Marketing Attribution Models Explained: Formulas & Examples

What is a marketing attribution model?

A marketing attribution model is the rule that decides which touchpoints get credit when someone converts. Someone might see a Meta ad, click a Google Shopping ad a few days later, open an email, then search your brand name and buy. The attribution model decides whether that sale belongs to the Meta ad, the Shopping ad, the email, the brand search, or all of them in some proportion.

That decision has real consequences. The model you pick determines which channels look profitable, which campaigns get more budget and which get cut. Two analysts looking at the same sales data can reach opposite conclusions about the same ad channel simply because they use different models.

This guide explains each model with its formula, runs all of them against the same $200 order so you can see the differences in actual dollars, covers which models Google and Meta still offer, and gives you a practical way to choose.

The attribution models at a glance

---

One order, seven models: a worked example

Rules of thumb are easier to trust when you can see the arithmetic. Take one customer who buys a $200 order after four touchpoints over ten days:

Here is how each fixed-rule model splits the $200:

The total never changes. Only the story does. The Meta ad is worth $200 under first-touch and $0 under last-touch for the exact same customer and the exact same order. If you only ever look at last-touch, the Meta ad that started this journey looks like it did nothing - and a budget review based on that report would cut it.

The rest of this guide shows where each of these numbers comes from.

---

First-touch attribution

First-touch attribution gives 100% of the credit to the first touchpoint in the journey.

Credit(first touchpoint) = 100%
Credit(every other touchpoint) = 0%

In the example, the Meta ad gets all $200 because it was the first thing the customer interacted with.

Use it to answer: "Which channels bring in new customers?" It is the right lens for judging prospecting and brand-awareness spend.

Watch out for: it gives zero credit to the email, retargeting and search ads that actually converted the customer. A channel that is brilliant at closing but never first will look worthless. It is also fragile in practice, because it depends on you reliably recording the first visit - which cookie limits and ad blockers often prevent.

---

Last-touch (last-click) attribution

Last-touch attribution, often called last-click attribution, gives 100% of the credit to the final touchpoint before the conversion.

Credit(last touchpoint) = 100%
Credit(every other touchpoint) = 0%

In the example, branded search gets all $200.

Use it to answer: "What closes sales?" It is simple, easy to audit, and is still the default view in many ad platforms and analytics tools.

Watch out for: last-touch systematically favors bottom-funnel channels such as branded search and retargeting. Those channels often harvest demand that other channels created, so last-touch rewards the harvest and ignores the planting. Over time, budgets drift toward the bottom of the funnel while the top of the funnel starves. For a deeper look at this trap, see What is multi-touch attribution?

---

Linear attribution

Linear attribution splits credit equally across every touchpoint. This answers the common search question "how does the linear attribution model calculate credit?":

Credit per touchpoint = 100% / N      (N = number of touchpoints)

Four touchpoints means 25% each, which on a $200 order is $50 each. Ten touchpoints would mean 10% each.

Use it when: journeys are long, many channels contribute, and you want a first, unopinionated view of the whole path before choosing something more selective.

Watch out for: linear assumes a passing impression matters as much as the decisive click. It also dilutes credit when journeys include many low-value touches, which makes every channel look average.

---

Time-decay attribution

Time-decay attribution gives more credit to touchpoints that happened closer to the conversion. The most common version uses an exponential curve with a half-life: a touchpoint's weight halves for every half-life that passes before the conversion.

Weight = 2 ^ ( -days before conversion / half-life )
Credit = Weight / sum of all weights

With the common 7-day half-life, a touchpoint on the day of purchase has weight 1.00, one seven days earlier has weight 0.50, and one fourteen days earlier has weight 0.25.

For the example journey:

Use it when: the purchase decision is made over days rather than months, and recency really does signal intent - promotional periods and impulse-leaning e-commerce are typical cases.

Watch out for: it still undervalues the channel that started the journey (the Meta ad gets about 14%), and the half-life is an assumption you choose. A 7-day half-life suits a one-to-two-week buying cycle, not a six-month B2B one.

---

Position-based (U-shaped) attribution

Position-based attribution, also called U-shaped attribution, rewards the two moments many marketers consider most important: how the customer first found you, and what finally got them to buy.

First touchpoint:   40%
Last touchpoint:    40%
Middle touchpoints: 20% split equally

In the example, the Meta ad and branded search get 40% each ($80), and the two middle touches split 20% ($20 each, 10% apiece). Short journeys need a rule of their own, because there is no "middle" to share. A common choice is a 50/50 split for two touchpoints and 100% for a single touchpoint.

Use it when: you want a balanced default that credits both discovery and closing and still acknowledges the middle of the journey. It is easy to explain to stakeholders, which matters more than people admit.

Watch out for: 40/20/40 is a convention, not something measured from your data. If your nurture emails are what actually convert people, a fixed 20% for the middle under-credits them.

---

W-shaped and full-path attribution

W-shaped and full-path models extend position-based thinking to B2B and lead-generation funnels where the journey has defined stages tracked in a CRM. Instead of two anchor points, they use milestone touchpoints.

HubSpot, for example, defines its two models like this:

• W-shaped: 30% each to the first interaction, the lead-creation interaction and the deal-creation interaction, with the remaining 10% shared equally by other interactions.
• Full-path: 22.5% each to the first interaction, lead creation, deal creation and the interaction that closed the deal, with the remaining 10% shared equally by other interactions. This is the answer to the common question "how does the full path attribution model calculate credit?" (HubSpot's definitions)

Use them when: you sell with a multi-stage pipeline (lead, opportunity, closed-won) and your CRM records which touchpoint created each stage.

Watch out for: they require clean CRM stage data and are not a fit for a typical online store, where the whole journey ends in one checkout. That is why most e-commerce attribution tools, SignalBridge included, work with the first five models rather than these.

---

Data-driven attribution

Data-driven attribution (DDA) replaces fixed rules with an algorithm. Instead of saying "the first touch gets 40%", it looks at the paths of people who converted and people who did not, and estimates how much each kind of touchpoint changed the likelihood of converting.

Google describes its version as using your account's own conversion data to calculate the actual contribution of each ad interaction across the conversion path (Google Ads Help). It is the default model for most Google Ads conversion actions and is the recommended model in GA4.

Use it when: you run a meaningful volume of Google Ads and want the platform's own bidding to value interactions more precisely.

Watch out for: it is a black box you cannot audit, it only sees the data inside that platform (Google's model cannot see your Meta or email touches), and each platform's model grades its own homework. Two platforms can each claim the same sale. If you have seen Meta and Google both report the same order, that overlap is why - see why Facebook and Google attribution never match.

---

Which attribution models can you still use in Google Ads, GA4 and Meta?

This is where many guides are out of date. The major ad platforms have been narrowing their attribution options, which pushes the more transparent rule-based models out of the platforms and into your own measurement.

Two takeaways:

1. If you want to compare first-touch, linear, time-decay or position-based credit, you have to compute them yourself - from your own first-party touchpoint data or a tool that stores it. The big platforms will not do it for you any more.
2. Platform attribution only describes that platform. Google's model sees Google interactions, Meta's reporting sees Meta interactions. Neither can tell you how your email, organic and paid channels combined to produce a sale.

Attribution model vs. attribution window

These two terms are often mixed up, and Meta's changes make the distinction important:

• The attribution model decides how credit is divided among touchpoints.
• The attribution window decides how far back a click or view can still receive credit (for example 7 days after a click).

A change to the window, like Meta's January 2026 removal of longer view windows, can make reported conversions drop without any change in actual sales. It changes what the platform is willing to count, not what customers did. For more on reading these numbers, see how to read Facebook Ads Manager after iOS changes.

---

How to choose a marketing attribution model

There is no universally correct model. The right choice depends on how your customers buy and what decision you are trying to make.

For most e-commerce brands the most useful setup is not a single model but a comparison:

1. Pick three models that disagree on purpose - first-touch, last-touch and one spread-out model such as time-decay or position-based.
2. List credit per channel under each. Look for the channels whose credit changes the most between models. Those are the channels where your budget decision is most sensitive to the model you chose.
3. Read the pattern. A channel strong in first-touch but weak in last-touch is probably an introducer. One strong in last-touch but weak in first-touch is probably a closer that harvests demand created elsewhere. These need different jobs and different success metrics.
4. Move budget in steps, not leaps. Shift a modest share, then watch blended results. Do not rebuild your media plan from a two-week model comparison.
5. Check big moves with an incrementality test. Attribution models split credit among touches that happened; they cannot prove a touch caused the sale. A geo or holdout test can. See What is conversion lift?

Related views that help: assisted conversions show which channels help without closing, and true ROAS puts spend next to the credited revenue.

---

Why your attribution data limits every model

Every model above assumes you captured the touchpoints. If a visitor's first ad click was blocked by an ad blocker, or lost when a browser cleared a short-lived cookie, then:

• First-touch credits whatever channel happened to be the first recorded one.
• Last-touch credits the last recorded one, usually a direct visit or branded search.
• Linear, time-decay and position-based divide credit over a path with holes in it.

Changing the model cannot repair missing data. The foundation is capturing touchpoints reliably on your own domain with server-side tracking and first-party data, so the journey you analyze is the journey that actually happened. For the scale of the problem, see What is ad blocker tracking loss?

---

Comparing the models in SignalBridge

SignalBridge's multi-touch attribution calculates five models side by side for every channel, so you can see the first-touch / last-touch gap described above without building a spreadsheet:

Credit is shown in both conversions and revenue per channel, built from server-side events and each visitor's recorded touchpoints. It also does not offer W-shaped or full-path models, which depend on CRM stages. Pair it with the conversion journeys and assisted conversions views to see the actual paths behind the numbers.

---

FAQ

What is an attribution model?

An attribution model is a rule or algorithm that decides which marketing touchpoints receive credit for a conversion. Some models give all the credit to one touchpoint (first-touch or last-touch), while others spread it across the journey (linear, time-decay, position-based) or let an algorithm estimate each touchpoint's contribution (data-driven).

What are the main types of marketing attribution models?

The seven you will meet most often are first-touch, last-touch, linear, time-decay, position-based (U-shaped), W-shaped or full-path, and data-driven. The first two are single-touch models, the next three are fixed-rule multi-touch models, W-shaped and full-path are CRM-stage models for B2B, and data-driven is algorithmic.

How does the linear attribution model calculate credit?

Linear attribution divides credit equally among all touchpoints. The formula is 100% divided by the number of touchpoints. A journey with four touchpoints gives each 25%, so on a $200 order each touchpoint receives $50.

How does the full-path attribution model calculate credit?

In HubSpot's definition, full-path attribution gives 22.5% of the credit each to the first interaction, the lead-creation interaction, the deal-creation interaction and the interaction that closed the deal. The remaining 10% is shared equally by all other interactions. It is designed for B2B pipelines with CRM stages, not single-checkout e-commerce.

What is the difference between first-touch and last-touch attribution?

First-touch gives 100% of the credit to the first interaction and answers "what introduced this customer?". Last-touch gives 100% to the final interaction and answers "what closed the sale?". They often point at completely different channels for the same order, which is why comparing both is more informative than picking one.

Which attribution model is best for e-commerce?

There is no single best model. Short-cycle stores often do well with last-touch compared against time-decay, while stores that run prospecting, retargeting and email together usually learn more from position-based or time-decay compared with first-touch. The most reliable approach is to compare several models and validate big budget changes with an incrementality test.

What attribution model does Google Ads use?

Google Ads offers data-driven attribution, which is the default for most conversion actions, and last click. The first-click, linear, time-decay and position-based models are no longer supported, and conversion actions that used them were upgraded to data-driven.

Which attribution models does GA4 support?

GA4 supports data-driven attribution, paid and organic last click, and Google paid channels last click. First click, linear, time decay and position-based were removed in November 2023.

Is last-click attribution still worth using?

Yes, as one lens, not as the only one. Last-click is simple and useful for understanding which channels close sales, but used alone it over-credits bottom-funnel channels such as branded search and retargeting and under-credits the channels that created the demand.

Can I use first-touch or linear attribution if Google no longer offers them?

Yes, but you need your own touchpoint data. Because Google Ads and GA4 no longer provide those models, you calculate them from first-party journey data or use an attribution tool that stores every touchpoint and computes the models for you.

---

Related reading

• What is Multi-Touch Attribution? - the concept, implementation steps and common mistakes
• Facebook Ad Attribution: The Complete Guide - how Meta attributes conversions and how to cross-check it
• Why Your Facebook and Google Ads Numbers Never Match - why platforms disagree about the same sale
• What Are Assisted Conversions? - the channels that help without closing
• What is Conversion Lift? - testing what actually caused a sale
• How to Calculate True ROAS - spend versus credited revenue when pixels miss conversions
• Shopify Attribution: How to Track Which Ads Drive Real Sales - attribution on a Shopify store
• Why Your Ads Are Lying to You - how platform reporting inflates results

Try SignalBridge free - capture every touchpoint server-side and compare five attribution models on your own data.