Attribution Is Astrology for People With Spreadsheets

Pull up your attribution report. Look at the number next to paid search. Now switch the model from last-touch to first-touch. Watch the number change. Now switch it to linear, then time-decay, then whatever your platform calls its proprietary blend. Watch it change three more times.

Same campaigns. Same spend. Same customers. Five different stories about who deserves the credit. And here is the part nobody says out loud in the QBR: you already know which model you’re going to present. You picked it before you opened the report. You picked the one where the channel you fought for last quarter looks like a genius move.

That isn’t analytics. That’s reading tea leaves and calling it data.

The dirty secret of every attribution model

An attribution model is not a measurement of reality. It is a rule you invented for splitting credit across touchpoints, and every rule is wrong in a different direction.

Last-touch says the final click did all the work, which means your branded search and your retargeting always look like heroes and your top-of-funnel content looks like a waste. First-touch says the opposite. Linear pretends every touch mattered equally, which no human being actually believes. Time-decay assumes recent touches matter more, because someone decided that sounded reasonable on a Tuesday.

None of these models know why the customer bought. They can’t. They only see clicks they were allowed to track, on the channels that bothered to pass a parameter, on the devices the cookie survived. The model is doing arithmetic on a partial, broken, deduplicated guess and handing you a confident pie chart.

The confidence of the pie chart is inversely related to how much it actually knows.

Why this keeps happening

Here is the mechanical reason attribution stays broken, and it has nothing to do with the models themselves.

Your customer’s actual journey lives in seven places. They saw a LinkedIn post (data in one tool). Clicked a Google ad two weeks later (data in another). Read three blog posts (web analytics). Got a nurture email (the email platform). Talked to sales (the CRM). Bought through a self-serve checkout (Stripe or Shopify). Renewed (somewhere else entirely).

No single tool sees that whole path. So each tool reconstructs the journey from the slice it can see, and each one quietly claims credit for the conversion because that’s what it was built to do. Your ad platform thinks it drove the sale. Your email tool thinks it drove the sale. Your CRM thinks it drove the sale. Add up the credit every tool reports and you’ve sourced 280% of your revenue. Congratulations, you’re a unicorn in a slide deck and broke in the bank account.

The models aren’t lying because the math is bad. They’re lying because the inputs are fragments, and you can’t reassemble a person’s decision from fragments that were never stitched together in the first place.

What the spreadsheet jockeys do next

The usual fix is to throw a human at it. Someone exports six CSVs, VLOOKUPs them against each other at 11pm, manually dedupes the contacts that exist in four systems with three different email addresses, and builds a “unified” attribution view in a spreadsheet that takes four hours to refresh and breaks the moment anyone touches a column.

That spreadsheet becomes the source of truth. One person understands it. They go on vacation and the company forgets how it makes money. This is not a hypothetical. You’ve met this person. You might be this person.

And even after all that labor, the output is still a model. Still a rule someone invented. Still astrology, just with more steps and a tired analyst.

Stop asking which channel gets credit. Ask a better question.

The honest move is to admit attribution will never be a clean number and stop pretending it is. Credit-splitting is the wrong obsession. What you actually want to know is more concrete and harder to fake:

• Which customers who bought this quarter touched which things, in what order, and how long it took?
• When we paused that channel for three weeks, did pipeline actually drop, or did nothing happen?
• What does the path look like for the deals that closed big versus the ones that churned in ninety days?

Those questions have real answers, but only if the data lives in one place. You can’t trace a journey across tools that have never spoken to each other. You can’t run a clean before-and-after when every system counts conversions differently. The bottleneck was never the model. It was that nobody could see the whole board at once.

One source of truth or keep guessing

When every touchpoint, every channel, every order, and every CRM record sits in one system, attribution stops being a religious argument and becomes something closer to a question you can interrogate. Not “which model do we believe,” but “show me the actual path these forty customers took.” You stop arguing about credit and start looking at behavior.

You’ll still never get a perfect number. Nobody will. But there’s a real difference between an honest “here’s what we can see across the full journey” and a confident pie chart built on seven tools each claiming the win. One is measurement with known gaps. The other is a horoscope you spent six figures to generate.

If you’re tired of presenting whichever attribution model flatters last quarter’s decision, the fix isn’t a smarter model. It’s getting your whole stack into one place so you can ask real questions and get answers that don’t change every time you click a dropdown. That’s the entire idea behind THE DASHBOARD. Not a better way to guess. A way to stop guessing about things you could actually just see.