Discussion starter: a publisher and community SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- AI Search
- News SEO
With AI answers absorbing explainer traffic, how are you splitting newsroom resources between evergreen explainers and breaking news?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
Short answer: I would stop splitting the newsroom into “evergreen” and “breaking” and split it by how easily an AI answer can stand in for the piece. The commodity explainer (what is X, who is Y, how does Z work) is the format I would treat as most exposed, and I would put it into maintenance mode: keep the pages that still earn up to date, merge duplicates, and stop commissioning new ones without an original angle. The hours that frees up are better spent on explanation attached to live stories and on reporting only your newsroom can do, rather than on more breaking news for its own sake.
Breaking news looks somewhat more sheltered in search: Google has said it aims not to show AI Overviews for hard news. But AI features now reach Discover too, and a wire story that every outlet runs is replaceable by a competitor even without AI. So the split that fits your newsroom should come from value per journalist hour by content type, measured on your own data. An industry ratio won’t reflect your beats, audience or revenue model.
What the evidence shows so far
Useful recent newsroom data comes from the Reuters Institute’s Journalism, Media, and Technology Trends and Predictions 2026, published in January 2026 and based on a survey of 280 media leaders in 51 countries and territories. It cites Chartbeat data showing Google organic search traffic to more than 2,500 sites fell by a third (33%) globally between November 2024 and November 2025, and by 38% in the US. It says publishers that specialise in “lifestyle or utility content such as weather, TV guides, or horoscopes” are more likely to have been affected, that Discover traffic is also down, and that respondents expect search traffic to fall by more than 40% over the next three years.
The same survey shows where editors plan to move effort. Net scores (the share planning to do more minus the share planning to do less) were strongly positive for on-the-ground reporting (+91) and “contextual analysis and explanation” (+82), and negative for service journalism (-42), general news (-38) and evergreen content (-32). The report’s explanation is blunt: “Some clearly recognise that AI chatbots will answer these questions more effectively than publishers in the future, even if some of the content was generated by them in the first place.” Notice the split inside that result: explanation is going up while evergreen is going down. I read that as editors separating explanation that needs reporting from reference content that doesn’t.
On the news side, Google’s Elizabeth Reid wrote in May 2024 that “we aim to not show AI Overviews for hard news topics, where freshness and factuality are important.” That is a stated aim for one feature, written before AI Mode launched, not a guarantee. As of September 2026, Google’s “Search generative AI performance” report covers AI Overviews, AI Mode and generative AI features in Discover, so news content can still meet AI summaries on surfaces that matter to newsrooms.
Two cautions before reading all this as “AI killed explainers”. The Chartbeat figures cover all Google organic traffic over a period that included several core updates, so they are not a clean measure of AI answers. And the survey records plans and beliefs, not results. Treat it as a direction, and measure the size of the effect on your own explainers before moving people.
Sort content by how replaceable it is, not by format
A generic explainer is like a dictionary entry. It’s useful, but once a dictionary is built into the search box, few people go looking for a second one. An explainer attached to a live story is closer to a correspondent’s briefing: it contains what happened in the last few hours, what your reporters found and why it matters to your readers, and nobody can summarise that until someone publishes it.
The technical version: AI answers synthesise information that already exists in indexed pages and model knowledge. Content whose value is stable consensus knowledge is easy to synthesise. Content whose value is new facts, original data, access or local detail is hard to synthesise until it exists, and the outlet that published it is at least a candidate for the citation, even though a citation is not a click.
The exposure ratings below are my working judgement, not measured values. Use them as a starting hypothesis and replace them with your own numbers.
Content type Exposure to AI answers Where its value comes from What I would do Commodity explainer (“what is inflation”) High: stable, consensus facts Past rankings and internal links Maintain pages that still earn, merge duplicates, commission new ones only with an original angle Utility and service content (schedules, weather, basic how-tos) High: the Reuters report flags lifestyle and utility publishers Habit and repeat visits Keep where you own the data or the habit; cut the generic versions News-attached explainer (“what we know”, timelines, “what it means for renters”) Lower while the story is moving; rises as it ages Your reporting and speed of updates Invest: owned by the desk covering the story, updated as it develops, linked from every related article Original analysis, data and investigations Lower: new information Unique facts, links, subscriptions Invest where it fits your beats Commodity breaking news (wire rewrites) Moderate in search; easily replaced by other outlets Speed and Top Stories visibility Stop producing SEO-only rewrites unless you add something Original breaking news and exclusives Lower at first Being first with verified facts Protect; this is where a newsroom’s advantage sits Measure value per journalist hour, by content type
The split is a portfolio decision, so it needs portfolio data. Tag every piece by content type in the CMS (the six rows above, or your own version), then join four sources:
- Search Console performance data: clicks from the Web search type, plus the Discover and Google News reports. Web totals include AI Mode data with no separate click breakdown.
- The “Search generative AI performance” report: impressions only (no clicks, CTR or queries), filterable by page, country, device and date. It shows which explainer URLs appear in AI features, not whether anyone clicked through.
- GA4 and revenue data: registrations, subscription starts, newsletter sign-ups, returning readers and ad revenue per piece. GA4’s newer “AI Assistant” channel group picks up referrals from assistants such as ChatGPT, Gemini and Copilot, though traffic without a referrer still lands in Direct.
- Cost: reporter and editor hours per piece. A month of rough time logging is enough to start.
Then compare clicks, conversions and revenue per hour by content type, and the trend in each over the last 12 to 16 months. The cleanest AI-absorption signal on explainers is falling CTR at stable impressions and position for the same group of queries. It is more convincing when you can compare queries where your rank tracker records an AI Overview with similar queries where it doesn’t.
Two measurement traps:
Google stopped supporting the
&num=100results parameter in September 2025, and many sites then saw desktop impressions fall and average position improve, while clicks were affected far less, according to Brodie Clark’s analysis. Year-on-year impression comparisons across that change overstate the decline, so compare clicks. Separately, practitioners report that clicks from AI Mode arrive in GA4 as google / organic, so a stable organic total can hide a shift from classic results to AI Mode.How I would move resources without betting the newsroom
- Baseline the current split. Hours by content type for one month, next to value per hour from the data above.
- Move hours in steps. Shift a fixed slice each quarter (for example, a tenth of commodity-explainer hours) into news-attached explanation and beat reporting, and reverse the step if value per hour says so.
- Give explainers to the news desk. The reporter covering a story owns its explainer, updates it as the story develops, and links to it from every related article. One maintained URL per ongoing story accumulates links and history; a new “what we know” URL every day splits them.
- Keep a small maintenance crew for evergreen. Update the explainers that still earn subscriptions or revenue, merge overlapping ones, and add original elements (your data, your experts, local detail) where a page is worth saving.
- Put breaking-news effort where you have an edge. That means beats where you have sources, local knowledge or specialist reporters. Google’s February 2026 Discover core update, which started with US English users, said it would show more locally relevant content from sites based in the user’s country, less sensational content, and more in-depth, original content from sites with topic-by-topic expertise.
- Build direct demand. Google has extended preferred sources to AI Mode and AI Overviews (May 2026), added an embeddable “preferred sources” button for publishers (August 2026), and says users are about twice as likely to click a preferred source. Where the feature is available in your market, asking loyal readers to add you costs little and helps news and explainers alike.
Hypothetical example:
A regional newsroom with 30 reporters logs hours for a month. Commodity explainers take 15% of reporter hours and still bring 25% of search clicks, but after a year of falling CTR they produce only 6% of subscription starts. News-attached explainers take 5% of hours and produce 12% of subscription starts. Per hour, the second group is about six times as productive for subscriptions. That justifies moving a slice of hours this quarter and measuring again. It does not justify scrapping the evergreen archive, which still brings a quarter of search clicks.
The AI opt-out doesn’t fix the explainer problem
Since 31 August 2026, every site can use the “Search generative AI control” in Search Console to opt out of AI Overviews, AI Mode and AI Overviews in Discover. It works at site level, so you cannot opt out your explainers and leave your news in. Opted-out sites get no traffic or impressions from those features, and Google says the setting won’t be used as a ranking signal outside them. Digiday’s June 2026 coverage described it as a choice publishers can’t safely use. For a newsroom, it is a policy decision for commercial and legal leads, not a way to protect explainer traffic.
What to test, and what would change the plan
You can’t randomise the news, but you can randomise the explainer archive. Take 30 to 60 commodity explainers whose CTR has fallen and randomly assign them to three treatments:
- Leave: no change (the control).
- Rebuild: add original elements such as your own data, expert interviews or local detail, as a genuine update.
- Fold in: merge the page into a maintained, news-attached explainer and redirect it.
Measure clicks, AI impressions from the generative AI report, subscription starts and hours spent per page. Start the clock when URL Inspection shows Google has recrawled the changed pages. Explainer traffic moves in weekly cycles and jumps unpredictably when related stories break, so I would want at least six full weeks after recrawl for those swings to average out across the groups, longer for seasonal topics, and I would not read results during a core update rollout. For the reallocation on news desks, a workable design is a before-and-after on the desk that changes against a comparable desk that doesn’t, over a full quarter. That is weaker evidence, and it should be presented that way.
- Rebuilt explainers recover clicks or subscription starts at a better value per hour than news-attached work: keep a larger evergreen team and rebuild more of the archive.
- Folded-in pages keep their traffic inside the news-attached explainer: consolidate more of the archive.
- Nothing recovers, and AI impressions stay high while clicks fall: the archive is serving AI answers more than readers. Keep it accurate for brand reasons, and move the hours.
Where I would land
If this were my newsroom, I would not set a fixed ratio. I would tag content by how replaceable it is, measure value per hour for a quarter, shift a slice of commodity-explainer time into news-attached explanation and beat reporting, and let the archive test show how much evergreen work is still worth doing. The format label matters less than whether each piece contains something an AI summary cannot produce until you publish it.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.