Marketing Attribution

Here's the thing: the average B2B buyer touches 36 different marketing assets before making a purchase decision. If you can't figure out which of those 36 touchpoints mattered, you're spending based on gut feel — and gut feel gets budgets cut.

Without attribution, marketing teams default to measuring what's easy (impressions, clicks, MQL volume) instead of what matters (pipeline, revenue, CAC). You end up doubling down on the LinkedIn campaign that generates lots of leads but zero closed deals, while underfunding the blog content that quietly drives 30% of your pipeline.

Good attribution doesn't just protect your budget. It changes how you allocate it. Teams with mature attribution models reallocate an average of 25-30% of their spend after seeing what actually drives revenue — and see pipeline increase by 15-25% without spending a dollar more.

Attribution models fall into two camps: rules-based and data-driven.

1. First-touch attribution. 100% credit to the first interaction. Good for understanding what fills the top of funnel. Terrible for everything else.
2. Last-touch attribution. 100% credit to the final touchpoint before conversion. The default in most CRMs. Easy to implement, but it ignores everything that built awareness and trust before that last click.
3. Multi-touch attribution. Distributes credit across multiple touchpoints. Linear (equal credit), time-decay (more credit to recent touches), U-shaped (40% first, 40% last, 20% middle), or W-shaped (33% each to first touch, lead creation, and opportunity creation).
4. Data-driven attribution. Uses machine learning to analyze all touchpoints and assign credit based on actual statistical impact on conversion. The most accurate, but requires significant data volume to be reliable — typically 300+ conversions per month.

The practical move for most B2B teams: start with multi-touch (W-shaped or time-decay), complement it with self-reported attribution ("How did you hear about us?"), and combine both views to make decisions. Neither source is perfect alone. Together, they're pretty good.

A B2B martech company was spending $120K/month across Google Ads, LinkedIn, content marketing, and events. Their last-touch attribution showed Google Ads driving 60% of pipeline. So they kept increasing Google budget while cutting content spend.

Then they switched to multi-touch attribution. What they found: Google Ads was mostly capturing demand that content marketing had created. Blog posts were the first touch in 45% of closed-won deals. Prospects read 3-4 articles, then Googled the brand name, clicked the paid ad, and submitted a demo form. Last-touch gave Google all the credit. Multi-touch showed the real story.

They rebalanced: shifted 30% of Google Ads budget back into content and SEO, added an AI chat agent (Salespeak) to engage blog readers showing buying signals. Six months later, total pipeline was up 22% on the same overall budget.

• Treating last-touch as gospel. Last-touch attribution in B2B is like crediting the cashier for the entire restaurant experience. It captures the final action, not the journey that led there.
• Ignoring dark social and untrackable channels. Podcasts, word-of-mouth, Slack communities, DMs — you can't cookie these. If 40% of your "direct" traffic is really from a LinkedIn post someone shared in a private group, your attribution is lying to you.
• Over-engineering the model. A perfect attribution model doesn't exist. Spending 6 months building a custom data-driven model when you have 50 conversions/month is wasted effort. Start simple, iterate.
• Not asking prospects directly. Add "How did you hear about us?" to your demo form. Self-reported attribution catches what software can't — and it's shockingly accurate when the field is open-text instead of a dropdown.