How do I track website changes over time?

Why a point-in-time audit is not enough

A single audit tells you the current state. It cannot tell you whether that state is new, and the second question is usually the operational one.

• Was that header always missing, or did it disappear on Tuesday?
• Did our accessibility score drop after the redesign, or before it?
• When did the certificate change issuer?
• Has the third-party script count been creeping up all quarter?

Each of these is trivial with history and unanswerable without it. Reconstructing the past from memory and deploy logs is much harder than it sounds, particularly for changes nobody deployed.

Reading the stored history

curl -s "https://outrings.com/api/v1/history?url=example.com" | jq '{
  snapshotHost,
  points: [.history[] | {at, score, grade}]
}'

This endpoint reads history only and never fetches the target, so it still works while the site is down — which is exactly when you most want to know what changed.

Comparing two points

curl -s "https://outrings.com/api/v1/changes?url=example.com" | jq '{
  comparable,
  scoreDelta,
  categories: [.categories[]? | select(.delta != 0) | {category, from, to, delta}],
  appeared: [.newProblems[]?.id],
  cleared:  [.resolved[]?.id]
}'

This re-audits, compares against the most recent stored snapshot, and reports the difference. comparable: false means there was nothing earlier to compare with — history starting, not an error.

What to keep an eye on longitudinally

The first row is the one that catches most teams. No single tag addition is significant, and twenty of them over a year is a different website.

A sensible cadence

1. Snapshot daily, or on every production deploy if you ship less often than that.
2. Alert on new high and critical findings, not on score movement.
3. Review the trend monthly rather than the daily numbers — that is the timescale at which drift becomes visible.
4. Before concluding anything from a comparison, check the ruleset version matches.

Snapshots are stored at most once per site every six hours, so a daily job is a natural fit and more frequent runs will not produce more history.