I built a competitor watch for a client in one afternoon. Three years ago that was a retainer
A reading list, a summary email and a weekly digest — assembled from tools I already pay for. The interesting part was deciding what NOT to automate.
A client asked how hard it would be to keep an eye on four competitors: pricing pages, feature announcements, hiring posts. Three years ago that's a research retainer or a distracted intern. Last Tuesday it was one afternoon and tools I already had — a chatbot with browsing, a scheduling tool, and an email address nobody would miss.
The build was the easy part. Every morning, a scheduled prompt checks the four competitors' changelogs and pricing pages, diffs them against yesterday, and writes two sentences if anything changed. Everything lands in one digest email at 8am. Total code: zero. Total cost: within the subscription I already pay. The client's reaction — 'that's it?' — was the correct reaction.
What made it actually useful was deciding what not to automate. No sentiment analysis, no AI scoring of 'threat level,' no auto-posting to their Slack with alerts. Competitor watching fails from noise, not from silence. A quiet email that says 'nothing changed at three of four; Competitor B moved their enterprise pricing' beats a dashboard that screams daily.
The failure modes taught me the limits. One competitor redesigned their site and the diff logic reported every page as changed for two days — a false alarm flood I had to ride out. Another publishes pricing only in PDF, which the daily prompt handles badly. Automation is brittle exactly where the web is hostile to machines, and pricing pages are where companies deliberately blur.
The meta-lesson: the value wasn't technical. It was asking a client 'what would you actually do with this information daily' and building only that. Most competitor-watching wishes die because the answer is 'nothing, daily.' Weekly digest, two sentences per competitor, one email. Anything more is a toy that pretends to be work.