Skip to content
Join the waitlist

QueryRev Terms

Approval Workflow

An approval workflow is the fixed path a proposed change follows before it goes live: who may propose it, who must approve it and in what order, and what is recorded. In SEO it decides which edits, such as internal links, may be applied without a person, and which, such as redirects or noindex, wait for a named approver.

  • Industry term

By QueryRev ResearchPublished Updated6 min read

Why it matters.

Approval is an old discipline of change control. NIST’s guide to security-focused configuration management describes configuration change control as the documented process for managing and controlling changes to a system. It involves the systematic proposal, justification, implementation, test and evaluation, review and disposition of changes, with each step clearly articulated along with the responsibilities and authorities of the roles involved[1].

The path has steps. A change is requested, recorded, analysed, tested and approved, usually by a configuration control board; authorised staff then make it, someone verifies it, and the request is closed. NIST describes the board as a group of typically two or more individuals with the collective responsibility and authority to review and approve changes. It says the process is not complete, and the request is not closed, until it has been confirmed that the change was deployed without issues[1].

Not every change needs the board. NIST says an organization may designate types of change that are preapproved and not sent to the board, and gives vendor-provided security patches and updated antivirus signatures as examples[1]. The rule is written in advance, by kind of change. It is how an SEO team can let internal links through and hold redirects.

Tools carry the same rule. GitHub lets a repository require that pull requests receive a specific number of approving reviews before they are merged into a protected branch, and that the most recent push be approved by someone other than the person who pushed it[2]. WordPress builds it into roles: an Editor can publish and manage posts, including those of other users, while a Contributor can write and manage their own posts but cannot publish them[3].

A rule is sometimes written by someone other than the site’s editors. In QueryRev’s study, 22 of the 5,637 robots.txt files the crawler read carry the comment “BEGIN Cloudflare Managed content”, which opens the rules Cloudflare writes when a site turns its managed robots.txt on[4]. Who may change such a file, and how the change is recorded, is a question for the workflow.

How it is measured.

Count the changes by route: applied without a person, waiting for one approval, waiting for two, and held. A workflow that sends every change to a person is a queue, and one that sends none is not a workflow.

Measure the wait. The time from proposal to decision, by kind of change, shows where the path is too narrow, such as a long wait on a harmless edit, and where it is too loose, such as a risky edit decided in seconds.

Count the changes that bypassed the path: edits found on the site with no record of a proposal or an approval. NIST lists requirements for retroactive analysis, testing and approval of changes made outside the change control process among the things its procedures should cover[1].

Measured by

  • Changes by route
  • Time waiting for approval
  • Changes made outside the path

A worked example.

Illustrative: an invented site, to show the method.

A coffee-equipment shop has written down what may go live by itself. On Monday five changes are proposed. Nine internal links to /burr-grinder and thirty meta descriptions fall in the kinds the policy allows, so they are applied at once and recorded.

A new title on /burr-grinder and a redirect from /old-grinders to /grinders are kinds that need the SEO Lead, so they wait. A noindex on /espresso-machines is proposed for a protected page, where it needs two approvals, so it is held. Two applied, two waiting, one held, and nothing reached the site outside the path.

Five changes, by policy

What to do, step by step.

  1. List the kinds of change your site gets: internal links, meta descriptions, titles, redirects, noindex, template edits.

  2. Sort them by what a mistake would cost and how easily it can be undone.

  3. Decide which kinds may be applied without a person, which need one approval and which need two. Name roles, not people.

  4. Mark the protected pages, the ones that earn the most, where every change needs approval.

  5. Record each change when it is proposed and when it is applied: who, when, what, and the page before and after.

  6. Check the path after a month: which kinds wait too long, and which slipped through.

Mistakes to avoid.

  • Approving everything by hand. A long queue of low-risk edits buries the few that need a decision. NIST allows types of change to be preapproved so that the board looks at the rest[1].

  • Approving nothing. A path with no approval for hard-to-undo changes, such as a redirect, a noindex or a robots.txt edit, relies on nobody making a mistake.

  • Letting the proposer approve. GitHub can require approval of the latest push by someone other than the person who pushed it[2]. The person who makes a change is the worst placed to check it.

  • Naming people instead of roles. WordPress attaches capabilities to roles such as Editor and Contributor[3]. A path written around a person breaks when the person leaves.

  • Stopping at approved. NIST says change control is not complete until the change is confirmed to have been deployed without issues[1]. An approved change may go live, and it is not a result.

How QueryRev treats it.

In QueryRev, every proposed change is checked against the rules you set: what may be applied automatically, what needs approval, and which pages are protected (governed execution). Nothing changes a customer’s website outside that policy. The assistant can explain a change and propose one, but it never approves or applies one. Changes reach a site through a direct integration or a pull request, and where there is no integration the change becomes a task for a person, recorded and verified in the same way. Each change that goes live is written to the Change Ledger with its before and after. A typed decision carries its evidence into the queue, a person or the policy decides, and the outcome is verified. See the guide to automated SEO reports.

Questions people ask.

What is an approval workflow in SEO?

It is the agreed path a proposed site change follows before it goes live. It says who may propose a change, who must approve it, in what order, and what is recorded. It lets low-risk edits go through at once and holds the risky ones for a named person.

Which SEO changes should need approval?

Those that are hard to undo or that can take pages out of search: redirects, noindex, canonical changes, robots.txt edits, and titles or restructures on the pages that earn the most. Internal links and meta descriptions are common candidates for being applied without a person.

Can an approval workflow be automated?

The routing can be: a policy sends each kind of change where it belongs, and applies the allowed ones at once. The approval itself stays with a named person for the kinds that need it, and every step is recorded so that the path can be checked afterwards.

Sources and citation.

Primary sources only: the search engines, the AI vendors, standards bodies and QueryRev’s own research. Each is cited in the text by its number.

  1. Guide for security-focused configuration management of information systems (NIST SP 800-128)NIST ·

  2. About protected branchesGitHub Docs ·

  3. Roles and capabilitiesWordPress.org ·

  4. State of AI crawler access, October 2026QueryRev Research ·

Cite this definition

A suggested citation. Its link goes to the definition, whose anchor does not change.

QueryRev Research. “Approval Workflow.” QueryRev AI glossary, 25 October 2026. https://queryrev.com/glossary/approval-workflow#definition

Related terms