The problem
We were spending on marketing without a reliable picture of which channels were bringing in signups. Gaps in the tracking made it hard to tell what was working.We were spending on marketing without a reliable picture of which channels were bringing in signups. Gaps in the tracking made it hard to tell what was working.
The people who built the setup were long gone, and the documentation that remained did not explain why anything worked the way it did.The people who built the setup were long gone, and the documentation that remained did not explain why anything worked the way it did.
Someone had to fix it. I wasn't a marketing analyst, and I wasn't an engineer. But I knew the business, and I knew I could learn the rest.Someone had to fix it. I wasn't a marketing analyst, and I wasn't an engineer. But I knew the business, and I knew I could learn the rest.
The approach
So I didn't try to become an attribution expert overnight. I built a brain that already was one:So I didn't try to become an attribution expert overnight. I built a brain that already was one:
everything in my head about the business, what was left of the documentation, and 47 expert sources on attribution, all in one place my agents could work from.everything in my head about the business, what was left of the documentation, and 47 expert sources on attribution, all in one place my agents could work from.
It was my first Team Brain, before I had a name for it. Then I worked with the agents at every step, from choosing the approach to building the attribution models.It was my first Team Brain, before I had a name for it. Then I worked with the agents at every step, from choosing the approach to building the attribution models.
I wasn't outsourcing the thinking. I was directing it: checking every decision, correcting every wrong turn.I wasn't outsourcing the thinking. I was directing it: checking every decision, correcting every wrong turn.
The result
I built an attribution model that covered about 95% of signups, combining tracked data with survey-informed estimates. Finance, marketing, and leadership could now see what each channel cost and what it brought back.I built an attribution model that covered about 95% of signups, combining tracked data with survey-informed estimates. Finance, marketing, and leadership could now see what each channel cost and what it brought back.
That's when it clicked. I was no longer limited to what I already knew how to do.That's when it clicked. I was no longer limited to what I already knew how to do.