Multi-Touch Attribution Models Explained
What multi-touch attribution is
Multi-touch attribution is a method of distributing credit for a conversion across all the touchpoints a buyer interacted with, instead of awarding it entirely to one. Where single-touch attribution names a single hero, the first click or the last, multi-touch acknowledges that a paid ad, a blog post, an email, and a demo request may all have contributed to the same deal.
Why split the credit at all
Most considered purchases, especially in B2B, involve many sessions across several channels and weeks. If you credit only the last touch, you systematically undervalue the awareness and nurture activity that created the demand in the first place, and you over-invest in bottom-of-funnel channels that merely capture intent others built. Multi-touch models exist to spread that credit so each channel's true contribution becomes visible.
Rules-based versus data-driven
The four models on this page are rules-based: a human decides the weighting in advance. A separate category, data-driven attribution, uses your own conversion data to learn weights statistically. Rules-based models are transparent and easy to explain; data-driven models can be more accurate but need volume and are harder to audit.
The four models and how each splits credit
Each rules-based model is just a different formula for dividing 100% of the credit across the touchpoints in a path. Picture a journey with four touches (an organic search, a paid social ad, an email click, and a direct visit that converts) and watch how each model treats them.
Linear
Linear attribution gives every touchpoint an equal share. In a four-touch path, each touch receives 25%. It is the simplest multi-touch model and treats the whole journey as a team effort with no most-valuable player. Use it when you believe every interaction matters roughly equally, or when you want a neutral baseline before introducing weighting.
Time-decay
Time-decay gives more credit to touches that happened closer to the conversion and less to earlier ones, decaying along a half-life you set (a seven-day half-life is a common default). In our four-touch path, the converting direct visit and the email click earn the largest shares, while the opening organic search earns the least. It suits short, fast-moving sales cycles where recency really does signal influence.
U-shaped (position-based)
U-shaped, also called position-based, rewards the bookends of the journey. A common split gives 40% to the first touch, 40% to the last touch, and the remaining 20% spread evenly across the middle touches. The logic: the first touch earns the introduction and the last touch earns the conversion, so both deserve outsized credit. It is a strong default when acquisition and closing both matter and the middle is mostly nurture.
W-shaped
W-shaped adds a third weighted milestone for funnels with a defined hand-off. It typically gives 30% to the first touch, 30% to the lead-creation touch, 30% to the opportunity-creation (or conversion) touch, and splits the final 10% across the remaining middle touches. By crediting the moment a contact becomes a lead and the moment they become a pipeline opportunity, it maps cleanly onto B2B stage transitions.
Models compared side by side
The fastest way to choose is to compare how each model weights the journey, what it rewards, and the journey shape it fits best.
| Model | How credit is split | What it rewards | Best-fit journey |
|---|---|---|---|
| Linear | Equal share to every touch | Consistent presence across the path | Long journeys where each step is roughly equal |
| Time-decay | Rising weight toward touches nearest conversion | Recency and late-stage influence | Short, fast cycles; nurture and retargeting |
| U-shaped (position-based) | 40% first, 40% last, 20% across the middle | Acquisition and closing equally | Two-sided journeys where the middle is light nurture |
| W-shaped | 30% first, 30% lead, 30% opportunity, 10% middle | The three key funnel-stage transitions | B2B funnels with defined lead and opportunity stages |
None of these is objectively correct. Each encodes a different belief about which moments deserve credit. The right one is the model whose assumptions match how your buyers actually move.
When to use which model
Pick the model that matches your sales-cycle length and how clearly your funnel stages are defined, not the one that flatters a particular channel. A short, simple rule of thumb follows.
- Use linear when journeys are long and you want an unbiased baseline, or when you simply cannot argue that any one stage matters more than the others.
- Use time-decay for short sales cycles and for retargeting or nurture programs where recent activity is the strongest signal of intent.
- Use U-shaped when both creating demand and closing it matter, and the middle touches are mostly supporting nurture. It is a sensible default for many marketing teams.
- Use W-shaped when you run a stage-gated B2B funnel and can reliably mark the lead-creation and opportunity-creation touches in your CRM.
A practical habit: do not crown a single model. Run two or three in parallel and watch where they disagree. Channels that look strong under last-touch but weak under U-shaped are usually intent-capture channels riding on demand someone else created, and that disagreement is itself the insight.
What data each model needs
Every multi-touch model depends on one thing above all: a complete, stitched, time-stamped path of touchpoints tied to a single identity. Without that, even the cleverest weighting is built on a broken journey.
The shared requirements
- Identity resolution: touches across devices and sessions, plus anonymous-to-known transitions, must resolve to one person, or the path fragments and credit scatters.
- Consistent tagging: reliable source and medium plus campaign values, usually from disciplined UTM parameters, so each touch can be classified.
- Time stamps: ordered, dated events, which time-decay needs directly and every model needs to sequence the path.
- A defined lookback window: a rule for how far back touches count, so a months-old visit does not silently absorb credit.
The model-specific asks
Linear, time-decay, and U-shaped need only the ordered path and its endpoints. W-shaped needs more: you must mark the lead-creation and opportunity-creation events explicitly, which means your CRM stage data has to be accurate and joined to the touch history. If those stage markers are unreliable, W-shaped degrades quietly into noise, so favor U-shaped until the stage data is trustworthy.
The verdict
Multi-touch attribution is not about finding the one true model; it is about choosing the weighting whose assumptions match your buyer's journey, then reading several models together. Start with U-shaped as a balanced default, move to W-shaped once your CRM stage data is clean, and lean on time-decay for short cycles. Whatever you pick, the model is only as good as the stitched, well-tagged path beneath it. Fix the data first, then the weighting earns its keep.
Frequently asked questions
What is the difference between single-touch and multi-touch attribution?
Single-touch attribution gives all the credit to one interaction, usually the first or last touch. Multi-touch attribution distributes credit across every touchpoint in the journey using a rule, such as evenly or weighted toward key moments. Multi-touch gives a fuller picture of how channels work together.
Which multi-touch attribution model is most accurate?
No rules-based model is objectively accurate; each is an opinion about which touches matter. W-shaped and time-decay tend to reflect complex B2B journeys better than linear, but the most defensible answer is data-driven attribution, which learns weights from your own conversion patterns rather than fixed percentages.
What data do I need to run multi-touch attribution?
You need every touchpoint stitched to one identity across the full journey, time-stamped, and tagged with source, medium, and campaign. That usually means consistent UTM parameters, reliable identity resolution across devices and sessions, and a defined lookback window so the model knows which touches count.
Is U-shaped the same as position-based attribution?
Yes. U-shaped and position-based describe the same model: it gives the first and last touches the largest share, commonly 40 percent each, and splits the remaining 20 percent evenly among the middle touches. The name U-shaped comes from the weighting curve that dips in the middle.
When should I just use single-touch attribution instead?
Single-touch is fine when journeys are short, when you mainly need a directional read, or when your tracking cannot reliably stitch a full path. First-touch suits demand-generation and awareness analysis; last-touch suits conversion optimization. Multi-touch earns its complexity only once journeys span several meaningful touchpoints.
Last updated: 14 June 2026