How to optimise a website for Google Discover

How to optimise a website for Google Discover, the article feed in the Google app and Chrome on phones that displays content without users searching for anything. For content sites, Discover can bring more traffic in a day than search brings in a month; it can also disappear just as quickly. This guide explains what you can and cannot control, covering Google's requirements and what we have observed on client sites receiving traffic from it.

A warning from the outset: Discover is not a channel you "optimise" like search, using keywords. It is a recommendation channel that chooses content for each user's interests; what you can do is meet the eligibility requirements and produce the kind of content the system recommends.

How to optimise a website for Google Discover

How it works

Discover matches indexed content to users' interests, inferred from their Google activity (searches, followed topics and Chrome history if enabled). There is no query, so there is no "position"; there are impressions in some users' feeds, clicks and, if users respond well, wider distribution. Content usually appears in the first few days after publication and disappears afterwards, except for evergreen articles that may be recommended for months. Any indexed page is eligible; no registration is needed.

Google's requirements

• An indexed page that complies with Discover content policies (no misleading content, no clickbait headlines, no adult content and no dangerous medical or financial advice).
• Large images, at least 1.200 pixels wide, allowed through the max-image-preview:large setting in meta robots or by using AMP. Without a large image, the article may appear, but with a small thumbnail and a much poorer click-through rate.
• Headlines that describe the content rather than hiding information to force a click. Google explicitly says it avoids clickbait headlines and exaggerated content.
• Trust signals: an identified author, a visible publication date and a clear source. These are the same E-E-A-T signals that search requires.
• A good mobile experience: Discover is almost exclusively mobile, so page speed and stability matter directly; details are in the article on page experience signals.

What content gets into Discover

From data on the sites we monitor, the patterns are consistent:

1. News and timely analysis on topics with broad interest: technology, health, personal finance, cars, sport and lifestyle. An article about a change at Google or to a law receives distribution in the first 48 hours.
2. Evergreen content with a personal or practical angle: "what I learned after", "how I solved it", guides with good images. It can return to the feed months later.
3. Lists and comparisons with distinct images for each item.
4. Articles that satisfy curiosity rather than a need: people read Discover in their free time, not at their desks with a problem to solve.

What rarely gets in: product pages, service pages, narrowly technical articles without a relevant image, press releases and content without an author. For an agency like ours, Discover brings traffic to articles about changes at Google and about AI, and almost none to cPanel guides.

What to optimise in practice

The image

The main image should be at least 1.200 pixels wide, in a 16:9 or 4:3 ratio, with a clear subject and no text overlay (text gets cut off when cropped). Declare it as og:image and in the article's structured data. Use a real photograph or a quality illustration, not generic stock imagery; images repeated across many articles receive less distribution. The meta robots max-image-preview:large setting is mandatory on all article pages; check that the SEO module does not limit it to "standard".

The headline and first paragraph

The headline says what the article is about and what readers will learn, in no more than 70 characters to avoid truncation, without "you won't believe" or capitalising every word. The first paragraph confirms the headline in two sentences; Discover users leave pages that delay the answer, and these quick exits reduce subsequent distribution, following the logic described in the article on interaction signals. Examples of the headlines we use are in the article on SEO headline examples.

The author and entity

A real name, an author page with a biography and links to profiles, and Person structured data linked to Article. Discover favours sources it can identify and that users follow; Chrome's Follow button relies on the site's RSS feed, so check that the feed exists and is declared in the head.

Frequency and topics

Sites that appear consistently in Discover publish regularly on the same topics; the system learns which subjects a source covers and whom to recommend them to. An isolated article on a topic unfamiliar to the site receives little distribution. Choosing the topics you cover systematically is a content strategy decision, described in the guide to SEO content strategy.

Structured data

Article or NewsArticle with headline, image (multiple aspect ratios), datePublished, dateModified, author and publisher. These are not mandatory for Discover, but help identify the image and author; validation is done using the structured data validator.

How to measure

Search Console has a separate Discover report in the performance section, which appears only if the site has had Discover impressions in the last 16 months. It shows impressions, clicks and click-through rate by page, without queries (there are none). A normal click-through rate in Discover is much higher than in search (5-15% on pages with good images), and impressions come in waves: an article receives 50.000 impressions in two days, then nothing. Monthly comparisons make sense only for totals and the number of articles that entered the feed, not for individual articles.

In Analytics, Discover traffic appears as direct traffic or as google / organic, depending on the app; exact separation is not possible without parameters, and adding parameters to addresses is not allowed for Discover. Use Search Console as the source of truth for this channel.

Why Discover is unpredictable and what to do about it

Distribution depends on users' interests, competition on the topic that day and the reaction of the first people who see the article. The same type of article can receive 100.000 impressions once and 500 the next time. Google says Discover traffic is "less predictable" than search traffic and recommends not building your business on it. We treat it as a bonus for content made for search anyway: we meet the requirements (image, headline, author, speed), publish regularly on our topics and do not change strategy after a peak or an empty week. What changed in the Discover algorithm in early 2026 and how publishers reacted is covered in the article on the February 2026 Google Discover update.

Discover for small businesses and service sites

The obvious question: is the effort worthwhile for a site that is not a news publication? The practical answer: partly. A service site will not receive Discover traffic on its service pages, but its articles can enter the feed if they are timely and have good images; that traffic brings brand visibility, not customers directly. For a small business, the eligibility requirements (large image, author, speed, structured data) are good for search anyway, so meet them at no extra cost and let Discover come when it comes. What does not make sense is changing your editorial direction to broad-interest topics just for Discover; that traffic does not convert and disappears with the first algorithm change.

Discover and Google's other channels

An article entering Discover gains no advantage in search, and vice versa; the systems are separate, with shared signals. Google News, for publications, requires registration and compliance with news content policies; Discover does not require registration. "Top stories" results in search follow their own freshness rules. Sites appearing consistently in all three have one thing in common: they publish frequently, on consistent topics, with identified authors, and do nothing special for any of the channels.

Mistakes that remove a site from Discover

• Clickbait or exaggerated headlines, even on good articles: the system detects them and reduces distribution for the entire site.
• Repeated stock images or images with text overlays.
• "Timely" articles republished with a new date but no new content; changing the date without changing the content is considered misleading.
• Pop-ups covering the content on arrival, on mobile; Discover users leave immediately.
• Content outside the site's topics, published solely for volume.

Discover rewards what search rewards, with an emphasis on images and timeliness. Sites with consistent traffic from it do nothing special for Discover; they produce good content, with good images, on time, and let the system find it. The remaining principles are in the guide to content that generates customers.

Sources

• Google Search Central, Google Discover: how it works, image requirements and official recommendations.
• Google Search Central, helpful, reliable, people-first content: quality criteria also applied in Discover.

Source references

• Google Search Central, Google Discover
• Google Search Central, conținut util, de încredere, scris pentru oameni