Publisher site audit in 2026: four pillars

«SEO has evolved. You're optimizing for presence, accuracy, efficiency, and quality» — that is how Harry Clarkson-Bennett, SEO Director at The Telegraph, frames it in his guide for Search Engine Journal. He breaks a publisher audit into four pillars: technical SEO, content, E-E-A-T and brand consistency as an entity.

The core shift: readers now arrive through three different channels, and mixing them up is a mistake. Passive — content found off-platform, in recommendation feeds, social and AI systems. Active — search itself. Owned — your site as brand headquarters. Up to 86% of clicks, per the figures cited in the piece, come from discovery, while Google still sends the open web over a trillion clicks a year. Nothing old disappeared; a second large entrance simply opened next to it.

On trust: confidence in news has fallen to 37% across 48 markets, the lowest since 2015, and off-site reputation signals now carry measurable weight — brands with a robust Trustpilot profile get cited by AI in 75% of cases, a basic profile gets 53.5%, no profile gets 1%.

On content, the author separates two modes. News lives by the golden hour: headline, image, authorship and Google News Sitemap presence have to be right immediately. Evergreen is judged differently — does it still pull traffic months after publication, and was it refreshed for demand spikes.

What to do in practice:

• confirm bots actually reach the content: robots.txt, crawl stats, indexation report and 302 chains eating crawl budget;
• clean up semantics — h1–h6 hierarchy and structured data, so machines don't have to guess;
• link fresh news to its topic hub straight away, not a week later;
• measure speed with field Core Web Vitals from Search Console, not Lighthouse alone, and benchmark against competitors;
• for Discover — a clickable OG headline (it can differ from the SEO one) and images from 1200 px;
• expose your authors: journalist profiles, editorial policy, contacts, About Us;
• reconcile brand data and Organization schema with SameAs everywhere — YouTube, podcasts, GBP, third-party mentions;
• prioritise by scale × impact × effort, and deliberately ignore the small stuff.