Keep a dated visual record of pages that will not stay the same

Archiving a web page sounds like a solved problem until you try to rely on one of your archives. The question is never "do we have a copy" — it is "does this copy prove what we think it proves". Three things decide that: completeness, consistency and provenance.

Completeness: capture the page, not the top of it

A viewport screenshot records the first screen. For an archive that is close to useless: the pricing table, the small print, the disclaimer and the footer contact details are all below the fold. Full-page capture scrolls the whole document and stitches it, so the archived image contains everything the page contained.

Completeness also means the page as rendered. Pages assemble themselves from many sources — fonts, images, scripts that inject content after load. Capturing in a real browser engine, with time allowed for the page to settle, is what makes the image resemble what a visitor saw rather than a skeleton of what the server first sent.

Consistency: the same page, the same way, every time

An archive earns its keep when you compare two captures and see what changed. That only works if everything except the page itself was held constant. Capture the same URL at the same viewport width, with the same device profile, the same banner handling and the same capture location, and a visual difference means the page changed. Vary any of those and you cannot tell.

This is the strongest argument for scheduling over manual capture. A person will use whatever browser window they happen to have open. A schedule uses the profile you configured, every single time, for years.

Provenance: when it was taken, and by what

An image on its own is weak evidence. What gives an archive weight is the surrounding record: the exact capture time, the settings used, and an unbroken series showing the page was captured on a regular cadence rather than cherry-picked after the fact. Automated captures produce that naturally — the timestamps are generated by the system, and the gaps, if any, are visible.

Where the files live matters too. Captures delivered straight into your own cloud storage inherit that platform's version history and retention controls, and survive independently of your subscription with us.

Deciding what to archive, and how often

The instinct is to archive everything. Resist it: a hundred schedules nobody reviews costs allowance and produces noise. In practice most teams archive three categories of page.

• Pages you are accountable for. Your own terms, privacy policy, pricing and product claims. Weekly is usually enough, plus a capture after each release.
• Pages you depend on. Supplier terms, API documentation, partner pricing — anything you have built a commitment on top of. Weekly or monthly.
• Pages that are evidence. A competitor's claim, a marketplace listing, a published statement. Daily while it is live, because these are the pages that vanish.

If your archive exists to satisfy a regulator rather than to settle an argument, the requirements are stricter — see compliance evidence for what changes. If you are archiving other companies' pages to watch how they move, that is competitor tracking, and the frequency and geography settings matter more than completeness does.

Getting started without over-engineering it

Archive one page first. Choose the page you would most regret not being able to prove — usually your own terms, or a competitor claim you are already arguing about — and put it on a daily full-page schedule delivered to a cloud folder. Look at it after a week. The right frequency and the right set of pages become obvious once you have something real to look at, and you can widen from there.