Search results and measurement
Google Discover: interest-based content recommendations
Google Discover is Google’s interest-based content feed. It can recommend articles and other content without the user entering a query at that moment. Appearance is personalized and variable; eligible indexed content does not automatically receive Discover traffic.
Also known as: Discover · Google Discover traffic
At a glance
- Discover recommends content rather than answering a newly entered keyword.
- Indexing and policy compliance support eligibility, not a guarantee of appearance.
- Interpret Discover spikes separately from sustained search performance.
On this page

How does a reader encounter Discover?
Imagine someone recently researching garden irrigation seeing an article about winter draining without searching that morning. Discover acts as an interest-based recommendation. In conventional search, the person would first enter a question and choose a result.
The editorial task differs. A search guide should answer a defined need; a recommended article's title and image also need to explain why it matters now. Neither situation justifies misleading headlines or deliberately withholding the substance.
What makes content eligible?
Google's Discover guidance describes indexed content meeting its policies as potentially eligible. There is no mandatory special Discover markup that switches recommendations on. Older material can appear when relevant to the person.
Helpful content might contribute an original diagram, documented experience, clear comparison or precise answer. Changing an old article's date alone does not make its advice current. Review the actual substance first.
What role does the image play?
Images help people understand feed cards quickly. Google's current recommendations include at least 1,200 pixels in width, more than 300,000 pixels overall and an appropriately composed 16:9 ratio. max-image-preview:large can permit large previews. These are recommendations, not inclusion guarantees.
An article about draining irrigation equipment should show relevant equipment or a clearly useful illustration rather than an unrelated attractive garden. Compression should preserve necessary detail. Image SEO, context and cropping matter more than repeating a keyword in several fields.
Example: two traffic patterns for one guide
Suppose a fictional gardening company's winterization article receives 600 search visits in October and another 1,800 visits associated with Discover during one weekend. These invented values are not expected performance.
Total traffic may fall afterward while query-based search visits remain stable. Calling that automatically a loss of Google rankings would confuse the surfaces.
Separate the traffic sources, then investigate whether the new readers were relevant and which parts of the explanation helped them. A temporary recommendation can be useful without establishing a repeatable monthly result.
How should it be measured?
Search Console can provide a dedicated Discover report where sufficient data and access exist. Its impressions and clicks concern that surface. An absent report is not proof that nobody ever saw the site there.
Website analytics examines activity after arrival, such as opening related guides or contacting the business. The systems measure different events, so exact agreement is not always expected.
What business conclusion is reasonable?
Discover can supplement organic traffic, but a short spike does not establish sustained demand. Improve titles, subject choice and imagery for readers, then observe recommendation traffic separately.
A briefly timely article can still be valuable. Keep the durable customer question distinct from a temporary wave of attention when planning the business.