Marketing Attribution Explained (and Why the Numbers Never Add Up)

Marketing attribution is the attempt to assign credit for a sale to the marketing touches that led to it — and the numbers never add up because every platform measures a different, overlapping slice of the same customer journey. Your ad platform, your email tool, and your CRM will each claim more credit than the others, and none of them are wrong. They're just incomplete.

What Attribution Is Actually Trying to Answer

At its core, attribution asks: "of everything this prospect saw or did before buying, which touches deserve credit, and how much?" That's a reasonable question. The problem is that a real buyer's journey — an ad seen but ignored, a colleague's recommendation, three blog posts read over two months, a direct visit that finally converts — doesn't come with a ledger. Attribution models are approximations layered on top of incomplete data.

The Four Common Models, and What Each One Assumes

• First-touch: 100% of the credit goes to whatever brought the person into your world first. Good for measuring top-of-funnel awareness spend, bad for anything that happens after.
• Last-touch: 100% of the credit goes to whatever happened right before conversion. Good for measuring closing-motion spend, bad for anything earlier in the journey — including the channel that actually created the interest.
• Linear: Credit is split evenly across every touch. Fair in theory, but treats a passive ad impression the same as a 30-minute demo call.
• Multi-touch (weighted): Credit is split unevenly based on a rule set you define. More realistic, and also the most arguable — the weights are a judgment call, not a measured fact.

Why the Platforms Disagree With Each Other

Your ad platform tracks its own click and view windows. Your CRM tracks whatever a rep logs, often manually and inconsistently. Your analytics tool uses its own session and cookie logic, some of which browsers now actively restrict. Each system is honestly reporting what it can see — none of them can see the whole path, so summing their numbers will always overcount the true total.

What to Do Instead of Chasing a Perfect Model

1. Pick one attribution model and apply it consistently across every channel and every report.
2. Treat the output as a weighting for where to invest more, not a court-admissible ledger.
3. Sanity-check the model against a few known deals — does the story it tells match what your sales team remembers actually happening?
4. Revisit the model only when your sales motion changes meaningfully, not every time a number looks off.

When "Good Enough" Attribution Is Actually Good Enough

For a lean team, the bar isn't a perfect model — it's a consistent one you trust enough to shift budget based on it. If switching $2,000 from one channel to another based on your attribution read would make you meaningfully nervous, that's a sign to sanity-check the model, not to keep adding complexity to it.

Pairs well with Pipeline Commander for tracking what happens to a lead after marketing hands it off, and the AI Overview Traffic-Loss Diagnostic if a chunk of your first-touch credit is coming from organic search that may be quietly eroding.