Auditing an agency, company or portfolio website

Four detected categories run the same module: agency, corporate, portfolio and personal. The category engine recognises an agency from paths such as `/services`, `/case-studies` or `/clients` and wording such as "our clients" or "case study"; a corporate site from `Organization`, `Corporation` or `LocalBusiness` markup and paths such as `/about`, `/company`, `/investors` or `/press`; a portfolio from paths such as `/portfolio`, `/work` or `/projects`; a personal site from `Person` or `ProfilePage` markup and paths such as `/about-me` or `/now`.

Because `/about` and `Organization` markup are common, many sites carry the corporate category as a secondary one, kept when it reaches 35 percent of the primary category's score. They then receive these rules alongside their main module.

No contact page — a URL containing `/contact` — is reported at high severity, the most severe rule of the module: for a business site, being reachable is the point. When a contact page exists but no page shows a form in a contact context, that is reported at medium severity.

No `tel:` link and no `mailto:` link anywhere is reported at medium severity, because a form alone leaves a visitor with no direct way to reach a person.

No portfolio, work, projects, case-studies or réalisations page is reported at low severity; no about page at low severity; no team page for information only.

No legal-notice page — `/legal`, `/mentions-legales`, `/imprint`, `/legal-notice` — is reported at medium severity. When the legal-document discovery finds the notice elsewhere on the site, the finding is withdrawn, and when the crawl stopped at the page budget first, it is downgraded to "could not be verified".

When no services page was reached, the module says the services listing could not be verified rather than reporting it missing.

`structure.cta` expects a contact or quote action — contact, quote, phone, email or a form — on agency and corporate sites, and on local businesses, and reports its absence. Portfolio and personal sites are not held to it.

None of these four categories is reweighted in the score: security, performance, SEO, accessibility and privacy weigh as they do for any site, and the commerce checks do not count unless the site is also detected as selling.

A FAQ is not expected of them either. `structure.faq` reports a FAQ as optional for these sites and never counts its absence against them.

The rules find pages by URL pattern in English and French, so a site whose contact page lives at an unusual address can be reported without one. The report shows which patterns were searched, and the finding is easy to dismiss when the page exists under another name.

The scan never submits a contact form, so it cannot tell whether messages arrive. A form that looks complete and delivers nothing is invisible to it — which is why a real test submission after every change is worth more than any audit rule.

Pages a visitor cannot reach from the home page are a separate question, answered by the structure checks: they report important pages — contact, legal, about — that exist but are not linked from the home page's navigation or footer.

Beyond this module, a shared baseline runs once on every audited site, whatever its category: favicon, viewport, canonical link, `lang` attribute, pages excluded from indexing, Open Graph and Twitter card tags, mixed content, a cookie-consent signal and a link to the sitemap in the HTML.

For an agency or company site, those baseline findings are often the ones that matter most: a missing canonical or an accidental `noindex` removes the pages the business depends on from search, and a missing Open Graph image makes every shared link look unfinished.

A canonical or sitemap-link finding that the main scan has already made is not reported a second time by the baseline, so the report does not show the same problem twice under two codes.