Analytics · September 21, 2026 · 8 min read
Google Discover: How Blog Content Gets Picked Up (and Why It Vanishes)
Discover traffic arrives without warning and leaves the same way — here is what makes a post eligible and which disappearances are worth diagnosing.
By FluxWriter Team
Google Discover is the least controllable traffic source most blogs will ever have, and it rewards a different set of signals than the keyword work you have been doing. Most owners meet it as an unexplained traffic spike, assume something finally clicked, and then watch it disappear inside a week. This covers what makes a post eligible, how to read the Discover report properly, and which disappearances are worth diagnosing.
Why Discover Ignores Your Keyword Work
Discover has no query. That single absence breaks most of the model you built for Search, and it is the reason keyword-first thinking produces nothing here.
In Search, a page is matched against words someone typed. In Discover, a page is matched against a person — their reading history in the Google app, their language and location, and whatever their interest profile is weighted toward this week. Quality signals still apply. They just stop being the thing that decides which of 40 eligible posts fills the slot.
Two consequences follow, and both are uncomfortable. You cannot target a placement the way you target a keyword, and you cannot reproduce one on demand. The same post, republished next quarter, may earn nothing.
There is also no submission form and no opt-in. Eligibility is inferred from your indexed pages, which puts all the useful work upstream — publish something the feed can pick up, then wait.
What Actually Makes a Post Eligible
Eligibility is mechanical, and the checklist is short. Interest matching never looks at your post until the basics are in place.
Indexed. Discover draws from the same index as Search. A page carrying noindex, or one the Page Indexing report shows as discovered but not indexed, is invisible in the feed.
A large image. Google's guidance asks for a content image at least 1,200 pixels wide, served with the max-image-preview:large robots setting so the full-size preview is allowed. Miss that setting and your post competes as a small text card against full-bleed photographs. The major WordPress SEO plugins set it by default. Custom themes and headless front ends often do not, and Search Console will not flag it.
A title that pays off. Discover's content policies treat withheld payoff and manufactured urgency more harshly than Search does. A title promising a revelation the article never delivers is the fastest way out.
Check each requirement against what your site actually ships:
| Requirement | Threshold to hit | Where to verify it |
|---|---|---|
| Indexing | Page reports as indexed | URL Inspection in Search Console |
| Image width | 1,200 pixels or wider | Any crawler's image export |
| Preview permission | max-image-preview:large |
The meta robots column of a crawl |
| Title accuracy | The payoff is in the article | Read the post against its own title |
Rows two and three are where blogs fail most often, and they fail quietly. A site can publish for two years, satisfy every rule but the image one, and never see Discover turn into a channel worth counting.
One image rule sits outside the table. Google asks publishers not to use a site logo as the preview image, so a theme that falls back to your logo on a post with no featured image is producing dead cards. Check what your template serves when that slot is empty.
Why the Traffic Vanishes
Disappearance is the normal state here, not the exception. Three distinct patterns hide behind drops that look identical in a traffic chart, and they deserve completely different responses.
Post-level rotation. One post arrives, peaks within hours, and is gone in 2–4 days. Nothing broke. The feed is built around freshness and leans hard toward the last few days. Older posts resurface sometimes. Treat that as a bonus, not a plan.
Site-level loss. Every post drops on the same date and nothing recovers. That is a different animal. Line the date up against known core update windows, then open the Manual Actions report — News and Discover policy issues are reported there, and they can hold a site out of the feed for months.
Gradual fade. Impressions thin across 6–8 weeks with no cliff anywhere. Usually you changed what you publish and the interest match went with it. Put your last 20 titles in one column and read them as a stranger scrolling would — the drift shows there long before it shows in the chart.
Rotation deserves none of your attention. A same-day, site-wide flatline deserves an afternoon.
How to Read the Discover Report
The Discover report sits under Search Console's Performance section. It only appears once a property has accumulated real Discover traffic. If the entry is missing, you do not have a measurement problem — you have an eligibility problem.
What the report gives you is clicks, impressions and click-through rate, across a window of up to 16 months. What it withholds is position and queries, because neither exists on this surface. That is not a gap in the tooling.
Click-through rate is the number worth watching, and the only honest benchmark is your own. A card carries a large image and no competing results beside it, so Discover CTR tends to sit well above the same site's Search CTR — published figures for the surface swing so hard by niche and country that none is worth planning against. Pull both for the same 28-day window. If Discover is not clearly ahead, the image or the headline is losing the tap.
Then open the Pages tab and export it. Compare the posts that earned impressions against the ones you assumed were your strongest. Where those lists disagree, the feed is telling you which half of your publishing calendar it can use. Keep that export.
Reconciling Discover With Your Analytics
Discover is a mobile surface — a feed, not a results page. It appears in the Google app and the Chrome new-tab feed on Android and iOS, so the sessions it sends are effectively all phone traffic.
That is what makes it identifiable. GA4 files the traffic into the organic bucket alongside Search, with no separate channel, so the device split is the tell — a one-day spike that is almost entirely mobile while your Search clicks stay flat is Discover until something else explains it.
Confirm it instead of assuming. Segment the spike by device and landing page, then check those URLs in the Discover report for the same dates. Pages on both sides settle it inside ten minutes. That is also the last question this surface answers — it will never tell you which reader, which slot, or why.
What to Change and What to Ignore
Start with the image. It is the thing you control that moves the most. Get to 1,200 pixels wide as a floor, then check the crop — the feed frames cards as a wide landscape rectangle, and a portrait shot with its subject near the top loses that subject to the crop.
Then read your titles on a phone, not a monitor. A card gives a headline two or three lines before it clips, and the cut point moves with the device and the font size — which is why a character count is the wrong target. Open your own feed and look. Front-load the specific thing.
Two things deserve less attention than they get. Schema markup is not a Discover lever, whatever a plugin's marketing implies. Publishing volume is not one either — doubling output on a site that fails the image check doubles nothing.
Treat any advice that promises entry as a tell. There is nothing to buy.
Keep the channel in proportion. A Discover visitor arrived mid-scroll with no intent to buy, so those sessions usually convert far below your Search sessions — far enough that a day worth tens of thousands of visits can produce fewer signups than an ordinary Tuesday. How wide that gap runs depends entirely on what you sell, so measure your own. Build the business case on Search. Bank Discover as upside on content you were publishing anyway.
FAQ
Why did my Discover traffic drop to zero overnight?
Check whether every post dropped or only the one that was performing. A single post falling out is routine rotation and needs no fix at all. A site-wide flatline on one date points at a core update or a policy issue, and the Manual Actions report is the first place to look.
Do I need Google News approval to appear in Discover?
No, and the two systems are separate. Discover pulls from the regular index, so any indexed page on any site can surface there. News publisher status affects the News surface and its carousels instead. Plenty of Discover traffic goes to sites that never applied for anything.
Can I make an old post get picked up again?
Rarely, and never on demand. A substantial update can put a post back in front of the freshness signals the feed weights, but the hit rate is low enough that it is a poor reason to schedule work. Treat resurfacing as a side effect of refreshes you were doing anyway.
The Practical Takeaway
Fix eligibility before diagnosing anything. Confirm your last 10 posts are indexed, get every featured image to 1,200 pixels wide or more in a wide crop, and verify the max-image-preview:large setting is present on the template — those three explain most sites that never appear. Then open the Discover report, export the Pages tab across the full 16 months, and separate single-post rotation from a site-wide drop before spending an afternoon on either one. Start with the five posts that earned the most impressions.
If you are publishing at a cadence steady enough for the feed to notice, tools like FluxWriter can help keep image-ready posts going out without a production bottleneck — but whether the crop and the headline earn the tap is your call, and so is how much of your plan Discover deserves.