Short answer: a publisher is ready for Google Discover when every article is technically eligible, visually previewable, clearly authored, specific to a recognisable beat, and measured as a traffic-and-revenue surface rather than as a search ranking. This benchmark turns those checks into a repeatable score. It is a diagnostic, not a promise of distribution.
Google Discover is difficult to manage because it does not behave like ordinary search. There is no query to target, no position to improve, and no keyword report that explains why one story appeared while another did not. That does not make the surface random. It means a newsroom needs a different operating model: remove eligibility failures, publish useful stories with clear context, and judge the trend over a meaningful window.
This checklist is designed for news sites, digital magazines, specialist publications, and ad-monetised editorial teams. It is intentionally small enough to run once a month and specific enough to give an editor, SEO lead, and revenue owner the same conversation.
How to score the benchmark
Score each line from 0 to 2:
| Score | Meaning | Action |
|---|---|---|
| 0 | Missing or cannot be verified | Fix before expanding the content programme |
| 1 | Partly present or inconsistent | Assign an owner and a review date |
| 2 | Present, documented, and checked on live URLs | Keep it in the monthly control set |
There are eight checks, so the maximum is 16. The number is not a forecast of Discover traffic. It is a way to distinguish a missing prerequisite from an editorial experiment. A score can move up while traffic stays flat, and traffic can move while a score stays flat. Record both.
Check one: the article is eligible to be discovered
Start with the page itself. It must be indexable, return a normal success response, and not be excluded by a noindex directive, a blocked resource, or a canonical pointing to another article. A page that is not in Google's index cannot be selected for Discover.
Use the URL Inspection tool for a sample of recent stories, not only the homepage. Verify the canonical URL, indexing state, and the rendered HTML. If the article appears only after a client-side request, fix that before debating headlines. A newsroom cannot optimise a page that a crawler cannot reliably read.
The Google Discover documentation is explicit that there are no special tags required for eligibility. Treat that as a useful constraint: do not spend the first week inventing a Discover-only markup layer while basic indexation is uncertain.
Check two: the preview image can carry the story
Discover is a visual feed. Google recommends large images, and its guidance calls out a minimum width of 1,200 pixels for large previews. The image also needs to represent the story rather than act as a logo, a collage of tiny text, or a generic stock background.
Check the image that the page actually exposes through og:image or structured data. Do not assume the largest file in the media library is the selected preview. Confirm that the URL resolves without a login, the dimensions are sufficient, the subject is clear at a small size, and the crop still makes sense on a phone.
The Google image-preview guidance explains the max-image-preview:large directive. If your site restricts previews with a robots meta tag or header, the restriction is a self-inflicted ceiling. Removing that ceiling does not guarantee a Discover card, but it makes a large preview possible.
Check three: the headline states what happened
Discover headlines should make the story understandable without withholding the essential fact. Avoid formulas that promise a shock, tease an unnamed person, or substitute outrage for information. A headline can still be interesting; it just needs to be specific.
Review the headline and the first sentence together. The reader should be able to tell the subject, place, and change or event. For example, “Council approves 400-home development on green-belt site” gives a reader a reason to care and enough context to decide whether to open it. “What the council just did will change everything” gives neither.
Keep a small sample of headlines each month and mark the pattern, not only the winner. If the editorial team cannot explain why a headline is accurate, it is unlikely to be a durable standard.
Check four: the publication has a recognisable beat
Discover learns from reader interests, so a publication benefits from being legible about what it covers. A finance title can cover markets, regulation, and company results without becoming a general-interest site. A regional newsroom can cover council, transport, and housing while keeping a clear local frame.
Map the last month of articles to a small set of beats. Mark each story as core, adjacent, or off-beat. Then compare the distribution with the publication's stated promise. This is not a request to publish fewer stories for the sake of a tidy chart. It is a way to see when a one-off traffic spike is pulling the editorial signal away from the audience the newsroom is trying to build.
Check five: authorship is a real entity
Every substantial story should have a byline that resolves to a real author page. The author page should state the person's role, the subjects they cover, and link to a body of work. A generic “staff writer” label leaves the clearest expertise signal unused.
Use Person markup only when it matches the visible page. Keep the name, role, and profile URL consistent between the article, author page, newsroom, and social profiles. If a story is written by a desk, name the desk and explain its editorial responsibility rather than implying an individual author who does not exist.
Check six: freshness is measured as a process
Discover is a recent-content surface, so publication and update times should be visible and truthful. Make the date readable to a person, expose it in the article markup, and record corrections when they happen. Do not change a date simply to make an old story appear new.
Measure the time from editorial approval to a crawlable URL. If a newsroom takes a long time to publish a finished story, the bottleneck is operational. If the page is live but not indexed, inspect the sitemap, internal links, server responses, and crawl controls. This check belongs to the SEO and editorial operations owners together.
Check seven: reporting separates surfaces and money
Do not put Discover, Search, direct, and social sessions into one traffic number and call it performance. Report the mix by month and place revenue per thousand pageviews beside it when the publication is ad-monetised. Discover sessions and search sessions can have different engagement and revenue patterns, so a larger total can conceal a weaker business outcome.
The minimum monthly table is:
| Surface | Sessions | Share of sessions | Revenue or value | Notes |
|---|---|---|---|---|
| Discover | Record | Record | Record if available | Rolling trend, not a daily verdict |
| Search | Record | Record | Record if available | Include branded and non-branded where possible |
| Direct | Record | Record | Record if available | Include newsletter and app referrals consistently |
| Social | Record | Record | Record if available | Separate owned from paid when the data allows |
Keep an update log next to the table. Mark major Google changes, template releases, paywall changes, headline policy changes, and unusual news cycles. The log does not explain every movement; it prevents the team from inventing a cause after the fact.
Check eight: the team can act on the result
A benchmark is useful only if someone owns the next move. Assign each zero or one to an editor, SEO lead, developer, audience lead, or revenue owner. Set a review date. Fix one eligibility or measurement failure before launching a large headline experiment.
For a small newsroom, the monthly meeting can be 30 minutes. Review the score, the surface mix, the highest-impact defect, and one editorial test. For a larger publisher, keep the same scorecard but split ownership by desk and template. The score is a coordination device, not a performance rating for individual writers.
Interpreting the total
A score from 0 to 5 usually means the publisher is still removing basic blockers. A score from 6 to 11 means the foundations exist but are inconsistent across templates or desks. A score from 12 to 16 means the publisher has a documented operating layer and can spend more time on beat depth, original reporting, headline patterns, and revenue experiments.
Those bands are management thresholds defined by this checklist, not industry benchmarks. They should not be presented as a probability of Discover inclusion. The useful comparison is your own score over time, alongside the actual Discover share and revenue line.
Four questions editors usually ask
Does a high score guarantee Discover traffic?
No. The score measures readiness and consistency. Story quality, audience interest, competition, seasonality, and platform changes still affect distribution. A high score tells you that a technical or process failure is less likely to be the explanation; it does not promise reach.
Do we need special Discover schema?
No special Discover schema is required. Use accurate Article or NewsArticle markup where it reflects the page, keep the visible date and author consistent, and make the image available through the normal preview fields. Follow Google's structured-data guidance rather than adding types that the page cannot support.
Should we change every headline after one story underperforms?
No. Review a pattern across a consistent sample and a rolling window. Change a house rule when the evidence repeats, not because one story had a quiet day. Log the test so the next editor knows what changed.
Where does an agency fit?
An agency should help the newsroom measure eligibility, indexation, previews, content patterns, and revenue context, then make the findings usable by editors and developers. GrowNexus supports news and digital publishers through SEO and search work, growth consulting, website and app development, and social media distribution. The editorial team remains responsible for the beat, reporting, and final publication decisions. Start with the Google Discover optimization guide and the Discover recovery guide for the deeper diagnostic methods.
The same discipline applies to the PR and communications program: make the entity clear, state the evidence, and measure the surface you are actually trying to influence. If the result cannot be reproduced from a live URL and a dated report, it is not a benchmark yet.
GrowNexus is based in Nepal: our Kathmandu studio is the primary working location and Birgunj is the registered office. We serve clients in Nepal, the United States, the United Kingdom, Australia, the UAE, India, and Europe; those are markets served, not additional offices.
The monthly worksheet
Copy these prompts into the team's recurring review:
- Which recent URLs fail indexation, canonical, or preview checks?
- Which templates expose a large, relevant image and a truthful date?
- Which headlines state the story clearly without withholding the fact?
- Which beats are core, adjacent, or off-beat this month?
- Which bylines resolve to a real author page?
- What is the Discover, Search, direct, and social mix?
- What revenue or value sits beside each surface?
- Which owner has the next action, and when will it be checked?
This is the benchmark's intended use: a small, honest control loop that helps a publisher decide what to fix next. It is not a traffic guarantee, a ranking promise, or a substitute for editorial judgement.
