The manual spreadsheet finally got a clock.
In October, I had spent most of a day comparing Mixpanel and Intercom because their product-usage numbers did not match. Mixpanel was the source I trusted more, but its exports were difficult to work with. The immediate answer had been manual account checks and a CSV from the easier system.
That was workable, not good.
The next step was a scheduled Mixpanel report. It would generate the CSV automatically and send it to the people using the information every Sunday at noon. I kept myself on the recipient list for a while so I could see what arrived and make sure it stayed useful.
The note announcing it was only one line: the automated report should be fixed up.
The small word “should” mattered. Scheduling a report did not settle every question about how events were recorded or whether the two analytics systems would agree. It removed the repeated export work and gave the team the same file on a predictable schedule.
That was enough progress for this stage. Client success could use the CSV in its scoring work. I could keep checking the underlying Mixpanel data when something looked wrong. Nobody had to remember to run the same job by hand before the week started.
The report was not a new data platform. It was a weekly task converted into code and a calendar entry.
Sunday at noon, the file would go out. For a little while, it would come to me too.