SEO · 06
News SEO Consulting.
Most of a newsroom's Google clicks come from a feed that swings and that nobody types into. I prove why yours moved, and train editors and developers to run it.
You hire me when
- Discover or Search traffic dropped, and nobody can say why.
- Most of your Google clicks come from a feed nobody on the team watches.
- A relaunch or a new CMS is coming for the newsroom.
- You want AI in the newsroom without losing quality.
What you get
- Discover, Google News and Search side by side: 16 months of Search Console data.
- Traffic forensics: every number traced to its source, every finding checked by a person.
- The article template checked: the 3 SEO tests, images, compression, NewsArticle and author markup.
- Newsroom training on your own articles, with tasks and an owner per desk.
- A scorecard the desks run without me: task, pass rule, owner, date, result.
Why news SEO is mostly not about search
A newsroom thinks in queries and rankings. Most of its Google clicks come from somewhere else: Discover, the feed under the search box on phones, where nobody types a query.
In six newsroom accounts I worked with from 2023 to 2026, Discover brought about 60 to 80 percent of all clicks from Google. It rewards clicks and interaction, like a social feed, and it swings. Google says so itself: “Traffic from Discover is less predictable or dependable when compared to keyword-driven search visits.” (Google Search Central, Discover and your website)
So a newsroom needs two things a search-only SEO does not give it: a way to prove why a channel moved, and a routine the desks run every week. As I put it: “For News Site SEO is not about Google Search, but Google Discover Optimization.” My goal: your newsroom knows which channel each click comes from, why it moved, and what the desks do next.
What goes wrong in newsrooms in 2026
These are the failures I measured in my newsroom audits and second opinions, and what I specified for each:
| What breaks | What I measured | What I specify |
|---|---|---|
| The relaunch gets the blame for Google’s update | About 75 to 80 percent of Discover back in every market but one; under 2 percent there | Split by market and URL pattern before anyone touches the CMS |
| Traffic gone, and the code is fine | JavaScript off: passed on every page type; indexed: every page type | Rule out before you fix; every number traced to its source |
| Compression that makes pages bigger | Brotli bigger than gzip | One CDN setting, one check per deploy |
| Images too small for the feed | Article images about 600 px wide | 1200 px, three crops, linked to the large version |
| Evergreen content in a news feed | Discover well under 1 percent of what web search brought | Discover for news, Search for evergreen, measured apart |
| Planning for AI traffic first | All AI sources: 0.02 to 0.04 percent of sessions | Measure it, plan around Discover and Search |
| Tasks done, result missing | A speed score that went down over three dates while the tasks were marked done | A pass rule per task, tested on the live page |
| A routine that stops | A routine that ran for a while, then stopped | One owner per step, a scorecard on every call |
The relaunch gets the blame for Google’s update
After one site change, a publisher lost a large part of its Discover traffic. The new system got the blame.
- I split the publisher’s own Search Console data by URL pattern and by market.
- In every market except one, the new URLs got about 75 to 80 percent of the old Discover clicks back. In one large market, under 2 percent.
- Same URLs, same code, opposite outcome by country. A broken URL pattern would have hit every market.
It matched Google’s February 2026 Discover update, which lists “Showing users more locally relevant content from websites based in their country” (source). The full case, with the click rate and the clickbait share, is on the Discover workshop page.
What I specify: a split by market, URL pattern and device before any verdict, a control market, and a stop rule for the market Google re-weighted.
Traffic gone, and the code is fine
One site lost a large share of its Google traffic. The first suspect was the code.
- I ran the page with JavaScript turned off on every page type. It passed on every one.
- Every page type was indexed.
- The cause was not in the code, and I said so.
In that forensic job every number on every slide was traced to Search Console, GA4 or the Wayback Machine. AI tools helped with the analysis, the handover named them, and every finding was checked by a person.
What I specify: rule out before you fix. Each hypothesis gets a test, a source and a verdict, including “not the cause”.
Compression that makes pages bigger
Brotli should be smaller than gzip. In one audit it was bigger, page type after page type. The cause was a streaming setting at the CDN, not the quality level. Set right, the page is much smaller than what was served.
Google’s crawlers tell the server which compression they accept: “The content encodings supported by each Google user agent is advertised in the Accept-Encoding header of each request they make.” (Google crawlers overview) Googlebot asks for Brotli, so it got the heavier file. The full case is on the technical SEO page.
What I specify: one CDN setting, no template change, and a check on every deploy: Brotli bytes smaller than gzip bytes.
Images too small for the feed
In the feed, the image is the biggest part of the card. In one audit the article image was about 600 px wide: half of what my image rule asks for.
What I specify: one source image, cropped to 1200 px in 16:9, 4:3 and 1:1, all three in the NewsArticle markup, max-image-preview:large, and the image linked to its large version. The full rules are in the Discover best practices.
Evergreen content in a news feed
On one site with mostly evergreen content, Discover brought well under 1 percent of what web search brought. The feed wants what is new.
That is not a failure of that site. It is the wrong channel for that content, and the wrong KPI if anyone measures it there.
What I specify: Discover targets for news desks, Search targets for evergreen pages. Evergreen stories get updated at least once a season, the URL never changes.
Planning for AI traffic first
Every newsroom asks about AI search. Across three news sites over the year to October 2026, Discover brought about 74 percent of all Google clicks. All AI sources together sent 0.02 to 0.04 percent of sessions, depending on the window.
What I specify: measure AI referrals every month, from your own analytics. Plan the desks around Discover and Search until your own data says otherwise.
Tasks done, result missing
One editorial team implemented almost every task from its workshop. I re-measured the core page types on three dates over three months. The speed score never reached the goal. It went down, not up.
The tickets were closed. Nobody owned the question whether the result arrived on the live page.
What I specify: every task gets a pass rule you can test on the live page. Done means the test passes, not that the ticket is closed. The full case.
A routine that stops
In a 2026 call, a newsroom described a routine it had set up after a workshop. They did it for a while, then they stopped.
As I tell every newsroom: “A process without owners stops after two weeks.”
What I specify: one owner per step of the weekly cycle, named in the workshop, and a scorecard that is the agenda of every follow-up call.
How I work with newsrooms
- Data first: 16 months, three channels. Before any workshop or verdict I pull 16 months of Search Console data for Discover, Google News and Search, side by side. Discover, Search and News are three reports, and I read them apart.
- Split before blaming. Every drop gets read by market, URL pattern and device, in impressions, clicks and click rate. In one 2026 case that split moved the blame from the relaunch to a Google update.
- Every number traced. In forensic jobs each figure carries its source: Search Console, GA4 or the Wayback Machine. If AI tools did part of the analysis, the handover says which, and a person checked every finding.
- The article template, tested. The 3 SEO tests (PageSpeed Insights, the page with JavaScript off, Google’s live test), images, compression and markup, re-measured on fixed dates. In one training I measured the template on three dates over three months.
- Markup questions answered from Google’s docs. Who goes into the author field, a person or the organisation: in 2026 I answered such questions for one newsroom by email, every point with a link to Google’s own documentation.
- Training on your own articles. Editors, desk heads, developers and the data owner in one room. I have trained editors at more than ten newsrooms and magazines. One newsroom client ran a four-module remote program: sessions every two weeks, tasks in between, a certificate when the tasks were done. For AI in the newsroom: an editorial AI workshop came with a tools guide for the editors.
- A routine with owners. You leave with tasks and processes per team, an owner for each, and a scorecard. The Discover workshop is the full version of that day.
This site, measured the same way
This site is not a newsroom, but my own Discover SEO article should pass my own rules. I checked its live HTML on 12 November 2026, found three failures, and fixed what my own code controls the same day.
What passed from the start:
max-image-preview:largeis set in the robots meta tag.- The hero image is 1536 px wide, with width and height, loaded eagerly with
fetchpriority="high", with a responsive WebPsrcset(800 and 1536 px). - The author in the markup is a Person with a profile URL.
What failed, and what I changed:
- The hero image was not linked to its large version. Now it is, like every photo on my service pages. The click is cancelled, a reader stays on the page.
- The markup carried one image, not the three crops. Now it carries 16:9, 4:3 and 1:1, each at least 1200 px wide, and the build fails if a crop is missing or too small.
- Brotli was bigger than gzip: about 13.8 KB against 13.7 KB. The cause was the CDN’s on-the-fly compression. My build now ships every page precompressed at Brotli’s highest level; on the www host it waits for one CDN setting.
The article is a blog post, so BlogPosting instead of NewsArticle is correct. Check it yourself: open the page, view the source, and compare curl -H "Accept-Encoding: br" with gzip.
Questions newsrooms ask me
Do we need structured data to get into Discover?
No. Google: “No special tags or structured data are required.” (source) I still specify NewsArticle with the three image crops, because it tells Google which image, date and author belong to the story. For Google News, Google says: “While there’s no markup requirement to be eligible for Google News features like Top stories, you can add Article to more explicitly tell Google what your content is about” (source).
Our Discover traffic dropped and we changed nothing. What now?
Read it apart first: impressions, clicks and click rate, per market and device. Then the update dates. Google: “Our ongoing work to improve Discover’s user experience means sites may see changes in their traffic that are unrelated to the quality or publishing frequency of their content.” (source) After the update has finished, take the new level as your baseline and test from there.
Who goes into the author field, the journalist or the newsroom?
Whoever the page shows as author. Google: “Make sure that all the authors that are presented as authors on the web page are also included in markup.” (source) For a story signed by the newsroom: “If the author is an organization, you could link to the organization’s home page.” (source)
Do we need a Google News sitemap?
Google does not require it for Discover. I do: without a news sitemap and managed visibility in Google News, Discover stays luck, not a channel. Keep it fresh. Google: “Only include recent URLs for articles that were created in the last two days.” (source)
How much traffic do we get from AI?
Measure it in your own analytics. In my 2026 data for three news sites, all AI sources together sent 0.02 to 0.04 percent of sessions. Discover brought about three quarters of all Google clicks.
Can you guarantee Discover traffic?
No. Google: “Note that being eligible to appear in Discover is not a guarantee of appearing.” (source) What I can give you is the proof of why it moved, and a newsroom that acts on it every week.
Sources for the Google quotes on this page, read on 2026-11-12: Discover and your website, Article structured data, News sitemaps, Google crawlers overview, Google Search Central, and the February 2026 Discover core update post. The client numbers come from my own audits, workshops, second opinions and handovers in 2024 to 2026.
From real jobs
- A Discover loss after a site change, split by market; the cause was a Google update, not the code.
- 2026: a large Google traffic loss; rendering and indexing ruled out on every page type, every number traced to its source.
- 2026: Brotli bigger than gzip; one CDN setting was the cause.
- An AI workshop for editors, with a tools guide for the newsroom.
- Discover and News training, the article template re-measured on fixed dates.
- 2023 to 2026: Discover about 60 to 80 percent of Google clicks in six newsroom accounts.
- Editors trained at more than ten newsrooms and magazines, including a four-module remote program with certificates.
- A Discover study with about 150 hypotheses on publisher data.
Teams like yours
Some of our customersWorkshops for this
All workshop formatsNewsroom SEO Workshop
Your editors learn Search, News and Discover on their own articles and data.
The workshopGoogle Discover & News Feed Workshop
Your newsroom builds its own Discover process with me, from the tasks to the feedback loop, on its own articles.
The workshopAI Newsroom Workshop
Your editors use AI daily and keep their quality and tone.
The workshopSomething else in the way?
How consulting works- External SEO lead
- Relaunch and migration SEO
- Technical SEO
- International SEO
- Web performance
- Content strategy and search demand
- Search data and traffic forensics
- SEO agency cleanup
- GEO and AI search
- Vibe coding training
- AI automation consulting
- AI-first transformation
- Custom tools for one job
- Design thinking
- Growth
- Growth strategy analysis
- SEO workshops and training
- Keynotes and talks
- Sparring and second opinion
- Dev, product and marketing mediator