Change detection

You cannot watch a page you did not know they built.

Page watchers are pointed at URLs by hand. That works right up until the competitor launches something you had no reason to expect. We track the shape of the whole site instead, so new pages surface themselves.

Watch a competitor
Baseline, 2026-08-24

9,806 URLs tracked on one site.

Every one of them watched for appearing and disappearing, without anyone naming them first.

What the site is made of

4,760
pages
3,486
blogs
900
products
644
collections
16
other
A jump in one of these columns is a strategy change you can see before the announcement.

Three kinds of change, one weekly note

What movesWhat it usually means
Pages appear or disappearA launch, a campaign, or a quiet retreat from a product line
Prices moveEvery tracked product, so you see the direction before the press release
The tech stack changesSwapping review app, email platform or support desk says where they are investing

Read this before you buy

The first run is a baseline, not a report. There is nothing to compare against until the second one. Any change-detection tool that shows you something interesting on day one is showing you the present, not a change.

We diff the set of pages and the values on the ones we parse. We do not take visual snapshots, so a redesign that keeps the same URLs and prices will not register as a change here.

Watch a competitor

Questions we get

How is this different from a page watcher?
A page watcher tells you when a URL you already thought of changes. We track the shape of the whole site, so new pages find you: a new collection, a new landing page, a price that moved, a legal page that appeared.
What counts as a change worth telling me about?
One that a competitor decided to make. We normalise away session ids, rotating banners and timestamps before comparing, because a diff that fires every week on noise is a diff nobody reads.
How far back does the history go?
From the day we start. Snapshots are never deleted, so the archive is the asset: the value of week one is small and the value of month six is large, which is the only argument for starting now rather than later.

Related

Price architecture, worked exampleSee every ad they runCompetitor tracking for SaaS
HomePrivacyContactExample data from a public brand, refreshed weekly. Last run 2026-08-24.