B2B marketing attribution: a practical guide

The short answer: B2B marketing attribution is the practice of tracing closed revenue back to the marketing activity that helped win it. Because B2B deals take weeks or months and involve several people, it only works when your ad clicks, form fills and CRM records are tied together. There is no single correct model, so choose one on purpose, report it next to a second view, and judge channels on revenue, not leads.

Why B2B attribution is harder than it looks

Online retail can follow a click to a checkout in a single session. B2B cannot. A buyer finds you through a search, reads two articles, leaves, comes back from a LinkedIn ad a month later, and fills in a form. Their colleague attends a demo. The deal closes in the CRM four months after the first click.

• Long cycles. The first touch and the closed deal can be months apart, longer than many platform lookback windows.
• Several people. A buying group visits from different devices and accounts. Analytics sees them as separate visitors.
• The result lives in the CRM. Qualified, proposal, closed-won and deal value are recorded in your CRM, not by the ad platforms.
• Small numbers. A handful of deals a month means each one moves the picture, and statistical models have little to learn from.
• Hidden touches. Word of mouth, podcasts, private communities and "I saw you on LinkedIn" rarely leave a click behind.

That is why B2B attribution is mostly a data-joining job before it is a modelling job. See also why ad platforms do not agree with your CRM.

The four inputs you need before choosing a model

A model can only share out credit for touches it can see. Check these four first.

• 1. A source on every lead. UTM tags and ad click IDs saved onto the CRM record when the form is submitted. Without them, the record shows "unknown" however good your analytics is. Start with what UTM parameters are.
• 2. Agreed stage definitions. What counts as a lead, a qualified lead, an opportunity and a win. If sales and marketing disagree, every report will too.
• 3. Deal values and dates. Revenue attribution needs an amount and a close date on each won deal, entered consistently.
• 4. One place that joins them. Ad spend, website behaviour and CRM outcomes have to meet somewhere, whether that is a spreadsheet, the CRM itself or a marketing data warehouse.

A free marketing data audit checks exactly these four, and the attribution gap calculator sizes the overlap between what platforms claim and what the CRM records.

Which attribution model fits B2B

An attribution model is a rule for sharing credit across the touches that led to a sale. The common ones, and where each helps or misleads in B2B:

• First touch gives all credit to the first known interaction. It shows what creates demand, and ignores everything that nurtured it.
• Last touch gives all credit to the final interaction before conversion. It flatters branded search, retargeting and direct, which tend to harvest demand that other channels created.
• Linear splits credit equally. It is fair and easy to explain, and it treats a passing blog visit the same as the demo request.
• Position-based (U-shaped) gives most credit to the first touch and the lead-creation touch, and shares the rest. It often suits B2B, because those two moments carry the most weight.
• Time decay gives more credit to touches nearer the sale. It suits shorter cycles and promotional pushes.
• Data-driven uses statistical modelling to assign credit. It needs enough conversion volume to learn from, which many B2B accounts do not have, and it can be hard to audit.

One practical note: GA4 and Google Ads now offer only data-driven and last-click models, so first-touch and position-based views have to be built in your own reporting layer, from the CRM and the tagged sources.

A sensible default for most B2B teams is to report two views side by side: first touch, to show which channels create demand, and position-based or last non-direct touch, to show which channels convert it. When the two views disagree, that disagreement is useful information, not an error. Compare them on your own data with the attribution model comparison tool, and read how each behaves in marketing attribution models, judged against your CRM.

Leads versus revenue: attribute to the thing that pays

Most dashboards attribute leads, because leads are quick to count. B2B budgets should be judged on pipeline and closed revenue, because the channel that produces the cheapest leads is often not the one that produces customers.

• Lead attribution tells you where enquiries come from. It is fast, and it rewards volume.
• Pipeline attribution tells you where qualified opportunities come from. It is the earliest reliable signal.
• Revenue attribution tells you where closed deals come from. It is the slowest and the most honest, and it needs the CRM.

Two terms come up in reporting. Marketing-sourced pipeline is where marketing created the first known touch. Marketing-influenced pipeline is where marketing touched an opportunity at any point. Influenced is always the larger number, and it is easy to inflate. Report both, label them, and agree the definitions with sales before the first meeting.

Where the numbers go wrong

• Platforms mark their own homework. Google, Meta and LinkedIn each claim a conversion they touched, so the total exceeds what the CRM closed. See Meta and Google both claiming the same conversion.
• Unmatched records. Leads that arrive by phone, email or a sales rep never carried a source. They sit in "direct" or "unknown", which makes some channels look weaker than they are.
• Consent and blockers. Visitors who decline tracking, or use blockers, do not appear in browser data. The gap grows when consent defaults are strict.
• Direct and dark traffic. A growing "direct" bucket often means untagged links, private shares or dark social being counted as nothing.
• Late CRM updates. If stages are updated weeks after they happen, monthly reports undercount the recent past.
• Changing definitions. Rewording a stage or changing a lookback window breaks comparisons with earlier periods.

A useful low-tech addition is a self-reported question on forms, such as "How did you first hear about us?". It catches podcasts, referrals and communities that no tracker sees. Use it alongside tracked sources, not instead of them, and keep the free-text answers.

A practical setup, step by step

• 1. Tag every campaign link. One naming convention, lowercase, written down. The UTM builder helps keep it consistent.
• 2. Capture the tags and click ID. Hidden form fields save source, medium, campaign and the ad click ID onto the CRM lead.
• 3. Define the stages. Agree lead, qualified, opportunity and won, and the fields that prove each one.
• 4. Send outcomes back to the ad platforms. Use offline conversion tracking so Google, Meta and LinkedIn can optimise toward qualified leads and deals, not form fills.
• 5. Land everything in one place. Pull ad spend, analytics and CRM data into a single reporting layer, and deduplicate across platforms.
• 6. Pick the model and write it down. Choose your primary and secondary view, the lookback window and the definition of a touch, and keep them stable.
• 7. Review quarterly. Check match rates and stage definitions, and compare model outputs, not only channel totals.

What to report each month

• Spend, leads, qualified leads, opportunities and won revenue by channel, on the same rows.
• Cost per qualified lead and cost per opportunity next to cost per lead.
• Customer acquisition cost and payback by channel; see CAC payback and LTV:CAC, explained.
• Share of leads with a known source, so everyone can see how much of the picture is missing.
• The two model views side by side, with a note on where they disagree.
• A short note on what changed in tracking, definitions or consent that month.

When you need a warehouse, and when you do not

You do not need a warehouse to start. With a few channels and a handful of deals a month, a well-kept CRM with captured sources and a monthly spreadsheet is a legitimate first version.

A warehouse earns its place when you run several channels with overlapping claims, when reporting takes days of manual exports, when you need to deduplicate across platforms, or when you want the data in your own account regardless of which tool you use next. It also stops you rebuilding the logic every time a platform changes. How that is built is covered on our Data and BI page, and an example of the result is in the Unsigned case study.

What attribution cannot tell you

Attribution assigns credit under a set of rules. It does not prove that a channel caused a sale. A branded search click may have been triggered by a podcast. A retargeting ad may have been shown to someone who would have bought anyway.

To test cause, use experiments: hold out a region or an audience from a channel for a few weeks and compare, or pause a channel for a short test and watch pipeline. With B2B volumes these tests are slow and imperfect, but even a rough one is a useful check on a model. Treat attribution as a decision aid, report how much of the data is missing, and avoid presenting the output as precise.

Questions people ask

Related: multi-touch attribution, conversion tracking: a practical setup guide and MQL vs SQL.