How to set service standards that define a credible source for SEO tools

What to take away

Google's guidance on reducing crawl rate has pushed server load into SEO service standards, not just hosting.
A usable standard is a written registeraccepted source, named owner, review date and the claims it supports.
Design and content guidance matter because an unindexable page makes every later metric unreliable.
A before and after comparison of the register shows which claims lost their source at the last review.
Contract renewal dates, not calendar quarters, should trigger the re-check.

Where service standards come from

Service standards for SEO tools are the written rules about which data a team will accept as evidence. They are deliberately dull. A standard built on taste cannot be audited. One that names a page, a date and a person can be checked in minutes.

From public rule to written standard

Cite public guidance, not personal preference
Write the rule into the register
Note the source and what it obliges
Copy vendor documentation into the register
Record version or date copied
Name who verifies the setting

Cite public rules, not preferences

Public guidance settles arguments that personal preference cannot. The design principles in the GOV.UK service manual cover usability and indexability, which are the floor for any ranking work. A site that fails there will not produce trustworthy position data. Treat the register as an internal policy informed by the GOV.UK Service Standard; that standard does not certify an SEO data source.

Write the rule into the register with its source, then note what it obliges you to do. If the principle says pages must be usable, the standard should say who tests that and how often.

Turn vendor documentation into a standard

Vendor documentation only becomes a standard when someone copies the relevant section into the register. If a supplier describes a crawl control feature, the standard should state what you do with it and who verifies the setting. Features nobody owns quietly stop being used.

Record the version or the date you copied, because documentation pages change without notice. The SEO tools operations and delivery guide covers how to sequence that work across a team, which helps once the rules exist but nobody has been given them.

Checking sources before they enter the register

Every new source gets the same treatment: read it, date it, and decide what it is allowed to prove.

Set up the register

First, list the SEO claims and decisions the team needs to support, such as search performance, crawl findings or content recommendations.
Next, choose a source that can support each claim and record its publisher, page title, URL, access date or version, and any relevant methodology.
Then, define the exact claim the source may support, its limitations, the owner and the review trigger.
Finally, have the owner and the analyst using the data sign off the entry. Remove or qualify any report claim when its source expires or no longer supports it.

Source intake checks

Read the source and date it
Decide what it is allowed to prove
List permitted crawlers and rate limits
Name who reads the server logs
Mark which tool outputs count as evidence
Treat scores as prompts, not findings

Crawl and server behaviour

Google Search Central's documentation page, “Reduce Googlebot crawl rate”, concerns Googlebot's requests to a site; it does not control third-party SEO crawlers. For a crawler such as Screaming Frog SEO Spider, set and record the crawl limits in that tool. The standard should list which crawlers are permitted, at what rate, and who reads the server logs.

Content and interface checks

Content tooling is judged the same way. A content toolkit such as Semrush's offers drafting and optimisation features. The standard must say which of its outputs count as evidence and which are only starting points. A score is a prompt, not a finding.

Criteria for a credible source

Use official search-engine documentation for how Google or Bing works, and use Google Search Console or Bing Webmaster Tools for the performance data each reports for its own search engine. Those sources do not establish performance across every search engine. Use a tool supplier's documentation to explain its features and metrics, not as independent proof that its results are complete or accurate.

Treat a vendor's marketing blog as a lead to investigate, not evidence on its own. A third-party study is more useful when it discloses its author, method, sample and period, and those details fit the claim being made. In every case, record the exact page and URL, and limit the claim to what that source actually supports.

The same distinction applies when comparing SEO tools: Ahrefs Site Explorer and Moz Link Explorer report data from their own link indexes, while a Screaming Frog SEO Spider crawl records what it found under the saved crawl settings. Neither is a direct report of Google's index. Semrush content toolkit suggestions can inform drafting, but a score is not evidence that a page will rank.

Recording change: before and after

The table below shows a typical register before and after one review round.

Before review

Source named: "Vendor blog"
Owner: Shared team inbox
Review date: Annual, unscheduled
Crawl rules: Not stated
Claims allowed: Any headline figure

After review

Source named: Named documentation page and URL
Owner: Named analyst
Review date: Fixed date tied to contract renewal
Crawl rules: Permitted crawlers and rate limits listed
Claims allowed: Only figures traceable to a cited page

What the register should capture

Each row needs enough detail for a new colleague to repeat the check. Store the URL, the date you read it and the sentence you are relying on. If a figure came from a report, note the sample and the period it covers.

A sample register entry makes those limits concrete:

Google Search Console — Performance report:owner, SEO analyst; allowed claim, the clicks and impressions shown for the selected property and report filters; review when the documentation or reporting changes.
Bing Webmaster Tools — Search Performance:owner, SEO analyst; allowed claim, the performance Bing reports for the selected site; do not present it as Google performance.
Ahrefs — Site Explorer:owner, link analyst; allowed claim, the backlinks reported by Ahrefs' index; do not describe that as a complete count of links on the web.
Screaming Frog SEO Spider — crawl output:owner, technical SEO specialist; allowed claim, findings from the saved crawl configuration; do not treat a crawl as proof that Google has indexed a page.

For each entry, the register should include the exact source-page URL and the date or version checked. The owners above are role examples, not named staff.

Use the SEO source quality checklist for every proposed entry: identify the publisher and whether the source is official, vendor-produced or independent; check its date or version; inspect the method and stated limitations; match the evidence to the precise claim; and record the page title, URL, owner and review trigger. Reject or qualify an entry if those checks do not support the claim.

Review points and ownership

One person owns the register and decides when an entry expires, usually the contract renewal date. Reviews that happen only when something breaks are repairs, not reviews. When an entry lapses, any claim built on it should come out of reports still citing it.

Common questions

Do service standards need legal review?

Only where they touch contracts or personal data. For personal data, the relevant UK framework includes the UK General Data Protection Regulation and the Data Protection Act 2018. Ask a solicitor when a standard affects what you may do with customer information. Otherwise the owner can approve the wording.

How often should a source be re-checked?

Tie it to the contract. Most registers work on a twelve-month cycle, with an extra check whenever a supplier announces a change.

What if a vendor changes its documentation?

Re-read the page, note the change in the register, and suspend any claim that depended on the old wording until someone confirms it still holds. Then decide whether the source returns to the register.

Who signs off a new source?

The register owner signs off, with the analyst who will use the data. Two names is enough, because committees slow the process without adding anything.