What a technical SEO selection checklist means for your next review

What to take away

Google's robots.txt specification page is updated as crawling rules change, so audits that quote robots.txt directives must cite that page, not a cached summary.
A selection checklist is a document set, not a scoremethod, data sources, evidence of skills and a written scoring rubric.
Every figure in a vendor's pitch needs a named source and a date, or it is a labelled illustrative example.
Shortlist only suppliers who will show how they test, not just what they found.

Why the checklist exists

Selection went wrong for English teams in a quiet way. Buyers compared findings, not methods, so the loudest deck won. When Google updated its robots.txt specification page, audits that quoted older rules aged badly within months.

The fix is procedural. Ask each supplier for the same four documents and score them on the same rubric. That is the core of a technical SEO selection checklist: identical questions, identical evidence, comparable answers.

The documents to request

Request these before any demonstration. A supplier who cannot produce them is not ready for a shortlist.

Four documents to request

Method statementcrawl configuration and exclusions
Data sourcestools, APIs, log files
Skills evidencetraining records for staff
Scoring rubricissue ranking by impact

Method statementhow the crawl is configured, what is excluded and why.
Data sourceswhich tools, which APIs and which log files feed the findings.
Skills evidencetraining records or course completions for the people doing the work.
Scoring rubrichow the supplier ranks issues by impact and effort.

For the wider comparison frame, the parent guide on technical SEO: reviews and comparison methods for 2027 sets out how these documents feed a side-by-side score.

Technical checks to add to the rubric

Use technical checks as rows in the supplier comparison. For each, ask for a sample finding and evidence of how it was verified.

Crawlability and crawl budget:Review crawl paths and robots.txt directives, and assess whether crawl activity is being spent on useful URLs where crawl budget is relevant. Request the crawl configuration and, where available, log-file evidence. Screaming Frog SEO Spider and Sitebulb can be used to inspect crawls; Screaming Frog Log File Analyser is an example of a log-file analysis tool.
Indexing and URL signals:Check indexability, canonical tags, status codes and redirect chains. Compare findings with Google Search Console indexing reports, and review XML sitemaps for accurate, indexable URLs.
International targeting:Check that hreflang annotations are valid and correspond to the intended language and regional versions.
Rendering and performance:Verify that important JavaScript-rendered content and links are accessible to search engines, and review Core Web Vitals evidence.
Structured data:Check whether structured data is present where appropriate, valid and consistent with the visible page content; ask the supplier to show its validation method, such as the Google Rich Results Test.

Rules and controls to test

Crawling controls are the easiest area to verify, because the specification is public. Ask the supplier to explain how Google interprets robots.txt directives, including the difference between a disallow and a noindex instruction, using the current robots.txt specification guidance as the reference point. A supplier who cannot talk through that page is guessing.

The same test applies to claims about regulation. If a pitch says a marketing practice is banned or permitted, the supplier should point to the regulator's evidence base. In England that means the Competition and Markets Authority for consumer protection rules and the Advertising Standards Authority for advertising codes and rulings. For consumer-facing claims, ask which provisions of the Digital Markets, Competition and Consumers Act 2024 apply; for advertising claims, check the CAP Code as well as ASA rulings.

The Federal Trade Commission's published studies are a US example of how a regulator documents the evidence behind its positions.

Skills and training evidence

Skills claims are the hardest to check, so treat them as a document request. Ask for completion records from recognised training providers. The Chartered Institute of Marketing's digital marketing courses give one benchmark for the kind of structured training that supports SEO and content work.

This matters because tool access is not skill. A licence proves a subscription, not an ability to read a crawl. Score the people, then the software.

Comparison table

Use this to score two or three suppliers on the same scale. Weights are yours to set; the point is consistency.

Supplier scoring rubric

Check

Method statement: Crawl configuration and exclusions
Data sources: Tool list, APIs, log files
Rules knowledge: Robots.txt and indexing walkthrough
Skills evidence: Training records for named staff
Scoring rubric: Issue ranking by impact and effort
Reporting cadence: Sample report with dated figures

Evidence requested

Method statement: 20%
Data sources: 20%
Rules knowledge: 20%
Skills evidence: 15%
Scoring rubric: 15%
Reporting cadence: 10%

Weight

Method statement
Data sources
Rules knowledge
Skills evidence
Scoring rubric
Reporting cadence

Evidence requested

Method statement: Written crawl configuration and exclusions
Data sources: Tool list, APIs, log files
Rules knowledge: Walkthrough of robots.txt and indexing controls
Skills evidence: Training records for named staff
Scoring rubric: Issue ranking by impact and effort
Reporting cadence: Sample report on a fixed cycle

Weight

Method statement: 20%
Data sources: 20%
Rules knowledge: 20%
Skills evidence: 15%
Scoring rubric: 15%
Reporting cadence: 10%

For example, a team paying £400 a month for an audit retainer should expect all six rows to be answered in writing before renewal.

Illustrative example: a supplier scores 3 out of 5 on the method statement because the crawl configuration is written but the exclusions list is missing. At a 20% weight, that row adds 12 points to a maximum of 20.

Red flags to record

Note these in the scoring sheet rather than debating them in the meeting.

Unsourced figures.
Findings that cannot be reproduced on a second crawl.
Refusal to name the people doing the work.
Rubrics that rank every issue as high priority.

One refusal is a question. Three refusals is an answer. How suppliers handle source requests is covered in our guide to technical SEO agency reviews: data and sources.

Scoring and sign-off

Score independently, then reconcile. Where two reviewers disagree by more than one point, the supplier must supply the missing evidence, not a better explanation.

Sign-off needs a named owner and a review date. Diarise the next check at twelve months, or sooner if the supplier changes its method statement or its tooling.

Common questions

How long should a selection checklist be?

Long enough to cover method, data, rules, skills and scoring, and short enough to complete in one scoring session. Six to ten weighted rows is workable for most English teams.

Do we need a checklist for a single supplier?

Yes. The checklist is a renewal and benchmarking record, so it still applies when there is no competition. It also gives you a baseline if the supplier changes hands.

What if a supplier refuses to share its method?

Treat it as a failed row, not a negotiation. You cannot compare or renew a service you cannot inspect, and the refusal itself is useful evidence for the score.

Should the checklist mention tools by name?

Only where the tool supports a specific check, such as log file analysis. Naming a product is not evidence of method, so keep the weight on the process.