The problem multi-touch attribution tries to solve
The buyer read your article in January. In February they saw a LinkedIn ad. In March they came to a webinar. In April their colleague mentioned you at a meeting. In May they Googled your company name and visited directly. In June they signed the contract.
Last-click attribution says: the deal came from direct traffic. First-click says: organic. Both answers are wrong, because neither channel closed the deal alone.
Multi-touch attribution tries to distribute the value of the deal across all touchpoints. That’s more honest, but harder - and it requires understanding what each model actually measures.
The four main models
Linear. Each touchpoint gets equal weight. If there were five interactions, each gets 20% of the deal value. Simple, transparent, requires no training data. The problem: it assumes all touchpoints are equally important, which is rarely the case.
Time-decay. The closer to the deal close, the more weight the touchpoint gets. The first contact six months before purchase receives almost nothing, the final one receives the maximum. The logic: what happened recently matters more for the decision. For B2B this is debatable: often it’s the first piece of content that builds trust and interest, while the last touchpoints are just reminders.
Position-based (U-shaped). First and last touchpoints each get 40%, the rest share 20%. A compromise: we acknowledge the importance of the brand introduction and the final touch, without ignoring the middle. Popular and intuitively understandable, but the percentages are arbitrary.
Data-driven. The model trains on your historical data and determines weights based on which combinations of touchpoints actually led to conversions. Technically the most honest - but requires data volume (Google recommends at least 400 conversions per month) and doesn’t work as a black box.
What actually works for a EU B2B team
Straight answer: none of these models will give you the truth. Here’s why.
Most B2B companies in the EU with fifteen to fifty-person teams don’t have enough data for a data-driven model. 400 conversions per month is a lot for a company with ACV above €30k. With five to ten deals per month, the algorithm won’t train.
All models only work with tracked touchpoints. The dark funnel (see separate article) - recommendations, community, podcasts - falls outside any attribution model. In B2B this can be forty to sixty percent of the real journey.
GDPR in the EU limits cross-session tracking without consent. If a user came in one session without cookies and in another with them, cross-device and cross-session attribution breaks.
A practical approach instead of one model
For a B2B team in the EU, what works is not one “correct” model but a combination of data sources.
Combine GA4 with CRM data. GA4 shows online paths. CRM shows real deals with manual source notes. The gap between them is part of the dark funnel.
Use self-reported attribution. The “How did you hear about us?” question in the form or on the onboarding call. Captures data that no technical model can.
Look at assisted conversions, not just last-click. In GA4 - the “Attribution” -> “Multi-channel funnels” report. This shows which channels participated in the path to conversion even if they weren’t the last touch. Organic and content often look “useless” in last-click but are key in assisted conversions.
For budget decisions - linear as the default. If you don’t have data for data-driven and no time to build a complex system, linear attribution is an honest compromise. It doesn’t overweight any one channel and gives at least a rough picture of distribution.
When it’s worth investing in attribution
Making attribution more sophisticated makes sense when you’re spending significant money across several channels and don’t know which one to turn off. If you have one main channel or a budget below €10k per month, complex attribution won’t give answers that will change decisions.
Before building a sophisticated model, answer the question: “What decision will change if I learn the truth about attribution?” If the answer is specific (for example, “I’ll turn off LinkedIn Ads or double the content budget”) - attribution is needed. If the answer is vague - first define the question.
What to implement right now
The minimum set for a B2B team without a data analyst:
UTM tags on all paid channels and email campaigns - basic hygiene, nothing works without them.
A “How did you hear about us?” field in the inquiry form - gives self-reported data immediately.
The assisted conversions report in GA4 once a month - takes twenty minutes and shows which channels participate in the paths to conversion, even if they’re not the last touch.
This is enough to make better decisions than relying solely on last-click.