Working in the search industry means we have all spent a lot of time learning to read Google Search. We know the process: study the queries, map the intent, watch the rankings, adjust.
Despite our trusted process, no one knows exactly what goes into a ranking. However, Search at least gives you something to work with: there’s a query we can see or a result to be measured that can help us define a cause and a subsequent plan of action.
Google Discover doesn’t work that way, and that’s what makes it a bit harder to work with.
There’s no query, so the user never tells Google what they want. Instead, Google decides what to put in front of them, based on what it believes they’re interested in.
What Google means by interests isn’t entirely a mystery: its own documentation says Discover draws on a user’s Web and App Activity to surface content it thinks they’ll care about. What stays hidden is how those signals get weighed for any given page.
Essentially, content is pushed, not pulled.
That single difference upends most of what Search has taught us, because the lever we’re used to pulling (matching a page to a stated intent) doesn’t exist here. Discover is interest-led, and you’re optimizing for a decision Google makes on the user’s behalf.
For a long time, that kept Discover as a somewhat peripheral bonus. It was extra traffic that was nice to have, but not worth building a strategy around since it couldn’t really be tracked or measured. However, that has now changed.
The times are changing: Why this matters more now than it did two years ago
The Search scope is getting smaller. According to Chartbeat data cited in the Reuters Institute’s Journalism, Media and Technology Trends and Predictions 2026 report, Google’s organic traffic to more than 2,500 sites fell by approximately a third globally last year.
AI Overviews and AI Mode are absorbing the informational queries that used to send a click your way. Users now get their answers on the results page and don’t really have a need to click through to the source sites.
As easy Search clicks erode, the spaces that don’t depend on a typed query are becoming more valuable. Discover happens to be one of them.
Publishers seeing steep Search declines reported in a Digiday.com article that Discover referral traffic has held up even as Search has fallen.
It isn’t a complete answer to a shrinking Search channel, but it’s one of the places your audience can still find you, and that makes it worth understanding properly rather than treating it as a happy accident.
The problem is that understanding Discover from the outside is genuinely difficult. Google doesn’t publish a ranking guide for it. So how do you get a read on a channel that won’t explain itself?
There are really two ways, and they work best together.
Understanding Google Discover and how it works
The first way is to listen to the people who work with Discover every day. They have a feel for it that no documentation provides.
Clara Soteras, an SEO consultant for news publishers, put the central challenge plainly when we spoke with her:
“The most difficult thing in Google Discover is to start, to appear, and to have impressions in this channel. For me two of the most important [things to get right] are the headline and the photo.”
It’s actually also a point Guillaume Giraudet, founder of 209 Agency, makes during his interview about Discover. He says you really only have two things working for you: your title and your image.
Hold onto that idea, we’ll come back to why it matters so much that two of the biggest levers in Discover are things a user sees before they ever click.
The second way is to audit your own data. Experience tells you what to look for; your own performance data tells you what’s actually happening on your site. It’s the most reliable window you have into a channel that otherwise keeps its reasoning to itself.
Google Search Console will show you the outcome: Discover clicks, impressions, and CTR. That’s a start, but it stops at the question that matters most: Why do some of your pages appear in Discover while others never do, and why do some that appear get ignored?
What you need to know next is what do the winning pages have in common. For the time being, GSC doesn’t know anything about your pages beyond their performance. It won’t tell you their depth, their internal links, their author, their image, their word count, or their technical health.
That’s the gap the Google Discover Lens is built to close.
What the Google Discover Lens does
The Lens brings your Discover performance data directly into an Oncrawl crawl analysis, so you’re never looking at clicks and impressions in isolation. It sits your Discover impressions, clicks, and CTR alongside everything the crawl already knows about each page: its place in your site structure, its technical signals, its content characteristics, the page group it belongs to, and any custom data you’ve scraped.
That combination changes the question you’re able to ask. Instead of stopping at “How much traffic did we get from Discover?”, you can ask the one that leads somewhere: “Which pages get Discover visibility, which ones actually get clicked, and what makes them different from each other?”
The rest of this article is about answering that question methodically. We’ll look at how to activate the data, how to read it, and how to act on it.
Activating Discover data in Oncrawl
Discover data reaches Oncrawl through the Google Search Console connector, so activating it means enabling the right options before you crawl:
- Open your Oncrawl project.
- Create or edit a crawl.
- Go to Analysis > Connectors.
- Enable Google Search Console.
- Enable Google Discover.
- Launch a new crawl.
- Open the Google Discover Lens once the crawl has finished.
One thing to plan for: Discover data only appears in the Lens after a fresh crawl run with the connector enabled. A crawl you launched before switching the option on won’t show Discover data retroactively, so treat a new crawl as part of the setup rather than an afterthought.
Start with the big picture
Before you start to segment anything, it’s useful to get an honest read on how the whole site is doing. The Lens gives you a set of KPIs that establish that baseline:
- Discover impressions: how often your URLs appeared in Discover
- Discover clicks: how many clicks Discover actually sent you
- Discover CTR: the rate at which those impressions turned into clicks
- Pages featured in Discover: pages that received any impressions
- Active pages: pages that received clicks
- Featured but not clicked pages: pages that appeared in Discover, but didn’t bring in any traffic
That last line is the one to sit with, and it’s where Clara and Guillaume’s points about the headline and the photo come back. A page that’s featured but not clicked is a page Google chose to show. It cleared the hard part, the part Clara called the most difficult thing in Discover. However, the user scrolled past it anyway. Google did its job, but the page didn’t.
That’s not a failure. On the contrary, it’s the most actionable signal in the report. The page is already eligible, already in front of people.
What’s missing is the part the user judges before clicking: the headline, the main image, the angle, how relevant the snippet feels, whether the format and the topic fit the person seeing it.
When you find a group of featured-but-not-clicked pages, you’ve found a useful optimization opportunity by way of a list of pages where a better headline or a stronger image could convert visibility you’ve already earned into traffic.
Segmentation tells a different but equally useful story
KPIs are diagnostic. Segmentation is where you start finding things you can act on, and the most natural place to begin is with your categories or topics, because that’s how editorial teams already think about what they publish.
Looking across categories, watch for the patterns that point somewhere:
- Categories driving the most Discover clicks
- Categories with strong impressions but weak CTR (visibility that isn’t converting)
- Categories that perform better than anticipated: few pages, high performance
- Categories that publish heavily but underperform in Discover
- Topics that show up in Search but not Discover, and the reverse
A finding that comes up often, and that lands hard with editorial teams, is that a single category can represent a small slice of everything you publish while driving a large share of your Discover clicks.
It’s a clear tell, backed by data, as to which topics deserve more of your team’s time because they have real Discover pull.
There’s a reason that pattern shows up so often. Guillaume describes Discover as working by cluster: even a large publisher can’t be strong across economy, sports, politics, and international all at once.
The move that works is to find the cluster you’re already strong in, maximize it, then expand from there once it’s established. So a category that over-performs relative to its size isn’t just interesting, it’s where you double down first.
Avoid assumptions: Comparing Discover against Search
It’s important to identify which pages show up in Search and not Discover and vice versa because that helps avoid assumptions. Considering that Search and Discover are both Google, it’s tempting to assume a page that does well in one will do well in the other. However, the data rarely supports that, and checking it is one of the more clarifying things the Lens lets you do.
You can separate four groups: pages visible only in Search, pages visible only in Discover, pages visible in both, and pages visible in neither. With those in front of you, you can start to look deeper at some useful questions.
Do your Discover-performing pages also rank in Search? Are there pages that perform only in Discover? Are your Search-first and Discover-first strategies actually different and should they be? Which topics show Discover potential even where their Search visibility is thin?
The principle to carry through all of it is the main point we mentioned earlier: Search is query-led, Discover is interest-led. Strong Search performance doesn’t explain Discover performance, and treating one as a proxy for the other will send you optimizing for the wrong signals.
Test what editorial factors actually control
Going back to the insight about the headline-and-photo, this is where it turns into something you can investigate for yourself. The Lens lets you test whether specific content characteristics correlate with Discover performance. Test your intuition about what works and check it against your own data.
Doing this usually means building custom segmentations, and most of them rely on fields you scrape during the crawl. You might segment pages by whether they carry a named author, by title style or punctuation, by the presence of clickbait-style phrasing, by publication date, by paywall status, or by the size and format of the main image. You can lean on default metrics too, like word count and title length.
Once those segments exist, you can compare Discover clicks, impressions, CTR, featured pages, and active pages across each one, and start answering the questions to adjust your editorial strategy:
- Do signed articles outperform unsigned ones?
- Are shorter or longer titles associated with better Discover CTR?
- Do question-based or clickbait-style headlines behave differently?
- Does short-form or long-form content do better?
- How does freshness affect visibility?
- Can paywalled content still earn Discover clicks?
The goal isn’t to crown one factor as the cause of Discover success. It’s to surface patterns reliable enough to guide how you write headlines, choose images, and decide what to publish. Given that the headline and the photo are two of the first things a user judges, this is often where the highest-leverage findings live.
Test what technical SEO controls
Discover performance has a technical dimension as well, and the Lens lets you examine it through signals you already collect. This part still uses segmentation, but most of what you need is available by default. You can often start from an existing segmentation template in the builder rather than building from scratch.
Useful technical cuts include internal linking (pages with more versus fewer inlinks), page depth (shallow versus deep), load time (fast versus slow), indexability (indexable versus not), status code (clean 200s versus redirected, broken, or blocked), and canonical status.
Comparing your Discover metrics across these groups helps you see what technical SEO adjustments could impact your Discover performance.
Throughout the audit, there is one underlying question you should focus on: Are technical or structural patterns limiting your visibility or your clicks, and which ones?
Turn the audit into action
Audits are only as useful as the results they help you achieve: none of it matters unless it changes something. At the end of your audit, the findings can be sorted into four kinds of action.
Improve visibility
For pages or sections that rarely appear in Discover, improve internal linking to high-potential content, review page depth and crawlability, check indexability, canonicals, and status codes, and identify whole templates or page types with low Discover visibility.
Improve engagement
For pages featured but not clicked, this is where you spend your headline-and-image effort: review titles and headline patterns, check the main image and image-preview rules, compare CTR across category, topic, and format, and isolate the pages drawing impressions but converting poorly.
Prioritize content
Use what the audit tells you about your best-performing groups to identify the topics and categories worth more investment.
You can also compare fresh content against evergreen content and short-form against long-form content. Review authorship and paywall patterns, and adjust what you produce accordingly.
Compare channels
Use the audit to understand whether Discover performance follows the same patterns as Google Search. Identify which of your pages are Discover-only.
Find the content that wins in Search but not Discover and adapt your priorities channel by channel rather than assuming one strategy serves both.
A few habits that make this work
Treat Discover analysis as an audit you repeat, not a report you read once. Things can change over time, so it’s important to track progress and evolution.
Use segmentations to test specific hypotheses and avoid drawing conclusions from a single metric. One honest caveat: segmentations reveal patterns and correlations, not proof. They show you where to look and what to test next. They don’t tell you that one factor alone caused a result, and any audit that claims otherwise is overreaching.
Read volume, CTR, and page count together. A category with great CTR and almost no impressions is a different story from one with huge impressions and weak CTR.
Run the analysis again over time to catch movement, and share what you find with both your SEO and editorial teams, because the actions almost always belong to both.
Define your Discover strategy
Discover doesn’t explain itself the way Search does, and waiting for it to isn’t a strategy. What you can do is stop guessing from the outside and start reading from the inside, combining what practitioners have learned with what your own data already knows about your pages. That’s what the Google Discover Lens is for.
You won’t make Discover transparent, only Google can do that. But you can stop treating it as a channel that happens to you, and start treating it as one you can read.
