity deserves credit for a conversion. A conversion is the action you are measuring, such as a purchase, a signed deal or a placed candidate. A touch (or touchpoint) is one interaction on the way there: clicking an ad, arriving from a search result, opening a newsletter link.
Most customers touch several channels before they buy. Someone might click a paid search ad, come back a week later from a blog post, then return from a social post and fill in your enquiry form. The attribution model decides how much of that eventual sale each of those three visits earns.
What are the six attribution models?
| Model | How it shares credit | Good for |
|---|---|---|
| First touch | 100% to the first recorded touch | Judging which channels introduce new customers |
| Last click | 100% to the last touch before the conversion | Seeing which channels close sales |
| Linear | Equal shares to every touch | A neutral view of everything that played a part |
| Time decay | More credit to touches closer to the conversion | Short buying cycles and promotions |
| U-shaped | 40% first touch, 40% last touch, 20% shared across the middle | Valuing both discovery and closing |
| W-shaped | 30% first touch, 30% the touch that created the lead, 30% last touch, 10% shared across the rest | Longer B2B cycles where becoming a lead is a milestone |
How does first-touch attribution work?
First-touch attribution gives all of the credit to the interaction that first brought the customer to you. It is the model to use when you want to know where new demand comes from. Its blind spot is everything that happened afterwards: the nurturing, the retargeting and the final nudge get nothing.
How does last-click attribution work?
Last-click attribution gives all of the credit to the final touch before the conversion. It is simple and widely used, which is why many ad platforms and analytics tools have leaned on it for years. It tends to over-reward channels that sit at the end of the journey, such as brand search and retargeting, because people who have already decided often arrive through them.
How do linear and time-decay attribution differ?
Linear attribution splits the credit equally, so a four-touch journey gives each touch 25%. It is fair in a blunt way, but it treats a passing impression the same as a detailed product demo.
Time-decay attribution gives more credit to touches that happened closer to the conversion. In uForge the weighting halves for every seven days a touch sits before the conversion, so a visit the day before a purchase counts for far more than one a month earlier. It suits short, promotion-driven buying cycles and undervalues slow-burn awareness work.
How does U-shaped attribution split credit?
U-shaped (or position-based) attribution gives 40% of the credit to the first touch, 40% to the last touch, and shares the remaining 20% equally across any touches in between. The conversion itself, such as the form fill or the purchase, is not counted as a touch; it is the thing being credited.
Take the journey above: paid search ad, then blog article, then social post, then the enquiry form. Under U-shaped attribution the paid search ad earns 40%, the social post earns 40% and the blog article earns 20%. With only two touches, each gets half.
How does W-shaped attribution split credit?
W-shaped attribution adds a third milestone: the touch at which the visitor became a lead, usually by filling in a form. The first touch, the lead-creation touch and the last touch each receive 30%, and the remaining 10% is shared across everything else. If no separate lead-creation touch can be identified, or it coincides with the first or last touch, uForge falls back to the U-shaped split for that conversion.
W-shaped suits businesses where getting the lead is its own achievement, such as B2B services or recruitment, and where marketing and sales want to see credit for both.
Which attribution model should you use?
Start from the decision you need to make rather than the model:
- Where should we spend to find new customers? Look at first touch and U-shaped.
- Which channels close? Look at last click and time decay.
- Is a mid-funnel channel pulling its weight? Look at linear and W-shaped, which give the middle of the journey a share.
The most useful habit is to compare. A channel that looks strong under every model is a safe bet. A channel that only looks good under last click is probably harvesting demand created elsewhere. A channel that only shows up under first touch may be doing valuable introductory work that a last-click report hides.
What limits every attribution model?
All six models share the same data, so they share its gaps. They can only credit touches that were recorded, which means visitors who declined analytics consent cannot have their visits joined into a journey. Offline conversations, word of mouth and a colleague's recommendation leave no trace at all. Treat attribution as a well-reasoned estimate, and be wary of any report that presents one model's output as the plain truth. Our guide to why analytics undercounts after cookie consent covers the consent side in more detail.
Where uForge fits
uForge's first-party tracker, Sadang, calculates all six models for every sale order, won deal and recruitment job it attributes, so you can switch model in a report and watch the picture change. Attribution runs once a day, so reports cover activity up to yesterday, and figures are split by the consent the visitor gave. See analytics and attribution in uForge, or look up any unfamiliar term in the glossary.