Analytics
Attribution Modeling: What Actually Works for B2B
Multi-touch attribution has been promised as the solution to marketing measurement for a decade. Here's what it actually delivers — and where it falls short.
The marketing attribution conversation has been happening for 15 years. The technology has improved significantly. The accuracy of the models has improved modestly. The organizational behavior change required to act on attribution data has improved almost not at all.
The Technical Reality
Multi-touch attribution in B2B has three structural constraints that no technology solves: long sales cycles with multiple decision-makers, offline touchpoints that don't get tracked, and the fundamental impossibility of observing counterfactuals.
A model that assigns credit across 15 touchpoints over 9 months is making educated inferences, not measurements. The model's output is directionally useful. It is not a precise accounting of cause and effect.
The goal of attribution isn't to solve a math problem. It's to build enough organizational consensus about where marketing is creating value to make confident budget decisions. A directionally accurate model that gets acted on is more valuable than a statistically rigorous model that doesn't.
What Actually Helps
- Start with first-touch and last-touch as baselines — they're wrong, but they're consistently wrong
- Layer in time-decay for long sales cycles — recent touchpoints should get more credit
- Run cohort analysis on content assets — which assets appear most frequently in closed-won journeys?
- Conduct win/loss interviews — human testimony is still the most reliable source for intent signals
- Build a consistent UTM taxonomy before worrying about the attribution model
The Organizational Question
Attribution modeling is ultimately an organizational alignment problem. Sales and marketing need to agree on what a qualified lead is, what data gets captured, and how credit gets assigned. Without that alignment, the model produces data that each side interprets in their favor.
Start with the organizational conversation. Then build the model that supports the agreed-upon framework. Doing it in the other order produces technically sophisticated reports that get used as ammunition rather than tools.
Work with Berchtold
If this is the gap you’re living with, let’s talk.
We work with businesses that have real marketing problems — not ones looking for a vendor to fill a slot. A consultation costs nothing. Clarity does.