- Google Discover is still one of the fastest ways publishers can get millions of extra clicks, but it is far less forgiving than it used to be.
- The recent Discover and core updates reward local, trustworthy, topic-focused sites and quietly punish clickbait, thin AI content, and scattered topic strategies.
- If you are a small or mid-size publisher, you probably have an invisible ceiling in Discover until you raise your authority, improve monetization, and narrow your focus.
- Most publishers will get more results from cleaning up content, tracking a tight set of keywords, and testing side domains than from pumping out more AI articles.
Let me answer the core question fast: if you want to grow with Google Discover and survive the latest updates, you need to stick to clear topics, avoid cheap click tricks, publish real perspectives on live conversations, and treat Discover as a volatile bonus channel, not the foundation of your business.
From there, the game shifts into two parallel tracks: one is how you earn and keep visibility in Discover, the other is how you turn that visibility into actual revenue without betting your whole business on one traffic source.
I will walk through what is really happening inside Discover, how the recent updates changed the rules, what is realistic for different sizes of publishers, and where AI fits without putting your domain into a kind of long-term algorithm jail.
What Google Discover actually is today (and why it feels so random)
Many people talk about Discover like it is just “Google traffic,” but it behaves in a different way from normal search and that confusion causes a lot of mistakes.
If you open the Google app or Chrome on your phone and you do not type anything, those recommended cards based on your past behavior, location, and interests are Discover.
Why Discover traffic feels addictive but unstable
Discover can send a site 3 to 4 million clicks in a single day, and some very large publishers see 20 to 30 million clicks per day from it, which makes search traffic sometimes look tiny beside it.
The catch is that Discover behaves more like a social feed than a search result; it reacts to short feedback loops and can flip from viral to invisible in days.
Think of Discover as Google’s version of a social feed that sits on top of your existing authority, not as a replacement for search traffic.
Discover uses its own crawling and ranking systems, and over the last few years Google has moved from trusting domains in bulk to judging articles much more individually.
That shift is why some publishers saw the whole domain dropped out of Discover after a run of low trust posts, while other sites only lost visibility for specific sections or topics.
The February Discover update in plain language
The recent Discover-specific update did two clear things: pushed feeds to be more local and tried to dial down obvious clickbait in images and titles.
If you are in Berlin, Google wants your feed to lean more toward Berlin and Germany, not just global tech or celebrity stories, and that favors publishers who actually cover local or regional topics with some depth.
On clickbait, the main targets were those exaggerated thumbnails and titles that promise drama or shock while the article itself is mild or off-topic.
Despite the update, many people still report odd AI images and “you will not believe what happened next” type cards in their feed, which suggests the rollout reduced some extremes but did not clean the whole thing up.
The real change is less about one update and more about a trend: Discover is slowly preferring local, consistent, first-hand coverage over generic, global, recycled content.
How big is Discover really, and who should care
Discover reaches billions of users monthly, but that does not mean every site can treat it like a reliable growth channel.
If your business model is content publishing, and ads or affiliate deals pay your bills, then Discover is worth fighting for because those sudden traffic spikes translate into real money.
If you run ecommerce or SaaS, Discover can still help, but the value per visit is usually far lower because most Discover users are not in a buying mindset when they tap into an article.
Some publishers tie 70 to 80 percent of their revenue to Discover, which looks great in a screenshot but is a fragile place to be if you care about the next 5 to 10 years.
Why losing Discover can hurt for years
There are plenty of cases where respected news brands fell out of Discover in 2022 and still have not come back, even after multiple core updates.
Once a site crosses whatever internal “trust boundary” Discover uses, it can be very slow to recover, because Google has so many alternative sources to fill the feed.
If Discover is more than 40 or 50 percent of your traffic, you are one update away from a business problem, not just an SEO problem.
Why sites get hit in Discover: topics, breadth, and trust
Most Discover drops people talk about publicly focus on big traffic swings, but underneath those charts there are usually deeper issues with topic focus and trust.
The danger of going too broad, too fast
Google tends to reward publishers that stay close to what the audience expects from them; when a site suddenly starts publishing in random directions, risk goes up fast.
If your site is known for SEO and paid media, then loading it with celebrity gossip, finance news, or health tips just because those topics are trending can backfire hard.
From a user point of view, it is jarring; people did not follow you for that, and from Google’s view it looks like you are stretching to chase any topic that might pop in Discover.
That combo of audience confusion and algorithmic suspicion is often what pushes a site out of the feed, even if the writing itself looks fine.
Expanding topics without blowing up Discover
Topic expansion is possible, but it needs to match reality in the rest of your marketing, not just in your CMS.
If a fashion brand sells only womenswear for five years and then adds menswear, Google is used to seeing everything online about that brand framed as women focused.
To make the menswear expansion stick, you would need to show that shift across the web: campaigns, social, PR, product pages, and maybe even how people talk about you in reviews or press.
Without that broader proof, Discover tends to treat those new topic clusters as off-brand or thin, and that will limit or kill the exposure you hoped for.
Clickbait and misleading packaging
Two kinds of clickbait stand out in Discover: images that lean on exaggerated faces, fake drama, or irrelevant visuals, and titles that hint at some shocking twist that the article never delivers.
For example, headlines like “She walked into the room and everyone gasped” on an article that is just a mild industry piece, or thumbnails that show people who are not mentioned in the story just to get a reaction.
Those tricks can work in the short term because Discover tests your content with small audiences and reacts quickly to taps, but Google is now using larger behavior windows and better models to detect that mismatch.
Once a domain crosses a line of repeated disappointment, the system can pull back visibility for large parts of that site, not only for the worst offenders.
Authority caps and invisible ceilings
One pattern I keep seeing is what looks like a soft cap on Discover impressions for small and mid-size publishers.
A site with a small team might repeatedly peak at something like 200k or 300k Discover impressions per month, then hit a wall even when they publish more content and see short spikes.
In practice, that feels like Google letting them play inside a certain range until their overall authority, brand, and trust moves up.
The only reliable way to break past that limit usually looks boring: more real PR, more strong links, clearer brand signals, and better coverage depth on a limited set of topics.
Discover vs news: automatic inclusion and different systems
Discover and Google News share some tech, but they are not the same thing and mixing them up leads to bad strategy calls.
News inclusion has its own signals, feeds, and even separate reporting segments in Search Console, while Discover is a personalized recommendation engine that may feature both newsy posts and evergreen explainers.
In practice, a site can be pulled into Discover without formal News inclusion, and an approved News site can see very weak Discover performance if it writes in a style that does not earn taps or repeat engagement.
If you are planning content, you need to be honest about which one you are actually targeting, because short deadline-driven news may not age well in search, and slow evergreen pieces may not ride fast social cycles in Discover.
Discover best practices that actually work in 2026
There is no perfect checklist for Discover, and anyone who claims they have one is guessing, but there are patterns that keep paying off for publishers that stay visible through multiple updates.
1. Stay inside the conversations your audience expects from you
The strongest Discover hits usually sit at the intersection of your existing authority and a live online conversation.
That means you watch X, Reddit, YouTube, TikTok, LinkedIn, or niche forums and look for topics in your space that suddenly spike in mentions and reactions.
When that happens, do not just embed a tweet and add two lines; write a short piece that explains what is going on, what your take is, and what this means for your readers.
Content that does well here is often 800 to 2,500 words, mixes context with your view, and feels like someone in your team is actually paying attention, not reacting with a template.
Examples of Discover-friendly angles
I will make up some examples that are close to what works but are not copies of anyone else’s work.
- A cybersecurity site writes “Why that new password leak is less scary than it looks (and where you still need to worry).”
- A parenting publisher posts “The viral ‘no-screens morning routine’ everyone is sharing, tested with my three kids for a week.”
- A local business blog covers “What the new rental rules really change for tenants in Austin, broken down by income level.”
Each of these uses a conversation that already has momentum and adds a grounded, first-hand angle that Google can see as more than a rewrite.
2. Write for curiosity, not tricks
Titles and images still matter a lot in Discover, but the bar is shifting from shock to relevance and clarity.
Instead of hinting at some secret you never reveal, state a clear hook and then actually follow through.
| Bad Discover title pattern | Better Discover title pattern |
|---|---|
| “You will not believe what this teacher did after class…” | “A high school teacher tried a 4-day schedule. Here is what changed for students.” |
| “This one simple trick will change your tax bill forever” | “Three legal ways freelancers are cutting this year’s tax bill by 8-15%” |
| “He said two words and the whole room froze” | “The two questions a negotiator used to settle a 9-month contract fight” |
The second versions are still interesting, but they respect the reader more and match what you can actually deliver in the article.
3. Match content depth to topic speed
Not every trending topic needs a 3,000-word guide, and not every update can be covered well in 300 words, so picking the right depth matters more than people admit.
If a story has a fast, shallow cycle, like a product launch or a minor celebrity incident, a short, clear piece with one or two sharp points can be enough for Discover.
If the topic affects policy, money, health, or behavior for months or years, longer explainers with timelines, data, and expert quotes tend to stand out and fare better with repeat engagement.
Google seems to reward both quick reactions and deep dives, but it is far less patient with lazy rehashes that only restate what people have already read in the first two sources they clicked.
4. Get the basic technical setup right, then stop obsessing
Technical checks help you qualify, but they do not guarantee anything, and I think some SEOs get lost in the checklist and forget the content part.
For Discover, you want at least these items sorted:
- Featured images at least 1,200 pixels wide with clear, relevant visuals.
- RSS feeds set up correctly so Google can ingest new posts quickly.
- A working XML sitemap and, if you are a news publisher, a news sitemap too.
- Pages that are fast enough on mobile and free of intrusive interstitials.
I have seen sites with messy tech still get Discover hits and sites with perfect tech get nothing, so I would treat this as hygiene, not the growth engine.
5. Accept that Discover is random at the page level
One of the strangest realities is that a 350-word post can suddenly score a million Discover clicks while a polished 2,500-word piece on a similar topic never moves.
My best guess is that Discover runs small tests where it shows a new article to a small number of people; if taps and dwell time are high in that tiny sample, it doubles down and the thing can just take off.
If early reactions underperform, that same article may stay in normal search and social, but barely touch Discover, even if it is better in every traditional SEO sense.
So you cannot judge your whole strategy by one viral post or one flop; you need to look at patterns over months and across many pieces.
6. Track your Discover funnel differently
Because Discover is high volume and low intent, it should not be measured in the same way as bottom-of-funnel search content.
In analytics, it helps to treat Discover as a separate top-of-funnel source and then track how many of those visitors come back through search or direct visits later.
If you run email or community plays, consider offering a soft, low-friction call to action on Discover-heavy pages, like “get a weekly recap of stories like this,” not “book a demo now.”
For many publishers, the real value of Discover is attention that can be turned into recurring relationships, not immediate sales.
AI content, spam updates, and what changed in March
Recent spam and core updates hit at almost the same time, which created a lot of noise, but there are some clear threads in how Google treated AI-generated content and large content libraries.
What Google actually says about AI content
Google’s public stance is fairly simple: AI content is allowed, spam is not, and scaled pattern abuse is the main problem, not the tool itself.
The trouble is that many teams heard “AI is allowed” and quietly ignored the second part, which is where the penalties and deindexing often show up.
What I keep seeing in logs and reports is not Google punishing one or two AI-assisted guides; it is Google going after entire clusters of near-identical AI pages that exist only to cover as many long-tail phrases as possible.
Scaled content abuse vs smart AI support
There is a big difference between using AI to help with drafts or research and setting up a pipeline that spits out thousands of generic pages a month.
Some venture-funded companies with strong brands are publishing hundreds of AI-heavy posts per week across dozens of bottom-of-funnel terms.
Right now, many of them still rank because they have huge authority from PR, links, and offline marketing; that does not mean the tactic is safe for everyone else.
If your domain does not already have serious authority, copying the AI scale of a unicorn brand is more likely to damage your site than to lift it.
Smaller companies that tried to copy this pattern often enjoyed two or three months of growth and then watched entire folders get deindexed once the spam update rolled through.
How the spam update played out on content-heavy sites
During the March spam update, a common symptom was a sharp drop in indexed pages over a very short window, especially for sites that had been pumping out AI content on autopilot.
If you compare coverage reports in Search Console before and after, many of the pages that disappeared were new AI posts from the last six to twelve months.
Clicks fell in parallel because there were simply fewer pages left to show; this felt less like a ranking demotion and more like a clean-up of pages Google no longer wanted to carry.
For human-written, slower content, changes were usually gentler: positional shifts, a few winners, a few losers, but not whole sections wiped out overnight.
Why performance marketers struggle with SEO and AI
A lot of this trouble comes from mindset; people used to paid ads often think in terms of “find a winning creative, then scale it as far as budget allows.”
When that logic gets applied to SEO, the “budget” becomes words and URLs, and AI makes it feel cheap to publish thousands of pieces at once.
In paid ads, once a creative works, you increase spend and expect results to scale in a fairly linear way; in organic, authority and trust do not scale like that.
Google can watch how your entire body of work behaves and decide that your sudden explosion of thin AI content is a quality signal, not a quantity win.
How long SEO really takes, and why that matters for risk
One uncomfortable truth is that SEO plays out on multi-year timelines, while many internal targets are quarterly at best.
If you abuse AI content to chase a quarter’s goal and end up burning your domain, you might be stuck in recovery for two or three years, long after the people who pushed for that experiment have left.
That is why I think SEO leaders need to push back when the board or the CMO asks for “X percent more output thanks to AI” without asking where that extra content will come from in terms of ideas and trust.
It is not anti-AI to say no to thin expansion; it is just protecting a channel that your company may rely on much longer than this planning cycle.
Cleaning up: when deleting content is your best move
On some sites, the most effective “SEO project” in the last year has been unpublishing or consolidating large chunks of weak content.
One B2B site grew from around 4,500 organic clicks a month to close to 30,000 mainly by removing or merging old, low-value posts that were cluttering their index.
In other cases, teams moved off-topic content to a separate domain so it did not drag the main site down and could target a different audience or experiment path.
This is not some secret trick; it is just admitting that every indexed page affects how Google judges your site, for better or worse.
Satellite domains and multi-site strategies
Running more than one site can feel heavy, but if you already have the capacity to publish several solid posts per week, it is not as hard as it looks.
Some companies keep the main domain focused on high-intent, brand-aligned topics while using a second domain for experimental content, broader educational pieces, or a different tone.
In AI search specifically, I am seeing cases where a secondary domain already accounts for 5 to 10 percent of non-branded prompts, which is enough to prove the model and ask for more investment.
There are two schools of thought on how to connect these domains in Search Console; some teams keep them fully separate, others manage them together so Google can see the relationship clearly.
Personally, I think if the domains are clearly branded as part of the same company, there is no need to hide them; the key is to give each domain a clear role, not to play a shell game.
Topic authority, AI search, and invented terms
As AI answers take more space on result pages, some brands have tried to invent new category names and hope to “own” them, but this often backfires.
If nobody searches the new term, and you do not have substantial content to back it up, even AI snapshots will struggle to anchor that phrase to you.
In some cases, competitors with better explanation content around that new concept end up ranking and being cited instead of the brand that coined the phrase.
It is usually smarter to build your SEO and content around what people actually search, and then layer your brand language on top rather than forcing the reverse.
| Brand move | SEO-safe version |
|---|---|
| Create a new term “FlowCommerce” and use it everywhere | Target “subscription ecommerce platform” and explain how “FlowCommerce” is your way of doing it |
| Call jeans “Model 914” in URLs and titles | Use URLs like “/slim-fit-jeans-black/” and mention “Model 914” in copy and on-page branding |
| Push “Revenue Cloud Orchestration” as main phrase | Optimize for “multi-channel revenue management software” and introduce your term inside the article |
AI search still leans heavily on the same underlying signals as normal search, so skipping the basics of demand-backed keywords to chase a fresh label is usually a losing bet.
Monetizing attention: ads, affiliate, and the missing SaaS play
Once you get traffic from Discover or search, the real question is how to turn that attention into money in a way that survives algorithm swings.
Why ads and affiliate still dominate publisher revenue
Across most content-heavy sites, two revenue models keep showing up: display ads and affiliate deals.
Discover works quite well for both, mainly because you get huge impression counts and a lot of casual readers who are fine with light ad experiences or product mentions.
In many cases, affiliate beats ads on revenue per visit, with ads acting more like a background layer that pays for the content operation.
But there is a hard limit; you are still renting your income from someone else’s platform, with margins and policies you do not control.
Why most publishers underuse their authority
Many publishers have a strange split: strong domain authority and brand recognition on one side, and very simple monetization on the other.
They run successful sections on topics like personal finance, travel, small business tools, health, or local services, but never move past sending traffic to partners for a small cut.
Some of this is internal politics; editorial and business units are separated, and nobody owns the bigger question of building direct products around that demand.
There is also some fear; launching a product feels riskier than signing another affiliate contract, even if the math favors ownership in the long run.
A quiet but powerful path: publishers launching products
If you already have topical authority, you are in a much better position than a cold SaaS startup that needs to buy every visitor at retail cost.
You can build or “vibe code” simple tools, services, or subscription offers that match your strongest content themes, and then send part of your organic and Discover traffic into those funnels.
Some examples that I think are realistic, but different from the usual ones people cite:
- A home energy blog creates a simple calculator and managed service that helps readers compare local solar installers and negotiates group discounts.
- A careers site that ranks for interview questions offers a low-priced practice platform with timed mock interviews and peer feedback.
- A regional food publisher spins up a paid “locals-only” card that gives discounts at partner restaurants and access to special menus.
Each of these keeps the publisher’s authority in play but moves more revenue into something they own instead of feeding everyone else’s funnel.
Defensive SEO: protecting the traffic and revenue you already have
Once your organic or Discover traffic is driving meaningful money, the hardest work is often not getting bigger; it is avoiding preventable losses.
That starts with something surprisingly rare: tracking your key queries and competitors week after week, not just glancing at total traffic once a month.
If a single page is tied to, say, 80k dollars of monthly revenue, someone in your team should know which competitors are trying to outrank it, what links they build, and how their content evolves.
It also means being willing to push back when new experiments threaten to pollute your strongest sections, even if the idea sounds exciting in a slide deck.
The mindset shift most teams need
Instead of asking “How do we double output with AI this quarter?” a better question is “Which 50 pages actually move money for us, and how do we strengthen and surround them?”
For Discover, the smarter question is not “How do we hit 5 million impressions next month?” but “How do we stay in Discover for years without gaming it so hard that we get kicked out?”
If you get those two mindsets right, AI becomes a helper, not a trap, and Google updates feel like weather you account for, not a random storm that wipes you out.
Long term, the publishers and brands that win are not the ones that publish the most, but the ones that combine clear topics, honest content, smart monetization, and enough patience to think in years instead of quarters.
You do not need to copy your biggest competitor’s tactics beat for beat; in fact, that is usually how you inherit their risks without their safety net.
If you stay focused on what your audience actually wants from you, cut the clickbait, use AI with restraint, and build at least one revenue stream you control, Discover becomes a powerful bonus layer on top of a business that can stand on its own feet.
Need a quick summary of this article? Choose your favorite AI tool below:
