Google Discover SEO in 2026: How to Earn Feed Traffic With No Keyword to Target

Administrator in Search Engine Optimization Jul 29 / 3 weeks ago

On February 5, 2026, Google started rolling out the February 2026 Discover Update. It ran for 21 days and 17 hours and closed on February 27. Google has named dozens of ranking events over the years. This was the first one aimed at Discover alone. Normal search results were not the target. The feed was.

Say you run content for a food publisher in Lyon with nine writers on staff. Discover sends 55 percent of your monthly sessions. Display revenue sits near 12,000 EUR a month. In the first week of February, the feed goes quiet. Sessions fall by two thirds inside ten days. Nothing on your site changed. No manual action landed. Your normal search rankings look healthy. You have no query to check, no position to track, and nothing obvious to fix.

That gap is the whole problem with Google Discover SEO. Every other channel hands you a lever. This one hides them. This guide shows you the levers that do exist, tells you plainly who should walk away from the feed, and lays out what the 2026 rules actually reward.

Here is the short version

This guide is built for people who need a decision, not another checklist. By the end you will know whether Discover deserves your time, what to change this week, and how to tell a real Discover problem from ordinary feed noise. Most Google Discover SEO advice skips that first question entirely and jumps straight to image sizes.

Four ideas here run against common advice. First, Discover traffic should never sit in a forecast as a baseline, because Google itself lists three reasons it moves that have nothing to do with you. Second, the famous 1200 pixel image rule is a gate, not a lever, and most decent sites already pass it. Third, chasing the feed with sheer volume is now a real risk under the 2026 spam policies, not a growth hack. Fourth, almost every Discover guide online was written by news publishers, and it quietly misleads B2B and SaaS teams who will never get picked.

Your likely objection is fair. If there is no keyword, can any of this be controlled? Partly. You control eligibility, topic focus, publishing rhythm, source clarity, and headline honesty. You do not control demand. Good Google Discover SEO means working hard on the first list and refusing to promise results on the second.

Scope note. This covers the feed in the Google app and on Chrome mobile, written for sites serving the United States and Europe. We touch AI summaries and zero click pressure only where they change a Discover decision.

What is Google Discover SEO, and why does the usual playbook break?

Google Discover SEO is the work of earning placement in Google's interest feed, the one people see in the Google app and on the Chrome mobile new tab. Nobody types a query first. Google guesses what a person may want, then serves it. That single fact breaks keyword targeting, rank tracking, and every forecast built on search volume.

Normal search is a pull. A reader wants something, types it, and picks from a list. You can study that want. You can map it, group it, and write straight at it. Our breakdown of how search intent splits into micro intents exists because that map is real and useful.

Discover is a push. The reader wants nothing in particular. They open an app while waiting for coffee in Chicago or on a train into Amsterdam. Google fills the screen. Your page either fits that person at that moment or it does not.

So the levers change shape. You cannot rank for a phrase. You can only make your page an obvious match for a subject that a large group of people already care about. That is a different craft, and most advice online blurs the two together.

How does Discover choose pages when nobody types a query?

Google states that content becomes eligible for Discover once it is indexed and meets the Discover content policies. No special tags or structured data are required. Discover then uses many of the same systems as Search to judge helpful, people first content, and matches that content against signals about each reader's interests.

Those interest signals come from the person, not from you. Google says it matches content to interests based on a person's Web and App Activity. In practice that covers what they search, watch, and open across Google's surfaces. Google describes the result as a feed of things new to you, not only things new to the web.

Your side of the match is simpler than it sounds. Be indexed. Be clearly about a subject. Be recognizable as a source for that subject. Recognition builds when you cover the same ground again and again, which is why building topical authority through content clusters pays off more here than any single strong page. Being a known name helps too, and brand mentions across the wider web feed exactly that signal.

Search and Discover, side by side

The table below compares the two surfaces using Google's own documentation for each, plus what the Search Console reports actually expose. It is a practitioner comparison, not a claim about internal ranking systems.

Who should chase Discover traffic, and who should skip it?

Discover rewards topics that big consumer audiences follow out of habit or pleasure. News, sport, food, travel, deals, health, personal finance, and consumer tech all do well. Narrow business subjects almost never get picked. If your buyer is a procurement lead at a mid sized manufacturer, the feed will likely ignore you forever, and that is fine.

Here is the part most guides skip. Discover advice is written by publishers, for publishers. A newsroom shipping forty stories a day is playing a volume game. A B2B blog shipping four posts a month has no raw material for that game.

Imagine a Berlin SaaS company selling warehouse software. Its blog covers pick rates, barcode standards, and stock accuracy. No consumer follows that for fun on a Sunday morning. Six months of image work and headline tweaks would return close to nothing. That team gets far more from a deliberate content distribution strategy and from winning the zero click surfaces it already appears in.

Test it with one question. Would a stranger tap this on a bus with no reason to? If the honest answer is no, Google Discover SEO is not your growth channel. Say so early and spend the hours somewhere better.

What did the February 2026 Discover Update change for publishers?

Google's Search Status Dashboard lists a February 2026 Discover update that began on February 5, 2026 and finished on February 27, 2026 after 21 days and 17 hours. It is the only Discover specific ranking event Google has ever confirmed by name. Google published no change list, so read the reported effects with care.

What Google confirmed is the date range and the surface. Everything else comes from third party tracking and publisher reports. Treat that as evidence, not as fact.

The reports do point one way. Trackers such as NewzDash recorded unique domains in the United States Discover top 1,000 falling from 172 to 158 after the rollout, roughly 8 percent. Publishers in Google's own Search Central threads described sharp losses, and a handful described sharp gains. Several analysts read the update as a push toward stronger, clearer sources and away from thin clickbait style posts. That reading fits Google's published Discover policies, which is why it holds up better than most update folklore.

It also landed before two confirmed events that hit many of the same sites. The May 2026 core update ran from May 21 to June 2. The June 2026 spam update ran from June 24 to June 26. If your feed traffic moved this year, check all three windows before you blame your own work.

Is a 1200 pixel image enough to get into Discover?

No. A large image is the entry fee, never the reason Google picks you. Google asks for images at least 1200 pixels wide, above 300,000 total pixels, ideally in a 16 by 9 shape, and switched on by the large image preview setting in your robots meta tag or by AMP. Miss that and you stay invisible. Hit it and you are merely eligible.

Google is blunt about the rest. It tells publishers to avoid generic images such as a site logo, and to avoid text heavy images. It also warns against preview content that promises details the page never delivers.

A solid setup takes about twenty minutes per template:

• Turn on the large image preview value, then confirm it renders with a free meta tag analyzer.
• Set the hero to 1200 by 675 pixels, a clean 16 by 9, with a free image resizer so the feed never trims your subject.
• Keep your social preview tags tidy while you are in there. Discover does not read the Open Graph image property, but an Open Graph tag generator stops your links looking broken when readers share them.
• Squeeze the file with an image compressor that keeps quality, because a heavy hero drags down your Core Web Vitals scores.
• Apply the wider rules in our guide to optimizing images for web performance and SEO.

Do this once at template level. Then stop thinking about images, because they have stopped being your bottleneck.

Which signals move Discover more than backlinks do?

Timing, topical consistency, and honest headlines move the feed far more than link counts. Discover is not ranking your page against ten rivals for one phrase. It is deciding whether one person will care right now. A story that lands the day interest rises will beat a better story published a week late.

Four things earn your effort here:

• Timing. Publish while interest climbs, not after it peaks. Hours matter on hot topics.
• Rhythm. Cover the same subjects on a steady schedule. A working content calendar does more for Google Discover SEO than any clever tweak.
• Clear sourcing. Real bylines, visible dates, an author who exists. Our guide to building E-E-A-T signals covers this properly, so we will not repeat it here.
• Honest titles. Curiosity is allowed and welcome. A promise your page fails to keep is a policy problem, not a style choice.

Now notice what is missing from that list. Link building barely registers as a Discover lever. A third party score like domain authority is handy shorthand for competitive research, and Google does not use it anywhere, least of all in the feed. Teams who pour budget into links hoping for Discover pickup are solving the wrong problem with the most expensive tool they own. Google Discover SEO rewards patience and topic focus, not spend.

Where do you find Discover data inside Search Console?

Search Console shows a separate Discover performance report once your site passes a minimum impression level. It reports clicks, impressions, and click through rate. It reports no position, because no position exists. You can filter by date, page, country, and Discover appearance type such as Web Stories, videos, or News Showcase.

A few details save hours of confusion later:

• An impression counts once per result per session, and only when the item scrolls into view.
• Data groups by canonical URL, not by the URL a reader actually landed on.
• The table caps at 1,000 rows, and the default window shows the last three months.
• No report at all almost always means too few impressions rather than a bug. Google does not publish the exact threshold. A figure near a thousand impressions circulates in SEO communities. Nobody has confirmed it, and Google has not published one.

Analytics is messier. Discover visits arrive under the organic Google source in GA4, mixed in with normal search. Split them out with a landing page and device view, using the same discipline we apply to tracking AI referral traffic in GA4. Google has also begun a generative AI performance report for Discover inside Search Console, rolled out so far to a subset of site owners. Sound Google Discover SEO starts with reading these two reports weekly and ignoring day to day wobble.

Why does Discover traffic vanish overnight?

Because the feed owes you nothing and never pretended otherwise. Google names three causes of Discover swings. User interests change over time. Discover adjusts the content types it favors. Search updates spill over, since Discover is an extension of Search. Not one of those three is something you did wrong.

This is the most useful idea in Google Discover SEO, and the one teams resist hardest. Treat the feed as a bonus channel. Never treat it as a baseline.

Two rules follow from that. First, never build payroll or a media budget on Discover sessions. A site earning, say, 18 USD per thousand sessions from a feed run is one algorithm change away from zero. Second, measure the floor rather than the peak. Take your worst four weeks of the last year and plan against that number.

A composite makes it concrete. Imagine a United States travel blog with 400,000 monthly sessions, 70 percent of them from the feed. On that strength it hires two full time writers. Six months later, the feed cools by half. The lost traffic hurts. The salary commitment hurts more, and it lasts longer.

Spikes do carry real value, but only when you convert them fast. Capture emails, push follows, and route those readers onto a channel you own before the feed moves on.

Can chasing Discover traffic get your site punished?

Yes, and this risk grew sharply in 2026. Discover runs its own content policies, and breaking them can pull your site out of the feed entirely. Google can also issue a Discover manual action, which shows up in Search Console. Search spam policies now cover scaled content abuse as well, and Google applies them to older pages.

The rules people trip over are unglamorous:

• Misleading previews. Google prohibits preview content that pulls readers in by promising details the page does not contain.
• Ads over content. Paid and promotional material must not exceed your actual content.
• Transparency. News style sites need clear dates, real bylines, company details, and a way to make contact.

Then came the 2026 shift. On May 15, 2026, Google rewrote the opening definition in its spam policies to include attempts to manipulate generative AI responses. Scaled content abuse now applies to AI answers as well as to blue links, and enforcement reaches back to pages published earlier. The June 2026 spam update, which ran from June 24 to June 26, was the first big sweep after that rewrite.

Why does this land hardest on the feed? Because Discover has always pulled publishers toward volume and headline games. The fastest route into Discover and the fastest route into a policy problem look nearly identical from the outside. That is an uncomfortable thing to say in a guide about winning Discover traffic, and it is true.

How do AI summaries in Discover change the math?

Google has been adding AI generated headlines and summaries to Discover through 2025 and 2026. Search Console now carries a generative AI performance report for Discover in limited rollout, which tells you Google treats the surface as measurable and permanent. For publishers, it means a reader can take the gist without ever tapping through.

This is the same squeeze that reshaped search results. We have already covered how Google AI Mode works in depth, so treat this as the feed specific version. It has one extra sting. Inside Discover, several sources sometimes sit grouped under a single summary, so one click gets shared across the group.

Reported figures on how much of the feed now carries AI summaries vary widely, and none of them come from Google. We will not put a number on it. Three things do hold up:

• Your headline and image now compete with a summary that already answered the reader.
• Pages carrying something a summary cannot carry still earn taps. Original data, a working tool, a real opinion, a strong voice.
• Name recognition in the source line matters more than it did two years ago.

What does a 90 day Google Discover SEO plan look like?

Ninety days is enough to fix eligibility, prove whether the feed will take you, and decide honestly if it deserves more. Spend two weeks on technical readiness, one month on topic focus and cadence, and the last six weeks reading the data. Then cut the project without drama if Discover says no.

Days 1 to 14: make yourself eligible

Confirm your pages are indexed. Switch on the large image preview value across every template. Fix hero images at 1200 pixels wide in a 16 by 9 crop. Check bylines, dates, and an about page a stranger would believe. Six to ten hours of work in total, most of it one time.

Days 15 to 45: pick three topics and hold them

Choose three subjects you can cover every single week without straining. Publish on a fixed rhythm. Refresh your strongest older posts too, because Discover pulls evergreen pieces back into the feed when a topic warms up again. Our work on content refresh, pruning, and decay maps straight onto this.

Days 46 to 90: read the feed's answer

Open the Discover report weekly. Study which topics got picked, not which single post won. If no report appears at all by day 90, you have your answer. The feed is not coming, and those hours belong elsewhere. Say it plainly in the review meeting instead of letting the project drift for another year.

How do United States and European teams get different results from Discover?

Geography carries more weight after February 2026. Trackers measured international publishers losing share inside United States feeds, from 8.52 percent to 7.04 percent, alongside city level personalization within the United States. Google also shipped Search profiles to the United States first, and Europe is still waiting. European teams also carry more language splits to manage.

Four practical differences worth planning around:

• Language. One English feed does not serve Germany, France, Spain, and Poland. Local language versions compete in their own feeds, against local publishers who know the ground better.
• Feature lag. Google launched Search profiles in the United States in June 2026, giving publishers a claimable page that people can follow, which routes their content into those followers' feeds. Google says it plans to expand without naming a date. Specific follower thresholds circulate widely in the trade press, and Google's own announcement only says a sizable following on a major platform. Plan against the version Google published.
• Revenue shape. A feed spike worth 20 USD per thousand sessions in the United States may be worth under 10 EUR per thousand in parts of Europe. Same traffic, very different value.
• Measurement. Consent choices under European rules thin your analytics sample, so lean on the Search Console Discover report with a country filter rather than on GA4 alone.

Split your Google Discover SEO reporting by country from the very first week. One global number hides exactly the differences that decide where your next hire or budget should go.

Frequently asked questions about Google Discover

How long does it take to get into Google Discover?

There is no set timeline, and anyone quoting one is guessing. Some sites appear within days of publishing a timely post. Others publish for a year and never surface. Google Discover SEO works on eligibility plus interest match, so the honest answer is that you control eligibility and Google controls the rest. Give it a full 90 days with proper images, real bylines, and steady publishing on three focused topics. If the Discover report still has not appeared by then, take that as your answer.

Do I need Google News approval or Publisher Center to appear in Discover?

No. Google states clearly that content is automatically eligible for Discover once it is indexed and meets the Discover content policies. No special tags, no structured data, and no Publisher Center submission are required. Publisher Center helps you manage how your publication appears in Google News, which is a separate product. Plenty of sites that have never touched Publisher Center get strong feed traffic. Spend that effort on your images, your topic focus, and your source transparency instead.

Does AMP still help with Discover in 2026?

AMP is not required. Google's documentation lists two ways to enable large image previews in Discover. One is the large image preview setting in your robots meta tag. The other is AMP. Most modern sites use the first option and skip AMP entirely, since AMP carries real maintenance cost and no ranking advantage. If you already run AMP pages, they remain valid for the feed. If you do not, there is no good reason to build them now purely for Discover.

Why is my Discover report missing in Search Console?

Almost always because your site has too few Discover impressions to cross Google's reporting threshold. Google does not publish that number. A figure near a thousand impressions circulates in SEO communities, but nobody has confirmed it. A missing report is not a bug and not a penalty. It simply means Google has not shown your pages in the feed often enough to build a report. Keep publishing on focused topics and check again monthly.

Does Discover use backlinks or Domain Authority?

Domain Authority is a third party metric from a commercial SEO tool. Google does not use it anywhere, in search or in the feed. Backlinks do carry weight in Google Search, and Discover shares many of the same underlying systems. But in practice, links move the feed far less than timing, topic consistency, image quality, and source clarity. A well linked site that publishes irregularly on scattered topics will lose to a smaller site that owns three subjects and publishes every week.

Can evergreen content get Discover traffic, or only news?

Evergreen content absolutely can. Discover shows people things that are new to them, which is not the same as new to the web. A guide published fourteen months ago can resurface when its topic warms up again, especially around seasons, events, or news cycles. This is why refreshing strong older posts matters more for the feed than most people expect. That said, timely content still gets picked far more often, so evergreen sites should expect slower and lumpier feed traffic.

What image size actually works best for Discover?

Start with Google's published specification. Images should be at least 1200 pixels wide, carry more than 300,000 total pixels, and work well in a 16 by 9 crop. Large image previews must also be enabled through your robots meta tag or through AMP. Beyond the specification, avoid two things Google names directly. Skip generic images like your site logo, and skip text heavy graphics. Compress the file properly too, since an oversized hero image damages page speed and hurts real readers.

Does posting more often increase Discover pickup?

Consistency helps. Raw volume, on its own, does not, and in 2026 it carries genuine risk. Google's spam policies now cover scaled content abuse explicitly, and the June 2026 spam update enforced them. Publishing thirty thin posts a week to fish for feed traffic is exactly the pattern those policies describe. A steady rhythm across three well covered topics beats an unfocused flood every time. Aim for the cadence you can hold for a year without dropping quality.

Can I recover after losing all my Discover traffic?

Often yes, though the path is slower and less clear than a search recovery. First check for a Discover manual action in Search Console, since that requires a specific fix. If none exists, compare your drop against Google's confirmed update windows. Then audit your headlines against the misleading preview policy, your ad density against the ads policy, and your bylines against the transparency policy. After that, keep publishing on your core topics. Many recoveries arrive quietly, without any announcement, weeks later.

Is Discover traffic worth counting in an SEO forecast?

Only as a floor, never as a trend line. Google itself lists three causes of Discover swings that sit entirely outside your control. Building a revenue forecast on feed sessions is the most common expensive mistake in Google Discover SEO. Model your worst four weeks of the past year as the expected level, and treat anything above it as upside. Report Discover separately from search traffic in every board deck so nobody mistakes a feed spike for durable growth.

The one thing to do this week

Back to that Lyon food publisher. The right move was not a redesign or a link campaign. It was three checks in one afternoon. Look for a Discover manual action. Compare the drop against Google's confirmed February window. Audit twenty recent headlines against the misleading preview policy. Then keep publishing on the same subjects and stop refreshing the dashboard hourly.

If you take one action from this guide, make it this. Open Search Console, find your Discover report, and write down your worst four week period from the past year. That number is your real baseline. Everything above it is a gift, not a plan. Honest Google Discover SEO planning starts there, not with a checklist of tags.

Here is a prediction worth holding loosely. As AI summaries spread across the feed, the gap between recognized sources and anonymous content sites will widen fast. Brand will do the work that clever headlines used to do. Publishers who build a name people follow on purpose will keep their feed traffic. The rest will keep riding a wave they cannot see coming.

So what does your Discover floor actually look like, and would your team still be fine if the feed went quiet tomorrow?

Sources and further reading: Google Search Central, Get on Discover, the Google Search Status Dashboard ranking update history, and the Google Discover content policies.