On-page and technical signals

After a page is crawled, SEO Perception evaluates a set of on-page and technical signals derived from the HTML and response. This article explains the categories of what we look for. It is not a complete, versioned checklist — rules and priorities can change as the product improves.

Data comes from the crawl described in How the crawler works. Search performance context from Google Search Console complements these signals for the same URLs where both exist.

SERPs and sharing

We consider whether titles and meta descriptions are present and roughly appropriate for typical search-result display, and whether Open Graph tags form a complete set for social previews. Pages that redirect (3xx) are flagged so you can prefer direct or canonical targets where it makes sense.

Indexation hints

Canonical URLs and robots meta tags help clarify how you intend a URL to be indexed. When a page is marked noindex, we generally avoid treating indexing-oriented suggestions (such as missing title) the same way as for indexable pages.

Content and structure

We look at word count bands, whether there is a single clear H1, and use of H2 headings to structure content. Very thin or extremely long pages are called out at a high level.

Images and accessibility hints

We distinguish images that lack alt text from those with empty alt (decorative intent is not inferred automatically — both patterns are reported for your review).

Internal and external linking

We summarize how many internal and external links appear on a page. On substantial pages with no internal links, we may suggest improving internal navigation. Where sampling runs, broken or redirecting outbound links can be surfaced so you can clean up dead or indirect targets.

When discoverable XML sitemaps list URLs on your site, we may flag URLs whose first HTTP response is not 200 (for example redirects or errors), so sitemap entries stay aligned with what crawlers receive. See Sitemap URLs and HTTP 200.

Structured data and mobile-related basics

We detect whether structured data (for example JSON-LD referencing schema.org) is present; we do not validate every schema type in depth in this layer. Typical viewport meta for responsive layouts and a declared document language (html lang) are also considered. Favicon presence is evaluated on the homepage only — see Favicons and touch icons for why icons matter and how to add them across devices.

How this shows up in the product

Summaries and per-URL metadata appear in the website workspace. Prioritized improvement ideas may appear elsewhere in the app as that experience matures.