Marketing attribution is the method you use to work out which marketing and sales touchpoints contributed to a deal, and how much credit each one deserves. Put plainly: attribution answers the question "which efforts actually produced this revenue?" It is how you get from a messy pile of touchpoints – an ad, a webinar, three blog posts, a sales call – back to a defensible answer about where your next euro should go.
This article covers why attribution is harder in B2B than in e-commerce, which attribution models exist, which one you should actually reach for in practice, and how attribution ties into RevOps and your reporting.
What attribution actually is
Picture a prospect who sees a LinkedIn post, reads a blog two months later, signs up for a webinar, and finally requests a demo that turns into a deal. Which of those four moments "caused" the revenue? Attribution is the framework you use to answer that – by assigning credit to one or more touchpoints across the journey from first contact to signed contract. The full definition lives in the glossary entry for attribution.
Why you want to know is simple: without attribution you steer your marketing budget on gut feel. With it, you can show which channels, campaigns and pieces of content genuinely produce pipeline and revenue – and which mostly produce activity and noise.
One distinction is worth holding onto: attribution is not the same as incrementality. Attribution splits credit across the touchpoints you can see; it does not tell you whether the deal would have closed anyway. A branded Google search picks up credit in most models, yet that click is usually the result of all your other work – not the cause of the deal. Keep that in mind before you crown a channel a hero purely because it "scores well".
Why attribution is so hard in B2B
In e-commerce, attribution is relatively tidy: someone clicks an ad and buys a pair of shoes within half an hour. In B2B, almost every assumption underneath that example falls apart. Three reasons:
- Long cycles. A B2B deal can easily run six to eighteen months. The first touch and the final signature sit miles apart, and plenty of tools have long since "forgotten" that opening interaction.
- Buying committees. You are not selling to one person but to a buying committee of roughly seven people on average, each of whom gets involved through a different channel. Attribution built around a single lead misses half the truth by definition.
- Dark social. Most of the real influence happens where you cannot measure it: a tip in a WhatsApp group, a recommendation on a podcast, a chat at a conference. It arrives labelled "direct" or "organic", while the actual cause stays invisible.
The consequence: every attribution model in B2B is an approximation, not exact truth. Anyone who pretends their attribution is accurate to the euro is selling themselves a story. The goal is not perfection – it is a model that points you in the right direction and that your team can agree on.
The main attribution models
Attribution models fall into two families: single-touch, where all credit goes to one moment, and multi-touch, where credit is spread across several.
Single-touch models
- First-touch. All credit to the first interaction. Answers "what brings new people in?" Useful for judging your top of funnel, but it ignores everything that was needed afterwards to close the deal.
- Last-touch. All credit to the final moment before conversion. Answers "what closes deals?" Popular because it is simple, but it overvalues the bottom of the funnel and undervalues everything that created the demand in the first place.
Multi-touch models
- Linear. Every touchpoint gets equal credit. Fair and simple, but it pretends every moment mattered equally – which is rarely true.
- Time-decay. Moments closer to the deal get more credit. Sensible for long cycles, but it penalises the first interaction that started the whole thing.
- U-shaped (position-based). The first contact and the lead-creation moment share the lion's share (often 40/40), with the remaining 20% spread across the middle. A decent balance between origin and conversion.
- W-shaped. Like U-shaped, but with a third peak at the moment an opportunity is created. In B2B this is often the most realistic model, because it recognises the three moments that matter most in the journey.
On top of these sit data-driven models that use statistics or machine learning to divide the credit. Lovely in theory, but they need a lot of clean data and volume – which is exactly what most B2B companies closing a handful of deals a month do not have.
Which attribution model should you use?
My practical take: there is no single correct model – there is a model that fits the question you are asking. Want to judge your top of funnel? Look at first-touch. Want to understand what closes deals? Last-touch. Want a fair overall picture for B2B? W-shaped is usually the best starting point.
Attribution is not accounting, it is a compass. It does not need to be accurate to the euro – it needs to steer you the right way when you decide where your next marketing euro goes.
More important than the model is the discipline around it. A few principles that make the difference in practice:
- Pick one model as your "source of truth" for reporting and stick to it. Switching models between meetings breaks every discussion.
- Measure at the account level, not just the lead level – otherwise you miss the buying committee entirely.
- Combine hard attribution with self-reported attribution ("How did you hear about us?"). For dark social especially, that one plain question is often more accurate than your entire tracking stack.
- Accept an unexplained share. If 20–30% comes back as "direct" or unknown, that is normal in B2B. Do not pretend it is zero.
Do you attribute leads, pipeline or revenue?
A mistake I see constantly: companies attribute on leads when they should be steering on revenue. A channel that generates a hundred cheap leads who never buy looks fantastic in a lead-based report – while it quietly burns your budget. So choose deliberately what you hang credit on:
- Leads. Easy to measure, but says nothing about quality. Only meaningful right at the top of the funnel.
- Pipeline (opportunities). Considerably better: you are measuring what actually became a real chance. For most B2B teams, the most useful layer to steer on.
- Revenue (closed-won). The ultimate measure, but because of long cycles it only arrives late. Pair revenue attribution with pipeline attribution so you can steer faster.
The practical middle ground: steer on pipeline attribution in the short term, and check periodically whether that pipeline actually converts into revenue. Channels that produce plenty of pipeline but little closed-won deserve a hard look – they may be pulling in the wrong people.
Attribution, RevOps and reporting
Attribution is ultimately a RevOps problem. It stands or falls on clean data, a consistent definition of your funnel, and an unambiguous handoff from one stage to the next. If your data model rattles – duplicate contacts, missing source fields, an MQL-to-SQL handoff nobody records consistently – then no attribution model can save you. Garbage in, garbage out.
That is why attribution belongs inside your broader approach to RevOps and the alignment between marketing, sales and customer success. If those three teams each run their own definition of "a won deal", you will still be measuring past one another. How to organise that alignment is something I work through in revenue operations and GTM alignment.
Want to build attribution your team actually trusts – one that steers your budget on revenue instead of clicks? Take a look at how I approach measurable B2B marketing, or run the free GTM Scan to see where your measurement is leaking right now.
Veelgestelde vragen
What is marketing attribution?
Attribution assigns credit for a conversion or deal to the marketing and sales touchpoints that led to it — the basis for understanding what really drives your revenue.
Why is B2B attribution so hard?
Long buying cycles, buying committees with multiple deciders, offline touchpoints, and dark social mean the journey is rarely linear. No model captures the full truth; the goal is better decisions, not perfection.