Getting Started
October 19, 2026
Multi-touch attribution: what it is, 4 models, and how to set it up
Vincent Gaemers
On this page
Key takeaways
- Multi-touch attribution splits the credit for a lead, deal, or sale across every touchpoint in the journey, instead of handing all of it to the first or the last one.
- The 4 rule-based models give 4 different answers on the same deal. On our €12,000 example, LinkedIn earns €0 in last touch and €5,600 in U-shaped.
- Google removed first click, linear, time decay, and position-based attribution from GA4 and Google Ads in 2023. What's left is last click and a data-driven model you can't explain.
- For long B2B cycles, default to U-shaped and check it against linear. A channel that ranks high in both is safe to scale.
- Most wrong numbers come from the setup, not the model: visitors who are never identified, colleagues who are never grouped into 1 company, a lookback window shorter than the sales cycle, or a messy CRM.
What is multi-touch attribution?
Multi-touch attribution splits the credit for a lead, deal, or sale across all the marketing touchpoints that came before it, instead of giving 100% to 1 of them. It answers "which channels made this revenue happen?" rather than "which channel was there last?"
Single-touch models keep 1 moment. First touch gives everything to the channel that started the journey. Last touch gives everything to the channel that finished it. Both are easy to read, and both are wrong the moment a buyer needs more than 1 visit to decide. In B2B and in big-ticket B2C, that's nearly every buyer.
How many visits are we talking about? One B2B buyer viewed 51 pages over 6 visits in 90 days before signing. In real estate, 1 lead easily carries 21+ touchpoints. A single-touch report keeps 1 of those and throws the rest away.
Why it matters more since 2023
In 2023, Google removed first click, linear, time decay, and position-based attribution from GA4 and Google Ads. What's left is last click and data-driven attribution: a model Google trains on your conversion paths, with weights you can't see, that stops at your website. It never sees the CRM deal or the invoice.
So if you want a split you can explain in a budget meeting, and one that runs through to CRM revenue, you need a tool built for it. That's what multi-touch attribution software does.
How does multi-touch attribution work?
Every setup has the same 4 parts.
- Touchpoints. Every interaction with your brand: an ad click, a page view, an organic visit, a visit from a ChatGPT answer. Page views in 1 sitting form 1 visit. A new visit starts after a 30-minute gap, or when someone arrives from a different campaign. Each visit carries its source, campaign, UTM parameters, and click ID.
- The journey. All of a person's visits up to the moment they become a lead: a form fill, a demo booked, a trial started. Only visits inside the journey earn credit. After that moment, sales, email, and phone take over.
- The outcome. What you're splitting: leads, CRM deals and pipeline, or revenue from payments. The further down the funnel, the more useful the answer. Lead credit tells you who fills in forms. Revenue credit tells you who pays.
- The model. The rule that decides how much credit each visit gets. The 4 common ones are below.
Two things decide whether the numbers hold up.
Identity. A cookie knows a browser, not a person. Here's a typical journey: a Meta ad brings an anonymous visit, a Google search brings a second visit 4 days later, and 3 days after that a brochure form puts an email on both. The brochure is opened on another device the same day, the CRM logs a meeting 2 weeks later, and the deal is won 40 days after that first ad visit. The form is the moment you identify the person, and every earlier visit, on any device, joins 1 journey. In B2B you go 1 step further and group colleagues into 1 company journey. That's how one team found that 9 reported leads were 87 at company level, and that about 50 of them came from LinkedIn without ever filling in a form. (How company attribution works.)
One form fill names every visit before it
A cookie knows a browser, not a person. Here is what happens when someone shares an email.
Example journey from this guide.
The revenue join. Credit is only as good as the revenue behind it. Deals from your CRM and payments from Stripe are matched to the journey that produced them. With cohort-aligned revenue (e.g., all leads first touched in March), revenue is reported in the period the marketing happened, not the month the invoice was paid. A 6-month sales cycle no longer makes last quarter's campaigns look like waste.
The 4 multi-touch attribution models, on 1 deal
Here's 1 buyer's journey. The numbers are illustrative, the shape is typical B2B.
- Day 1. Clicks a LinkedIn ad.
- Day 9. Reads a blog post from organic search.
- Day 23. Clicks a LinkedIn retargeting ad.
- Day 40. Arrives from a ChatGPT answer.
- Day 52. Searches your brand on Google, clicks the ad, and books a demo.
The deal closes at €12,000. Here's how each model splits it.
First touch
All credit goes to the first visit. LinkedIn gets €12,000.
Use it to see which channels create new demand. Blind spot: it ignores everything that moved the buyer from curious to ready.
Last touch
All credit goes to the last visit before the demo. Google Ads gets €12,000.
Use it when people buy in 1 visit. Blind spot: brand search and retargeting look like heroes, and the channels that created the demand look like waste. This is the model that gets LinkedIn budgets cut.
Linear
Every visit gets the same credit: 5 visits, €2,400 each. LinkedIn (2 visits) gets €4,800. Organic, AI, and Google Ads get €2,400 each.
Use it as a neutral check on the other models. Blind spot: a 10-second bounce counts as much as the visit that started it all.
U-shaped (position-based)
40% to the first visit, 40% to the visit where the lead converted, and 20% spread over the middle. LinkedIn gets €5,600 (€4,800 + €800), Google Ads €4,800, organic and AI €800 each.
Use it when cycles are long and opening and closing both matter. That's most B2B. Blind spot: the middle gets little credit. If nurture is your strategy, compare it with linear.
Same deal, 4 answers
Same deal, same data: LinkedIn earns €0 or €12,000
One €12,000 deal, 5 visits, 4 rules for splitting the credit.
Illustrative example: one €12,000 B2B deal with 5 visits.
Same deal, same data, and LinkedIn swings from €0 to €12,000 depending on the rule. The lesson: never read 1 model on its own.
Other models you'll run into
- Time decay gives more credit the closer a visit is to the conversion. Fine for short, promotion-driven cycles. On a long B2B cycle it behaves a lot like last touch.
- W-shaped adds a third peak at opportunity creation, often 30/30/30 with 10% for the rest. It's only as good as the opportunity dates in your CRM.
- Data-driven models learn weights from paths that did and didn't convert. They need volume to be stable, and you can't explain the weights in a budget meeting.
Spectacle runs the 4 models above because each one fits in a sentence your CFO can repeat.
Which multi-touch attribution model should you use?
| Situation | Use | Check it against |
|---|---|---|
| Sell in 1 or 2 visits | Last touch | First touch |
| Want to know which channels create new demand | First touch | Linear |
| Run long B2B or high-consideration cycles | U-shaped | Linear |
| Need to defend budget for a top-of-funnel channel | U-shaped | Last touch, to show the gap |
| Run a nurture-heavy program (webinars, email, content) | Linear | U-shaped |
Our take: default to U-shaped and keep linear open next to it. A channel that ranks high in both is safe to scale. A channel that swings between models deserves a look at the actual journeys before anyone touches its budget.
What changes when you stop reading last click
4 situations from Spectacle accounts, anonymized, where a different view changed a budget decision.
Long-cycle B2B: the LinkedIn budget that almost got cut
In the last-click report, LinkedIn "didn't convert". In the journeys, it touched about 43% of pipeline and about half of 12-month revenue. At another company, a $16M deal traced back to 1 organic LinkedIn post, 4 months before the first form fill.
Action: judge LinkedIn on company-level pipeline, not on form fills. How company-level LinkedIn tracking works.
Long-cycle B2C: the "direct" sale that wasn't
A €29,000 purchase showed up in GA4 as direct. Traced over about 90 days of visits, it had a full journey behind it that GA4 never connected to the sale.
Action: treat direct as a measurement gap, not a channel, and keep every touchpoint for as long as the decision takes.
Real estate: buyers vs brochure collectors
Measured per CRM contact instead of per form fill, 1 channel cost €878 per contact and another €128. The visitors who used the price calculator were the strongest buyers.
Action: cut the expensive channel and retarget the people who used the price calculator.
Subscription and membership: the first payment isn't the value
A €49-a-month subscriber pays €588 in a year, and €539 of it lands after Meta's 7-day click window has closed. Cohort revenue (e.g., all members who joined in March) follows each group for 12 months, so renewals land on the campaign that brought the member in instead of on nothing.
Action: track a separate event per product type, and judge channels on 12-month revenue. Stripe attribution: the cheapest trials were worth the least.
Common multi-touch attribution problems (and how to fix them)
Most wrong numbers come from the setup, not the model. These are the ones we see most in tracking reviews.
On a 7-month funnel, GA4 and Meta never see the first touch
Both count back from the conversion. Anything older than their window drops out of the report.
Source: GA4 lookback cap and Meta's 7-day click window. 1 month = 30.4 days.
- A lookback window shorter than the sales cycle. GA4 caps its lookback at 90 days, and Meta stops counting 7 days after the click. On a 7-month funnel that erases the first touch. Use a tool that keeps every touchpoint, with no lookback limit.
- "Everything is direct." Direct is a measurement gap, not a channel. Consent walls, device switches, and redirects that strip UTMs all create it. In 1 review, 2 cookie banners on the same site wiped the same visitor twice. Route tracking through your own domain with server-side tracking, identify on every form, and keep UTMs consistent (UTM guidelines, or check yours with UTM Scout). For Safari click IDs, see the GCLID fix.
- Tracking that quietly breaks or doubles. After a plugin change in 1 account, only about 60% of form submits were still firing, and the report looked normal. In another, the same purchase reached Meta 3 times through overlapping tags. Write down every event with a definition and a destination, use 1 tag per platform, and send a transaction ID with every revenue event.
- People counted, companies missed. Without grouping, 3 colleagues are 3 partial journeys. Use group() and turn on company attribution.
- Revenue that's 100x off. Revenue sent in euros instead of cents, or 2 CRM amount fields fighting each other. Send revenue in cents and clean the CRM first (step 1 below).
- Tracking on some pages only. If the snippet is only on the homepage, journeys break at the first click. On single-page apps, call page() on every route change, or every session counts as 1 page view.
- Reading credit as cause. Multi-touch attribution shows which touchpoints were present. To prove a channel caused lift, run a holdout test.
How to set up multi-touch attribution in 5 steps
Most teams are done in about 30 minutes. The CRM step takes longest, and it's the one that decides whether you trust the numbers.
1. Clean 5 CRM fields first
1 currency and 1 amount field per deal, 1 record per company, a defined lead stage, source fields nobody overwrites, and real closed-won dates. Every attribution tool, Spectacle included, reports garbage on a messy CRM.
2. Track every page
Add the JavaScript snippet to the head of every page, use the Google Tag Manager template, or send events through Segment or the REST API. Route it through your own domain with server-side tracking, so ad blockers and browser limits don't cut journeys short.
3. Identify early, group in B2B
Call identify() whenever someone shares contact details: forms, signups, demo bookings. With form auto-tracking on, any form that holds an email does this for you. In B2B, call group() too, so colleagues roll up into 1 company, and switch on company attribution.
4. Set the moment a lead converts
Pick the events that mark a visitor becoming a lead: a form fill, a demo booked, a trial started. Everything up to that moment earns credit, everything after belongs to sales. Leave it empty and Spectacle uses the date of the person's first deal. There is no lookback window to set: every touchpoint stays on the journey for as long as the cycle runs. (Core concepts.)
5. Connect revenue and ad spend, then put the report in a meeting that already exists
Connect your CRM (HubSpot, Salesforce, Dynamics 365, Odoo), payments (Stripe), and ad accounts, so spend and revenue import on their own. Then put the report in the weekly sales meeting. Adoption is scheduled, not assumed.
Multi-touch attribution vs other ways to measure marketing
| Method | What it answers | Best for | Watch out |
|---|---|---|---|
| Multi-touch attribution | Which touchpoints were in the journeys that produced revenue | Weekly budget calls per channel, campaign, and keyword | Shows presence, not cause |
| Marketing mix modeling | How spend levels relate to outcomes over time | Large budgets, offline channels, yearly planning | Needs years of history, slow to update |
| Incrementality testing | What happens when you switch a channel off | Proving a channel causes lift | 1 channel at a time, and the test costs spend |
| Self-reported ("How did you hear about us?") | What the buyer remembers | Podcasts, word of mouth, communities | Memory favors the last thing someone saw |
They work best together. Use multi-touch attribution for the weekly decisions, and a holdout test when a big budget call needs proof.
Multi-touch attribution FAQ
What's the difference between single-touch and multi-touch attribution?
Single-touch gives 100% of the credit to 1 touchpoint, the first or the last. Multi-touch splits it across every touchpoint in the journey, so the channels that open, nurture, and close a deal all get counted.
What's the best multi-touch attribution model for B2B?
U-shaped is the best default for most B2B teams. It credits the channel that started the journey and the one that turned a visitor into a lead. Check it against linear: a channel that ranks high in both is safe to scale.
Does GA4 have multi-touch attribution?
Partly. Since 2023, GA4 offers data-driven attribution and last click only. Data-driven does spread credit, but its weights aren't visible, its lookback stops at 90 days, and it never sees CRM revenue.
How long should the attribution window be?
As long as your sales cycle. A 30-day default on a 7-month funnel cuts off the first touch, and some high-consideration B2C journeys start 3+ years before the purchase. Spectacle keeps every touchpoint with no lookback cap.
Does multi-touch attribution still work with cookie restrictions?
Yes, if you collect first-party data on your own domain and identify people when they share their details. Identification joins visits across devices into 1 journey, and grouping joins colleagues into 1 company.
How much does multi-touch attribution software cost?
Spectacle starts at €99 a month. Company-level journeys, cohort revenue, and HubSpot, Pipedrive, and Odoo are on the €349 plan. Salesforce, D365 include customer success, SSO and SLA for our clients who require more. Enterprise attribution tools usually start elsewhere around €2,000 a month. See pricing.
See every touchpoint behind your revenue
Your last-touch report is giving 1 channel credit for work that 4 channels did. Spectacle shows all 4 models on every report, follows each journey into your CRM deals and Stripe payments, and rolls people up into companies. It starts at €99 a month and takes about 30 minutes to set up.
Start a free 14-day trial or book a demo.
Related reading
- Revenue attribution analysis explained
- Cohort-aligned revenue: a different attribution model
- How to connect ad spend to revenue
- Google Ads attribution: the complete revenue guide
- HubSpot revenue attribution reporting
- Multi-touch attribution for product-led companies
- The 7 best marketing attribution tools for SaaS
- Stripe attribution: the cheapest trials were worth the least
Latest insights