B2B Marketing Attribution: Models That Actually Work

Attribution is the question of which marketing activities are responsible for revenue. It should be simple. It isn't, and it never quite will be. Not in B2B, where buying decisions involve multiple people over months, across channels that can't all be tracked, with relationship and reputation factors that never appear in a dashboard.

The problem isn't that attribution is unsolvable. It's that most businesses are using the wrong models and drawing the wrong conclusions from them.

What the standard models actually measure

First-touch attribution assigns 100% of the credit for a deal to the first marketing touchpoint. If someone clicked a LinkedIn ad, came to your website, and later became a customer, that LinkedIn ad gets all the credit for the revenue. Everything that happened between that click and the signature is invisible.

First-touch tells you what channels are best at generating initial awareness. That's useful. But it tells you nothing about what kept people engaged, accelerated their decision, or converted intent into action.

Last-touch attribution assigns 100% of the credit to the final touchpoint before conversion. Usually this is a direct visit or branded search: the person typing your company name into Google and clicking the homepage. It credits the moment of intent without crediting anything that built that intent in the first place.

Last-touch attribution systematically understates the value of upper-funnel activity . It rewards the final step and ignores everything that made the final step possible. Businesses using last-touch tend to over-invest in branded search and retargeting, and under-invest in content, events, and channels where the buying journey begins.

Why multi-touch models don't fully solve it either

Linear attribution spreads credit equally across all touchpoints. Time-decay weighting gives more credit to recent touchpoints. Position-based models give more to first and last. Each of these is a different way of distributing credit, but they all suffer from the same underlying problem: they can only measure what you can track.

In B2B, a significant portion of the buying journey is invisible. Your prospect reads an article about their problem and remembers your brand. Their colleague mentions your name in a meeting. They search your company on LinkedIn before replying to an email. They see a conference talk you gave six months ago. None of these touchpoints show up in your attribution model. They happen in what's often called "dark social": the private, untracked interactions where most actual influence occurs.

Multi-touch models are better than single-touch models. They're not complete. The longer and more complex your sales cycle, the more invisible touchpoints you have, and the less accurately any tracking-based attribution model reflects reality.

What to do instead: influenced pipeline tracking

The most pragmatic approach for most B2B businesses isn't to chase perfect attribution. It's to track influenced pipeline alongside your other metrics .

Influenced pipeline asks: across the deals in your pipeline and closed-won revenue, which marketing activities were involved in the journey? Not which activity gets the credit, but which activities touched the opportunity at some point.

This is a fundamentally different question. It doesn't try to solve the credit allocation problem. It asks whether your marketing activities are showing up in pipeline, which is ultimately what you care about.

In practice, this means tracking at the opportunity level: what channels did this contact come through? What content have they engaged with? What events did they attend? When you pull this data across a cohort of closed deals, patterns emerge. The channels and activities that show up most consistently in successful deals deserve investment. The ones that rarely appear probably don't.

Getting leadership aligned on imperfect data

The organisational challenge with attribution is that leadership often wants a clean answer: which channel is producing ROI, and which ones should we cut? Attribution models feel like they provide that answer. They don't. They provide a model of the answer, with structural biases depending on which model you've chosen.

The healthier conversation is: what do we know, what are we unable to track, and what does the combined evidence suggest? This requires more nuance from the marketing team and more tolerance for ambiguity from leadership. It also tends to produce better decisions.

The metric framework we typically use with clients combines: influenced pipeline (which activities show up in deals), pipeline-to-close rates by channel (which sources produce the highest-quality opportunities), and closed-won revenue by self-reported source (asking customers where they heard about you, which captures dark social better than tracking). No single metric tells the whole story. Together, they give a cleaner picture than any attribution model does alone.

A practical starting point

If you're currently using last-touch attribution and using it to make channel investment decisions, the most important thing you can do is add one question to your CRM : "Where did this contact first hear about us?"

That single data point, captured consistently across a few months of deals, will tell you more about where your pipeline is really coming from than most attribution models will. It's imperfect (self-reported data always is) but it captures the parts of the journey that tracking misses.

From there, you can build a more complete picture over time: influenced pipeline by channel, conversion rates by source, content engagement patterns in closed deals. Attribution is never finished. But you can make progressively better decisions with progressively better data.