Skip to content

Documentation Website impact review

Choose a guide from the library.

Website impact review

Compare a ready release in your website before deciding to publish. The website owns its renderer; no framework is required by the CMS.

What a collection report covers

A website impact report freezes published collection content and the proposed result of a ready release. Your website renders these snapshots. The proposed side uses the release engine inside a rolled-back transaction: capturing does not publish or send webhooks. REST, GraphQL, RPC and client reads work without registering any website.

How to use it

  1. Prepare a ready release containing the exact saved entry, translation and media actions you intend to evaluate.
  2. Choose a configured destination and one language. Inspect its collection and ID-based entry routes, including changes caused by references or media.
  3. Use your website's comparison renderer to inspect both snapshots at desktop and mobile sizes. The Astro journal is an optional reference implementation.

Good practice

  • Read coverage notes and unmapped-item warnings before treating the route list as complete.

Avoid

  • Coverage is limited to 25 published entries, a six-entry first collection page and four latest items, with an optional descending sort field. Slug-based routes, additional queries/pages, interactions, live remote values and automated screenshot analysis need separate checks.

Configure your website renderer

This is an optional integration. The Owner chooses a content type and language, relative collection and entry routes, and a renderer SHA-256 revision. Any server framework can implement the preview exchange and comparison payload contract; no website source code is uploaded to the CMS.

How to use it

  1. Implement the server-side preview exchange and frozen comparison renderer on your website; keep its environment credential private. Ordinary SDK draft reads do not require this integration.
  2. In Workspace settings → Website preview, choose any content type. Enable website impact review and enter paths such as /products and /products/{id}. Only {id} substitution is supported.
  3. Optionally choose a sortable field for the four latest items. Leave it blank to use the normal collection order. No title, date or body fields are required.
  4. Hash your rendering inputs with SHA-256; configure that revision in the destination and have the website verify it. Update both and capture a new report whenever the renderer changes.

Good practice

  • Use the exact code running in the consuming website, not an unrelated branch's hash. Hash inputs exclude secrets and build output.

Avoid

  • A mismatched renderer fails closed. The Owner's configured hash is a declared renderer identity, not independent proof that an external deployment is trustworthy.

Capture and review before/after routes

Website review is separate from publishing readiness and entry approval. It records a human's checks, not an automated judgment of visual quality.

How to use it

  1. Open a ready release, choose the destination and review language, and select Capture website comparison. An exact-language destination overrides its all-language default.
  2. Open Before & after. The site starts a private session lasting up to ten minutes; secrets stay in its HttpOnly cookie/server configuration.
  3. Choose each affected route, select Desktop and Mobile in turn, and compare both sides. Use actual size for fine details; fit mode scales the fixed viewport frames to your screen.
  4. Return to the studio, check Reviewed desktop and Reviewed mobile for each route, acknowledge the coverage limits and Record website review.
  5. Decide separately whether to publish or schedule the release through its normal controls.

Good practice

  • Check missing pages, text wrapping, image crops, references and language. Reopen a fresh session if cookies are blocked or it expires.

Avoid

  • Do not check a viewport you did not inspect or claim the comparison tested navigation; links inside frozen views are intentionally disabled.

Keep revision-bound receipts meaningful

Reports retain immutable before/after content, routes, source/model versions, release version, language and renderer identity. Human review stores its actor and timestamp separately.

How to use it

  1. If source content, model/policy, destination settings or the release changes, read the Stale indicator and capture a new report.
  2. The private session rechecks current actor permissions, application-token scope and source freshness on every read; revoked access cannot keep reading a frozen report.
  3. Keep up to three reports per release and fifty per environment. Remove unneeded reports to free capacity; their preview sessions end.

Good practice

  • Keep historical receipts as evidence of the specific revisions reviewed, and repeat review after a change.

Avoid

  • A receipt is not a screenshot archive, a production monitoring service or authority to bypass approvals. Publishing a release makes its pre-publication comparison historical.
Need a hand? Check troubleshooting. Documentation describes the locally verified pilot; production qualification remains in progress.