Know exactly what is holding a page back.

Most SEO health checks produce a report with two hundred warnings, of which four matter. The useful version is ordered by consequence: check the things that make a page invisible before you check the things that make it slightly less good.

Order the checks by consequence

The sequence matters more than the checklist. Work down, and stop treating anything lower as urgent until everything above it passes.

1. Can the crawler reach it? robots.txt, redirect chains, server errors.
2. Can it be indexed? noindex meta, X-Robots-Tag header, canonical target.
3. Do the signals agree? Sitemap membership versus robots meta.
4. Is the structure sound? Headings, title, description.
5. Are the entity signals clean? Structured data, sameAs, schema matching the page.
6. Can answer engines read it? AI crawler access.

Items 1 and 2 are binary and they are where the expensive failures live. A page with a stray noindex is not a page that ranks badly. It is a page that does not exist as far as search is concerned, and it looks completely normal to every human who visits it.

The checks that find real problems

Contradictory indexation signals. A URL in your sitemap that also carries a noindex tag is telling Google to crawl something you then tell it to drop. This happens constantly after a site restructure, and neither signal is wrong on its own, so no single-purpose tool flags it.

Canonical pointing elsewhere. A canonical tag is a strong hint that the real version of this page lives at another URL. Pointed wrongly, it hands your page’s relevance to a different one. Check that the canonical on a page resolves to that same page unless you deliberately intend otherwise.

Heading levels that skip. An H2 followed by an H4 breaks the document outline that both screen readers and extraction pipelines use to understand structure. It is trivial to fix and it improves how cleanly your content can be quoted.

Schema that contradicts the page. Structured data claiming a price, a rating, or an organisation name that does not appear in the rendered content is worse than shipping no schema at all. Validate that it parses, then read it against the page.

AI crawler blocks. Many robots.txt files still block or omit GPTBot, PerplexityBot, and ClaudeBot by inheritance from an old template. If a client wants to be visible in AI answers, that file is the first gate, and it is frequently closed by accident rather than by decision. This part of the checklist is automated: the free AI-readiness check tests eight named crawlers against a live robots.txt and returns a pass or fail for each, along with sitemap, structured data, llms.txt, title, meta and H1.

The scoring sheet

Record one row per page so results are comparable across audits and across clients:

URL | Crawlable | Indexable | Sitemap agrees | H1 count | Heading skips | Schema valid | sameAs present | AI crawlers allowed | Blocker?

The Blocker? column collapses everything into the only field a client cares about: is this page capable of ranking right now, yes or no. Keep the detail for your own remediation queue.

Doing this across a client roster

Running this on one site is an afternoon. Running it across twenty client domains every quarter, then tracking which findings were actually fixed, is the part that does not scale by hand. That is what AltoRank automates: crawl, check, generate what is missing, and publish behind an approval gate so nothing changes on a client site without a human saying yes. See the alternatives comparisons for how it compares, or read how to read a SERP if your problem is keyword selection rather than technical health.