I keep a file of the changes I ignored
The competitor changes you dismiss are worth more as a record than the ones you acted on.
Dismissed competitor changes look like noise one at a time. Kept in a dated list, they become the earliest timestamp you have on a real strategy shift.
Eleven months ago a competitor added one word to their footer: a link labelled "Partners." I looked at it, decided it was housekeeping, and moved on. I also wrote it down — date, URL, one sentence on why I was skipping it. Last quarter their entire go-to-market turned out to run through resellers, and that dismissed footer link was the earliest timestamp anyone on our side had.
The practice isn't a system. It's a text file with dates in it.
A dismissal is a judgement call, and judgement calls should leave a trace
Competitive intelligence work is judged on the changes people acted on. The pricing move that reshaped a deal. The feature that showed up in a lost-deal note. Nobody audits the other pile — the changes that got a glance and a shrug — because at the moment of the shrug they genuinely did look like nothing.
They usually were nothing. That's not the point. The point is that the shrug contains information you will want later: what you saw, when you saw it, and what you believed at the time. Six months on, the question is never "did this matter?" It's "how long has this been true?" A dated line of your own reasoning answers that in a second. Memory doesn't — you can't diff against your memory, and that goes for your own past opinions as much as for their pricing page.
Individually noise, collectively a direction
Three dismissed changes on their own are three pieces of housekeeping. Three dismissed changes that all point the same way — a compliance page, a new SSO mention on the sign-in screen, a footer link to partners — are a company moving upmarket in slow motion.
You only see that if the dismissals sit next to each other. Scattered across Slack threads and half-remembered glances, they never form a shape. In one list, sorted by date, the shape is obvious and often embarrassing: the story was legible months before anyone told it.
This is the honest split in the work. Seeto watches public surfaces continuously and hands you each change as a discrete, timestamped event, so nothing depends on whether you happened to open the tab that week. It does not decide what matters — that call is yours, and the file is where you keep the calls you made. Deciding a change is safely ignorable is real analysis. So is ruling out a false positive. Both deserve a line.
What it costs
Two minutes a week, and no tooling beyond somewhere to type. Date, surface, what changed, one clause on why you passed. No tags, no categories, no scoring — the moment it needs maintenance it stops getting written.
Review it once a quarter. Most of it will still be nothing. Once or twice a year, three lines will suddenly read as one sentence, and you'll know exactly when the sentence started.