Discover quietly became one of the largest referral sources on the open web. An article that lands in a user's Discover feed gets more traffic in a day than the same article sitting at position six in search gets in a month. And unlike ranked search, there's no keyword you can optimize for. Discover is a feed ranking; you're either eligible or you aren't.
The Discover Readiness Audit is a 14-signal check against the eligibility rules. Paste an article URL. You find out in 15 seconds.
What fails most often
The same three things, in this order:
1. The hero image isn't wide enough. Google's requirement is 1200 pixels minimum on the longest side, and the image has to be the one Discover actually pulls. Usually og:image, sometimes the first <img> the crawler sees. If your CDN serves a 600-pixel responsive variant to mobile user agents, you lose eligibility even though your desktop site looks fine. The audit fetches the image with a HEAD request and flags width when it can read it.
2. max-image-preview:large is missing. Without this meta tag, Discover shows a thumbnail-sized image instead of the full card. The difference in click-through is roughly 3x. Add it to the <head> of every article: <meta name="robots" content="max-image-preview:large">.
3. There's no NewsArticle or BlogPosting schema. Discover strongly favors articles with schema that declares datePublished, author as a Person or Organization, and image. A bare <article> tag with no JSON-LD is eligible in theory and never chosen in practice.
What else matters
The audit also checks: publish freshness (30 days is the soft cutoff), author byline visibility, NewsArticle vs BlogPosting schema choice, viewport zoom allowed, word count, X-Robots-Tag header for a noindex override, and dateModified as a secondary freshness signal.
The output is a numeric score (0-100), a list of failed checks grouped by severity, and an AI fix prompt that takes all of them to Claude for exact code.
How Discover differs from Search
You can't keyword your way in. Discover ranks on user-interest match against the article's entities, quality signals, and hero imagery. Three implications:
- Thin articles don't ever make Discover, regardless of keywords. Under 400 words is usually a dead stop.
- The
NewsArticle/BlogPostingdistinction matters more than in Search. Pick the right one; don't hedge. - Your author entity. The
Personschema withsameAsandknowsAbout. Carries weight. That's why the E-E-A-T Audit pairs well with this one.
The book reference
Chapter 28 of The $20 Dollar Agency. Mobile-First & PWA. Is the context. Discover is a mobile-first feed, ranked by mobile crawling. Chapter 14 of The $100 Network. Dynamic OG Images, Social Cards, and Visual Assets. Covers the hero-image pipeline you need to hit the width requirement reliably across a large content base.