This is the least visual project I have and one of the ones I’m proudest of. So bear with me.

When you’ve been a public company since 2012, your databases get heavy. We had something like 900,000 contacts in the system: old creator signups, old product users, and plenty of people who had nothing to do with where the business was headed. We were paying for a lot of that.

So I went in with custom filters and a lot of analysis and cut it down to roughly 40,000 contacts that still had a real reason to be there. It saved real money, but more importantly, it gave us a database we could actually work with again. Segmentation got cleaner. Reporting got clearer. The whole system got easier to manage because the list finally reflected the business we were actually running.

Then I went after the thing underneath it, which was attribution. We had more than fifty lead sources spread across HubSpot and Salesforce. Most were inconsistent, more than half were unused, and none of them were trustworthy. I collapsed them down to about twelve clean sources that rolled up into a few real categories. Then I personally went back through two years of deals, every single one, and reassigned each to its true origin. It was painstaking and forensic and deep in the weeds, and I liked it.

The payoff was that we could finally answer a simple question honestly: where does our business actually come from? Google Ads, or gated content, or events? You can’t make smart spending decisions until you can answer that, and now we could.

I also built a workbook where you drop in a Salesforce export and it gives you back revenue by lead source automatically. I’m a little too proud of that spreadsheet.

It’s my pattern. Find the problem. Go deep. Leave the whole thing better than I found it.