Every day, millions of users open Google on their smartphones and, before they've even typed anything, a stream of personalised articles appears. This is Google Discover. And behind this apparently simple experience lies a recommendation algorithm remarkably complex. How does Google decide which articles to display? What are the real selection criteria? And above all, how can your business benefit from this? Here's a complete overview.
Google Discover: a very different algorithm from traditional SEO
Before looking to optimise your presence in Google Discover, But there's one fundamental thing you need to understand: you're not dealing with the same engine as Google Search. The rules of the game are different, and the strategies that work in Traditional SEO do not apply directly here.
Discover vs Google Search: two opposing rationales
In Google Search, The user expresses an intention. They type in a query and Google looks for the most relevant pages for that query. This is a model known as jumper The user pulls the information towards him. Google Discover, it's exactly the opposite: a model push. Google anticipates what the user might be interested in - without them having asked for anything - and pushes content at them. This inversion changes everything, including the way you design your articles.
A feed based on interests rather than keywords
There is no target keywords in Discover. The algorithm builds an interest profile for each user based on their browsing history, past searches, location and behaviour in other Google applications (YouTube, Maps, Gmail). Your article will be suggested not because it responds to a specific query, but because its content matches the thematic profile of a user segment. It is a SEO without keywords, or almost. We went into more detail about this change in our article on the transition from SEO to GEO in the age of AI.
Why your content can explode... and then disappear
Le Google Discover traffic is by nature volatile. An article can generate tens of thousands of hits in 48 hours, then fall back to zero. This phenomenon can be explained by a combination of two factors: the freshness of content (Discover favours recent articles) and the emotional resonance (a strong headline, a striking visual, a hot topic). Understanding this dynamic is essential to building an effective editorial strategy.
The 9 steps to selecting an article in Google Discover
How to Google Discover selects articles in concrete terms? The process involves a number of successive technical stages, which can be described as follows content selection pipeline. Here are the 9 key steps.
1. Crawling and indexing content
It all starts with the crawl. Googlebot crawls your site, discovers your new pages and passes them on to the indexing pipeline. An article that is not indexed cannot appear in Discover. Indexing speed depends on the frequency of crawls for your domain, which in turn depends on your site authority and the quality of your technical SEO structure. An up-to-date XML sitemap, correct canonical tags and optimised loading times are the essential prerequisites.
2. Extracting entities via the Knowledge Graph
Once indexed, the page is analysed by Google's entity extraction engine. Via the Knowledge Graph, Google identifies the people, places, organisations, events and concepts present in your content. These entities act as a bridge between your article and users' areas of interest. An article that is well structured around recognised entities is much more likely to be distributed to a relevant audience.
3. Analysis of metadata (title, image, author, language)
At the same time, Google is extracting metadata the page: page title, Open Graph tag, Schema.org structured data, main image, content language, author information. This data is essential for the building the Discover card - the visual format in which your article will be displayed. A strong title and a high-quality image (minimum 1200px wide) are key factors in the click-through rate.
4. Classification into clusters
The article is then classified in one or more thematic clusters. Google has an internal taxonomy of interests (sport, technology, finance, health, politics, etc.) with very detailed sub-categories. The more clearly your content is anchored in a specific theme, the easier it will be to classify - and therefore to distribute to the right users. This is known as’thematic authority, a fundamental concept in the Discover content strategy.
5. Content filtering (site + URL)
Before entering the matching phase, the item goes through a double filtering : at domain level and at URL level. Google applies its content policies filtering: anything that comes under the heading of misinformation, sensitive content, excessive clickbait or deceptive SEO practices can be filtered. This filtering can be manual (via Google Search Console) or algorithmic.
6. Matching with user interests
This is the heart of the system. The engine of customise your Google feed - sometimes referred to as NAIADES in the patents filed by Google - matches the thematic clusters of your article with the interest profiles of users. This matching is real-time and highly personalised: two people with slightly different profiles will not see the same articles, even on the same themes.
7. Ranking (server-side ranking)
Once the matching is complete, the candidate articles enter a phase of ranking. Google classifies them according to several signals: freshness, commitment predicted (estimated CTR, likely reading time), reputation of the source site, thematic diversity of the feed already built up. Actual engagement signals - clicks, shares, saves - also play a role in the Discover ranking over time. This is why an article that performs well continues to be distributed, whereas an article that disappoints quickly disappears.
8. Stream construction and display
The final feed is dynamically assembled in real time each time the Google application is opened. It mixes fresh content and slightly older but highly engaging articles, to provide a balanced experience. The display takes the form of a card containing the title, the image, the name of the site and the estimated reading time. The visual quality of this card is directly linked to the metadata you have implemented.
The key role of structured data and content
The mechanics of Discover are largely based on Google's ability to understanding and qualifying your content. Structured data is your main lever for facilitating this work.
Why Schema.org takes precedence over Open Graph
Many sites simply implement the Open Graph (og:title, og:image, og:description) for social sharing and think this is sufficient. For Google Discover, Schema.org is clearly preferable. The NewsArticle, Article, BlogPosting allow Google to extract the precise author, publication date, main subject and entities mentioned. This structuring directly improves the quality of thematic matching. Open Graph remains useful, but it is Schema.org that drives algorithmic understanding.
The importance of images and metadata
Google Discover is a visual-first. The image is the first thing the user sees, even before the title. Images less than 1200px wide are often excluded from the large card format (which generates more clicks). The image must also be declared using the max-image-preview:large in HTTP headers or via the robots tag, and that it is correctly referenced in the Schema.org or Open Graph markup. A good image is often the difference between an impression and a click.
How Google really understands your content
Beyond the tags, Google evaluates the semantic consistency of your page: do the title, content, metadata and internal links all tell the same story? An article optimised for Discover should have a clear theme, well-defined entities and a content structure that is easy to understand. Google also uses the Knowledge Graph to check whether your domain is associated with a particular theme - what we call topical authority. We explain it in detail in our article on the evolution of search engine optimisation in the age of AI.
Dual filtering to determine your visibility
One of the least documented - but most impactful - mechanisms of Google Discover is its two-level filtering system. Understanding this system is essential to avoid finding yourself excluded from the channel without knowing why.
Domain-level filtering
Google evaluates your domain as a whole even before analysing a specific article. A site that regularly publishes quality content, has a good E-E-A-T reputation (Experience, Expertise, Authority, Reliability) and does not show any negative signals (high bounce rate, user complaints, massive duplicate content) will benefit from a distribution credit higher in Discover. Conversely, a domain that is penalised or has a low reputation will have its articles blocked upstream.
URL filtering
Each URL is also assessed individually. In particular, Google checks compliance with its content policies, the absence of misleading practices (clickbait titles that bear no relation to the content, abusive interstitial pages, intrusive pop-ups), and the quality of the user experience on mobile. Visit Core Web Vitals plays a role here: too high an LCP (Largest Contentful Paint) or a degraded CLS (Cumulative Layout Shift) can penalise the URL in the Discover ranking, even if its content is excellent.
Why a single negative signal can rule you out
Discover filtering is non-linear A sufficiently strong negative signal can be enough to exclude an article or an entire domain, regardless of the quality of the rest. This is why a global approach to site quality - technical, editorial, UX - is essential. Unlike SEO Search, where problems are often compensable, Discover applies a threshold logic: below a certain level of quality, the door is closed.
What Google Discover means for your SEO strategy
The emergence of Discover as a massive distribution channel - it has over a billion active users - means that we need to be more vigilant. in-depth revision of editorial strategies. Le Discover SEO follows a different logic to SEO Search, and the two approaches need to coexist in your content strategy. Google's move towards AI, which we describe in our article on AI in Google Search and the evolution of SEO further amplifies these changes.
Moving from keyword-based SEO to attention-based SEO
In Discover, your competitor is not the article that is positioned on the same query as you. Your competitor is everything in the user's feed at that very moment: a YouTube video, an article in the national media, a viral post. You are entering a war for attention, not a war for positioning. This means that your content must be intrinsically desirable - it must make people want to read it, even if they have no intention of doing so beforehand.
Create content that makes people want to click
Le click-through rate in Discover is one of the most important quality signals. An article that generates a lot of impressions but few clicks will gradually be downgraded. Investing in attractive headlines, well-crafted images and effective hooks is no longer optional: it's at the heart of the Discover content strategy. There's a fine line between an engaging headline and clickbait: Google punishes the latter, but rewards the former.
Focus on emotion, timing and storytelling
The articles that perform best in Discover share three characteristics: they create a emotional resonance (surprise, immediate usefulness, curiosity, identification), they are published in the good time (linked to a hot news item or a seasonal event), and they tell a story rather than simply providing information. The narration is a powerful lever for maintaining attention and triggering sharing - two favourable signals for the algorithm.
How to optimise your articles for Google Discover in 2026
Over and above a theoretical understanding of the algorithm, here are the main points to bear in mind concrete actions to maximise your chances of appearing in Google Discover.
Creating attractive but honest headlines
A good Discover title must combine several qualities: be catchy (it has to stand out in a busy flow), be precise (no vague or exaggerated promises), and be aligned with content (Google and users penalise disappointment). Phrases like «What no one tells you about...», «The method the experts use to...» or «Why X will change in 2026» work well - provided the article delivers what it promises. To help you structure your production, our best SEO automation tools can save precious time.
Use strong visuals (≥1200px)
Your image is your Discover shop window. It must be original (no generic photo stock), high resolution (at least 1200px wide to activate the large card format), relevant (it really illustrates the subject of the article) and declared correctly in the metadata. Avoid images with superimposed text, which is often poorly rendered on mobile. A good visual can double your CTR in Discover.
Publishing at the right time (hot news)
The freshness is a determining factor in Discover. Publishing an article linked to a current event or an emerging trend multiplies the chances of it being distributed. That's why a active thematic monitoring is essential for editorial teams seeking to attract new readers. Discover traffic. Note: evergreen articles can also perform well on Discover if they are regularly updated and republished with a new date. In fact, reactivating old content is one of the most effective ways of getting people to read it. an underestimated source of traffic which we document in a dedicated article.
Building a thematic authority
Discover rewards sites that are clearly experts in a particular field. The more regularly your site publishes on a specific theme, the more credit Google will give it for distributing content on that theme. This means defining content clusters, A generalist site that publishes everything will find it harder to emerge in Discover than a specialist site that publishes less but in a more targeted way. A generalist site that publishes everything will find it harder to emerge in Discover than a specialist site that publishes less but in a more targeted way.
Iterates, your partner for capturing Google Discover traffic
At Iterates, a firm of SEO consulting in Brussels, We help companies who want to go beyond traditional SEO search and build a sustainable digital presence across all Google channels - including Discover. Our approach combines technical expertise, strategic vision and performance-driven content production.
Discover editorial strategy
We can help you define a editorial strategy which incorporates the specific features of Google Discover: choice of high-potential themes, publication calendar linked to current events, tested headline and tagline structures, visual guidelines adapted to the card format. All in line with your brand positioning and business objectives.
SEO optimisation + content + data
Our approach is fundamentally data-driven. We analyse your current performance in Google Discover via Search Console, identify articles that have already performed well and those with untapped potential, and deploy technical optimisations (Schema.org, Core Web Vitals, metadata) and editorial optimisations (rewording titles, improving visuals, semantic enrichment). Every action is tracked and measured.
Creating content with high viral potential
We also produce content with high distribution potential in Discover: topical news articles, analysis, engaging case studies, high value-added service content. Our editorial team has mastered the codes for high-performance content in Discover, and works closely with your teams to ensure consistency of voice and message. As a’SEO agency in Brussels rooted in the realities of the Belgian and European markets, we understand the specific local conditions that influence Discover performance.
Ready to generate traffic with Google Discover?
Google Discover now represents a huge traffic opportunity - and one that is still largely under-exploited by most businesses. Understanding how Google Discover selects articles is the first step. The second is to put in place a concrete strategy to benefit from it.
If you want to build a solid presence in Discover and turn this channel into a regular and measurable source of traffic, Iterates is your partner.
→ Let's discuss your SEO and content strategy with Iterates
