Google Discover is the personalised feed inside the Google app and on the mobile google.com homepage. It shows people articles, videos and posts based on their interests, before they type anything. Most guides treat it as a ranking game with a secret checklist. There isn’t one. Google’s own documentation says any indexed page that follows its content policies is already eligible, and no special tags or structured data are required. Whether you get traffic comes down to three things you control: your images, your headlines and how much real topical depth you have.
What Google Discover Actually Is
Google describes Discover as “a part of Google Search that shows people content related to their interests,” based on their Web & App Activity. According to Google Search Help, people find it in the Google app, in the browser on google.com on Android phones and tablets and, on some devices, by swiping right from the home screen. Google also says the Discover performance report includes traffic from Chrome.
The key difference from Search is direction. Search is pull: someone types a query and Google matches pages to it. The Discover feed is push: Google predicts what a person will want to read next from their activity, the topics they follow and the sources they have engaged with. That explains most of how Discover behaves, including why traffic is so uneven.
Discover vs Search: The Differences That Matter
| Google Search | Google Discover | |
|---|---|---|
| Trigger | A typed or spoken query | The user’s interests and activity, with no query |
| What you target | Keywords and search intent | Topics and audiences |
| Traffic shape | Steadier, follows search demand | Spiky; Google calls it “less predictable or dependable” |
| Main visual | Title link and snippet | A large image card with a headline |
| Search Console data | Queries, pages, countries, devices | Pages, countries, dates and appearance types, with no queries |
| Entry requirement | Indexed and relevant to a query | Indexed and compliant with Discover content policies |
The missing query column is the detail most guides skip. You cannot do keyword research for the Discover feed, because nobody searches in it. You can research which topics your audience follows and where your site already shows depth.
Discover Eligibility: What Google Confirms
Google’s “Get on Discover” documentation on Search Central sets a short list of conditions for discover eligibility:
- The page is indexed. Content is “automatically eligible to appear in Discover if it is indexed by Google.” There is no submission form, feed or opt-in.
- It meets the Discover content policies. These cover clear disclosure of ads and sponsored content, no misleading previews, and transparency (dates, bylines, author and publisher information, and contact details for news content).
- No special markup is needed. Structured data can help Google choose an image, but it is not a requirement.
Google adds that eligibility “is not a guarantee of appearing.” Discover also uses many of the same systems Search uses to judge helpful, people-first content. Older content can appear too, so a strong evergreen guide can surface when it matches someone’s interest. Some formats are a poor fit by design: Google says Discover might not recommend job applications, petitions, forms or code repositories.
How Discover Decides What to Show
Google publishes no Discover ranking formula. What it has said publicly comes down to four inputs:
- Interest matching. Content is matched to topics a person cares about, inferred from their activity and from the topics they follow.
- Source preferences. People can follow publishers and creators, tell Google they are “not interested” in a topic, or choose “Don’t show content from” a source.
- Helpfulness signals shared with Search. The quality systems that rank pages in Search also inform Discover.
- Topic-level expertise. In its February 2026 Discover core update post, Google said its systems identify expertise topic by topic. A site can qualify through deep focus on one subject or through strength across several.
The practical takeaway is that Discover rewards a recognisable editorial identity. Ten deep, original pieces on one subject send a clearer topical signal than one shallow piece on each of ten subjects. This is the same principle behind E-E-A-T, applied to a feed instead of a results page.
The February 2026 Discover Core Update and the Follow Feature
The Discover core update. On February 5, 2026, Google announced on the Search Central blog the first core update aimed specifically at Discover. It listed three aims: showing more locally relevant content from websites based in the user’s country, reducing sensational content and clickbait, and showing more in-depth, original and timely content from websites with expertise in a given area. It launched for English-language users in the US, and Google said it would expand “to all countries and languages in the months ahead.” As with any core update, Google warned that Discover traffic may fluctuate.
Following publishers and creators. In September 2025, Google announced on its official blog, The Keyword, that signed-in users can follow publishers and creators directly in Discover and preview their recent articles, YouTube videos and social posts before following. It also said Discover would show more posts from X, Instagram and YouTube Shorts. Together, the two changes favour sites people deliberately return to over sites that win one-off clicks.
The Image Rules: Where Most Sites Quietly Fail
Discover is a visual feed, and Google’s documentation is specific about images. It recommends large, high-quality images that are:
- At least 1200 px wide
- More than 300,000 total pixels
- A 16:9 aspect ratio
Large previews must also be enabled. You do that with the max-image-preview:large robots rule (or by using AMP). Google’s robots meta tag documentation defines large as “a larger image preview, up to the width of the viewport.” Without it, Google may show only a small thumbnail, and in a feed of large cards, a small thumbnail rarely gets tapped.
Google also recommends specifying a representative image through schema.org markup or the og:image tag, and warns against generic images such as your site logo, or images full of text. If you crop images yourself, keep the key subject inside the version you specify.
A worked check. Take a blog post whose og:image is a 1200 x 628 social card. It is 1200 px wide and has 753,600 pixels, so it clears the first two thresholds. But it is roughly 1.91:1 rather than 16:9, and if the card is mostly headline text on a coloured background, it fails Google’s “avoid text-heavy images” advice. A better choice is a 1280 x 720 photo of the actual subject. Now check the robots meta. If an SEO plugin or a theme setting outputs max-image-preview:standard or none, the image work above is wasted. You can generate the correct tag with the Meta Robots Tag Generator and confirm which image a page actually declares with the Open Graph Preview tool.
Headlines: Honest Curiosity, Not Clickbait
Google’s guidance is to “use page titles and headlines that capture the essence of the content,” and to avoid misleading or exaggerated titles, snippets and images. The Discover policy on misleading content goes further. It covers previews that promise details the content does not deliver.
The test we use is simple: would a reader who clicked feel the headline was paid off in the first screen? “Why Your Tomato Leaves Curl in July, and the One Fix That Works” passes if the article names the cause and the fix early. “You Won’t Believe What’s Killing Your Tomatoes” fails twice. It hides the subject from interest matching and over-promises. The 2026 update named clickbait directly, so this is no longer a style preference. Our guide to writing headlines that get clicks covers the craft side, and the free Headline Analyzer is a quick sanity check.
A Five-Gate Readiness Check Before You Chase Discover
Most “Discover tips” lists mix critical blockers with minor polish. Run these gates in order, because a failure at gate 1 makes gates 2 to 5 irrelevant.
- Indexed. Is the page indexed, with a self-referencing canonical? Search Console assigns Discover metrics to the canonical URL, so a duplicate page will show zero.
- Policy-clean. Check the Manual Actions report for Discover or News violations. Make sure ads don’t swamp the content and that news pieces show dates and bylines.
- Image-ready. Confirm
max-image-preview:large, plus a relevant, non-logo, low-textog:imageat least 1200 px wide. - Topically credible. Do you have a body of in-depth, original work on this topic, or is this a one-off?
- Honestly compelling. Does the headline name the subject and deliver on its promise early?
Gates 1 and 3 are technical, and you can audit them across a whole site at once. The SEO Rocket site audit runs a real crawler (Screaming Frog-backed, with JavaScript rendering), so it flags noindexed and non-canonical pages and shows what each URL actually serves to Google. Gate 4 is where most sites actually fall short.
How to Build the Topical Depth Discover Rewards
Because the feed works by interest, the most effective google discover seo work is building clusters, not optimising single pages. Pick the two or three topics where you can be genuinely expert. Cover them more deeply than anyone else, consistently enough that a reader would want to follow you.
Search demand data still helps, because volume across a topic is a reasonable stand-in for how many people care about it. In SEO Rocket you can ask the chat to map a topic into clusters using real Ahrefs keyword data (volumes, difficulty and intent), then run a competitor gap analysis to see where established sites in your niche have depth that you lack. The article writer then drafts against validation gates for length, title and meta limits, so each piece meets a baseline before your editor adds the original reporting and first-hand experience Discover rewards. It is the same playbook proven across 1,000,000+ ranking pages: depth on a topic first, then distribution. For a fuller publisher checklist, see our Google Discover optimization guide.
Timeliness matters too: “timely” is among the qualities the 2026 update rewards. Our news SEO guide covers the overlap with Top Stories and Google News.
Measuring Discover Traffic (and Why It Swings)
Search Console has a dedicated Discover performance report. According to Search Console Help, it appears only after your property reaches a minimum number of Discover impressions, so a missing report usually means low volume, not a broken setup. It shows clicks, impressions and CTR. You can group them by page, country, date and Discover appearance (such as Web Stories and videos), and the default range is the last three months. The Discover performance report guide explains how to read it in detail.
Expect swings. Google says Discover traffic is “less predictable or dependable” than keyword-driven visits and should be treated as supplemental. It gives three reasons: changing interests, changes in the types of content shown, and updates to Google Search. Never build revenue forecasts on a Discover spike; use Search, email and direct audiences as your baseline and treat Discover as upside.
Who Should Prioritise Discover, and Who Shouldn’t
Honestly, Discover matters less to most businesses than most Discover guides suggest. It fits sites that publish frequently on topics people follow for fun or for work: news, entertainment, sport, food, travel, personal finance, technology and hobbies. A local plumber, a B2B SaaS feature page or an ecommerce category page is rarely what anyone wants in an interest feed. Those sites get more from search; our guide on how to increase blog traffic covers the levers.
The middle ground is a business with a real editorial blog. Getting the image and robots basics right costs little, so do it, then keep most of your effort on search, where demand is measurable.
Frequently Asked Questions
Can I submit my site to Google Discover?
No. There is no submission process. Google says content is automatically eligible once it is indexed and meets the Discover content policies.
Do I need AMP or structured data for Discover?
No. Google says no special tags or structured data are required. AMP is one way to enable large images, but the max-image-preview:large robots rule does the same job on a normal page.
Why did my Discover traffic suddenly drop?
Discover traffic swings with user interests, with changes in the content types shown and with updates such as the February 2026 Discover core update. Check the Manual Actions report first. Then compare recent headlines and images against Google’s guidance before assuming you have been penalised.
