Signal feed toggles, tracking changes, budget shifts, site migrations – aka the things most likely to move a client’s results – happen constantly, but are scattered across TeamworkFX and Gmail with no consistent record of when. When RCFX shows a lift or a dip two weeks later, there’s no easy or quick way to check what actually changed and tie it back to these actions.
Manually annotating every relevant change, for every client, every week, doesn’t scale. So most of it never gets logged, and most results shifts go unexplained.
Developed a Claude Cowork task that runs weekly to scan TeamworkFX and Gmail across my full client roster, classifies what it finds into four result-impacting categories, and matches each finding to the right client by a site ID-based roster lookup.
High-confidence findings get written directly into that client’s RCFX profile as a private, dated annotation. Everything else lands in a Slack digest for a five-minute manual review.
All annotations are logged in the Slack thread, so there is a running, sourced paper trail connecting initiatives to results (with citations of where Claude found these insights).
The win isn’t just the automation; it’s that every annotation is now a testable claim. Two weeks after a signal feed goes live, I’m not guessing whether it mattered. I can point to the exact date it changed and read the results against it.