Google Discover

Definition

What is Google Discover?

Google Discover is Google's personalised news feed in the Google app and on the Google start page. The feed delivers content without an active search query, based on user interests, location and behaviour in Search, Chrome and YouTube.

Where does the Discover feed appear?

On Android devices Discover appears in the pre-installed Google app, on Pixel phones additionally by swiping right on the home screen. On iOS the Google app opens the feed; a desktop test has been running since 2025.

Discover gets its biggest reach via mobile: for German news publishers the feed is now the most important mobile Google channel, ahead of the classic search results page.

How does Discover differ from Google Search?

Search responds to a typed question; Discover runs without a search query. Google decides which article gets into the feed by rating three factors: topic relevance for the user, article quality and freshness.

From a publisher's point of view that means: reach does not come from keywords but from topical authority, headline, image and precise timing. Exactly these levers are what the rest of this page describes.

Rating logic

How does Google Discover rate articles?

Google Discover rates articles on three core factors, topic relevance for the user, article quality and freshness, and matches them against the reader's interest profile.

Which signals from Chrome and YouTube does Discover evaluate?

Google compiles an interest profile per user from several data sources: search history, YouTube consumption, location, web and app activity and past Discover clicks. The profile updates almost continuously; a reader who opens business pieces for a week sees more business topics in the feed on day eight.

For newsrooms that means: an article lands in the feed when the site's own topic profile matches the interests of many users in a market segment. Anyone who stretches a profile too far weakens their position in the feed.

How quickly does the algorithm decide on visibility?

Discover predicts in advance, per article and reader, how likely a click is. As soon as many users leave a URL after seconds, the next delivery is quieter; a piece's freshness window ends after roughly a week.

Within this window it is above all the image and headline pair that decides on clicks. Weak headlines cost the article its top spot in the feed, even if the topic fits.

Traffic share

How much traffic does Google Discover deliver to publishers?

According to a NewzDash analysis across more than 400 publishers worldwide, Google Discover delivers news publishers 67.51 percent of their mobile Google traffic, more than classic Google Search.

What Discover CTR is realistic?

According to NewzDash analyses from the first quarter of 2025, the average Discover CTR is around 8 percent, roughly four times the value in classic Google Search. Individual articles pull several hundred thousand to millions of clicks per piece.

This click-through rate creates an economic lever far beyond the search results page. A strong Discover wave lifts the page RPMs of a mobile news domain so clearly that the reach channel has a direct effect on ad sales, subscription funnel and brand contact.

Why does Discover beat organic search for news?

The NewzDash series shows a clear trend: the Discover share of mobile Google traffic rose from 37 percent in 2023 to 67.51 percent in Q4 2025. In the same period the share of classic Google Search fell from 51 to 27 percent.

Other observers see the trend too: the Dutch publisher CMS provider Newsifier reports in its 2026 optimisation guide that about nine out of ten of the news brands it supports now build more reach via Discover than via classic search.

Topical authority

What role do top entities play for Google Discover?

Top entities, people, places, organisations or events that Google particularly trusts a brand on, decide which topics bring a publisher into the feed.

How do newsrooms identify their top entities?

Anyone who wants to find their top entities compares in Google Search Console the performance data of the last sixteen months in parallel across Discover, Search and News. The URLs with the most traffic produce a list from which people, places, brands and topic areas can be read that Google attributes the domain as a source for.

Local publishers regularly see municipal utilities, regional companies, sports clubs, hospitals and authorities high up in the cluster. Business media find industry brands and indices, trade titles their core topics and recurring study providers; the list marks the reach assets that topic planning should prioritise.

Why do random trending topics dilute the profile?

As soon as a newsroom moves into topic areas in which the domain has historically built no authority, the probability of Discover delivery falls, even if the chosen topic has high search demand. Reach only arises where search demand and the site's own topic trust coincide.

Random celebrity or trend pieces beyond the core dilute the profile further. Google reads this mix as a blurred topic signal and subsequently downgrades the brand on material the domain has so far carried reliably. More on the mechanism behind this on the page What is publisher SEO.

Ranking

Which ranking factors decide Google Discover?

Discover rates E-E-A-T signals (experience, expertise, authority, trust), headline and image quality and the page experience. According to Google itself, trust is the most important part of the E-E-A-T concept.

What role does E-E-A-T play?

Discover delivers content especially when Google can clearly assign the author to a subject responsibility. Author profiles with job title, career and topic focus belong in every section, linked to the matching topic pages.

The signals are particularly critical in YMYL topics such as health, finance, law and politics. Anyone who publishes here without a recognisable profile loses in Discover faster than on soft topics. How newsrooms implement that in daily work is described in the best practices for editorial SEO.

How do headlines and images work in the feed?

Headline and image are the only window in the feed. A Discover-ready headline raises a question, combines seasonal relevance with concrete benefit or sets a number as an anchor, without giving away the main point in the title or tipping into clickbait.

Images meet technical minimum standards: 1,200 pixels minimum width and the meta tag max-image-preview:large in the head. Emotional subjects with people and action beat interchangeable stock photos; details also in journalistic SEO.

Channel comparison

How does Google Discover differ from Google News?

Google News draws on content for active search queries and rates above all prominence, freshness and source authority. Discover runs without a search query and responds to user interests; both are separate logics that each carry differently.

Which content lands in which channel?

In Google News land hard news, breaking news, investigations and background on current events. The channel assigns visibility by freshness and authority logic.

In Google Discover land explainer formats, service topics, people pieces and soft content as well as breaking material, as soon as they match a user's interest profile. The feed favours visually strong and emotionally engaging topics.

Which workflow serves both surfaces?

Before publication the newsroom sets the goal: news anchor for Search and Top Stories, or Discover anchor for the feed. The image and headline pair is built differently depending on the goal: factual and keyword-close for search, curiosity-driven and click-strong for Discover.

How news and Discover strategy overlap operationally is described on the page What is Google News. Successful newsrooms do not treat the two channels separately but as two goals of one shared publishing process.

Refresh strategy

How do publishers use refreshing and republishing for Discover?

Publishers keep Discover reach through two levers: refreshing updates the content on the same URL with a new modified date. Republishing additionally sets a new publication date and classifies the article as a new piece for Google.

What distinguishes refreshing from republishing?

Refreshing suits evergreens with existing rankings and news pieces that need a new image or a sharper title; the URL authority is preserved. Republishing suits news texts that have fallen out of the feed but carry again in substance.

Both levers need a real content update: new figures, added paragraphs, a revised headline. Without this substance the date swap works short term but backfires long term on the republishing effect. The transition to evergreen optimisation provides the logic for maintaining existing content.

What risks do the 2025/26 spam updates bring?

Since the Discover core update of February 2026, Google has clearly tightened its assessment of fake updates. Three patterns lead to visibility losses particularly fast: a moved publication date without a content intervention, old articles on a new URL without a 301 redirect, and headline or image changes that run without a substantive revision of the text.

Practical safeguard: anyone who republishes adds a short update note at the end of the text with date and bullet points on the intervention. That creates transparency for readers and gives the algorithm an additional trust signal that protects against downgrades.

Best practices

Which best practices work in Google Discover?

Successful Discover articles follow click-strong headline patterns, rely on high-quality images from 1,200 pixels wide and separate the OG title from the meta title for separate click surfaces.

Which headline patterns perform in the feed?

Discover headlines work with a question format, number triggers or seasonal relevance. They spark curiosity, name a concrete benefit and do not give away the main information in the title; otherwise the click probability falls.

Practical trick: maintain the OG title as a separate CMS field. It may be up to 120 characters long, compared with around 70 for the meta title, and may be phrased to spark curiosity instead of factual and keyword-close. That way one URL serves both click surfaces cleanly.

Which image strategy draws clicks?

Images meet three criteria: minimum width 1,200 pixels, the meta tag max-image-preview:large in the head and a subject with a person or action instead of an interchangeable stock photo. Generic topic images directly cost the article reach in the feed.

In longer articles images belong embedded in the body layout so readers do not meet walls of text. Discover measures engagement signals after the click too; anyone who lets readers bounce back into the feed early loses visibility on the next story.

Daily routine

How do publishers organise the Discover day?

Successful newsrooms anchor Discover in a 30-minute morning routine, a midday shift for title and image follow-up and an evening review of the traffic data.

What does the 30-minute morning routine look like?

Before the morning conference an editor opens the live monitoring and sorts the running Discover pieces by reach and click-through rate. He intervenes on pieces that slip below 7 percent CTR or are just tipping out of the feed: adds paragraphs, swaps images or headlines, resets the date published as soon as the text has really been expanded in substance.

A rule of thumb helps not to overdo the setting: pieces that are taking off anyway, nobody touches. Every intervention in a running Discover hit cuts off the wave; the algorithm re-rates every CMS change.

Which time slots do newsrooms work during the day?

The midday slot is for sharpening: a second shift pulls the live data of the morning pieces and revises title, lead image or teaser text as soon as the click-through rate falls. In the late afternoon someone evaluates the traffic winners and losers and thereby sets the topic list for the next morning conference.

How these shifts fit into a clean news and republishing workflow is described in news optimisation for publishers. Newsrooms with this routine get two to three times as much reach out of the same topic area as houses without a fixed daily schedule.

Monitoring

How do publishers measure Discover traffic?

Google Search Console shows Discover performance over up to 16 months; live tools such as Chartbeat, Marfeel and NewzDash add second-by-second steering and topic benchmarks.

What does Google Search Console show for Discover?

As soon as a domain receives Discover traffic, a separate Discover report appears in Search Console. It delivers clicks, impressions, CTR and position per URL, in a 16-month comparison, filtered by country or device.

Particularly valuable is the comparison of two periods: three months now against three months before. That way the newsroom recognises topic areas that gain or lose visibility, the basis for refresh, republishing and hub decisions.

Which tools complement the monitoring?

Live tools such as Chartbeat, Kilkaya, Piano, NewzDash, Trisolute and Marfeel show second by second which article is running in the feed, including dwell time and engagement. For visibility analysis, SISTRIX, Semrush and Ahrefs.

Marfeel additionally publishes its own Discover Monitoring with benchmarks for feed shares, AI tiles and video hosts. Concrete evidence of what such tool setups trigger in publishing houses can be found in the Case studies.

Volatility

Why does Google Discover fluctuate so much?

Google Discover fluctuates clearly from day to day because the feed follows every user interaction. Rule violations in the newsroom and technical problems amplify the fluctuations; in extreme cases the referrer disappears for weeks or months.

Which editorial mistakes cause drops?

Four triggers keep showing up in the volatility pattern: clickbait headlines, the pure date swap without a substantive update, suddenly adopted trending topics without topic trust, and paywall barriers directly behind the Discover click. In every case many readers bounce back into the feed, and Google reduces the brand on the next wave.

Anyone who builds reach exclusively via Discover sits on a brittle leg with every Google update. A diversified mix of direct, newsletter, push and social remains mandatory; matching evidence is provided by the trade articles and podcasts of the past months.

How do Core Web Vitals affect delivery?

Weak Core Web Vitals, especially poor CLS values (layout shifts) and INP values (input responsiveness), measurably lower Discover visibility. Typical causes lie in mobile ad sales: lazy-loading ad slots, unsized slot dimensions or scripts that recalculate the layout while scrolling.

A technical audit checks per template which ad slots create layout jumps and which scripts burden INP. Only clean values secure lasting Discover visibility; the content alone is not enough.

AI outlook

How is AI changing Google Discover?

AI is changing Google Discover in three places: AI summary tiles in the feed, localisation of the delivery logic and a shift of clicks to video hosts such as YouTube.

How do AI Overviews affect the Discover feed?

According to Marfeel Discover Monitoring, AI summary tiles in the US already made up 21.6 percent of impressions in positions 1 to 5 in December 2025. Clicks on these tiles led in around 77 percent of cases to YouTube, not to publisher domains.

In Germany the pressure is increasing further: since 8 October 2025 Google has been serving AI Mode in the German-language market too. Anyone who wants to keep visibility builds content to be citable, with answer-first paragraphs, solid facts and clean schema markup.

Why is Discover becoming more local and more visual?

The first standalone Discover core update from 5 to 27 February 2026 shifted visibility towards local providers. In the US market local news brands clearly gained Discover share, while internationally oriented reach brands lost ground, a warning signal for publishers with a national or international profile.

In parallel, Discover is shifting click streams to video platforms: moving image is therefore no longer a social add-on but a mandatory part of every reach strategy for the feed. Anyone who wants to check the mechanics behind this finds the operational basis on the Facts.

Transparency

Sources used

• Own experience and learnings from six years of Google Discover work for publishers.
• Google Search Central: "Get on Discover".
• NewzDash: Discover share analysis Q4 2025 across more than 400 news publishers worldwide, via Search Engine Roundtable, 23 December 2025.
• Marfeel Discover Monitoring: analysis of AI summary tiles and video host clicks.