57% Saw AI Summaries in Search Last Week. Rewrite Your Content Plan
More than half of people surveyed in nine countries saw AI-generated answers or summaries in search in the previous week, and weekly use of standalone AI tools has risen from 34% to 51%. For a small product, that means planning content for quotation and for owned distribution, not only for a blue link.
Key takeaways
- 57% of people across nine countries saw AI-generated answers or summaries in search in the previous week.
- Weekly use of standalone generative AI tools rose from 34% to 51% in the six countries with trend data.
- 38% now use AI to get information, up from 11% in 2024, and convenience is the main reason.
- Only 14% used an AI agent for a real task, so search habits still matter more than agent habits.
- Build answer-first pages, measure AI referrals and keep an owned channel that does not depend on a result page.
On this page 9
- How large is the shift in the report?
- What does the 57% figure actually measure?
- Why do people use AI instead of searching?
- Do AI users actually search less?
- Are AI agents about to replace search?
- What should a content plan do differently?
- What this means for bootstrapped founders
- Where ZeroRaised fits
- Sources
The Reuters Institute's Generative AI and News Report 2026 found that 57% of respondents across nine countries had seen AI-generated answers or summaries in search engines such as Google and Bing in the previous week (the Reuters Institute report). Weekly use of standalone generative AI tools rose from 34% to 51% in the six countries with trend data. The survey covers news, not software buying, so treat it as a signal. The signal is clear: the answer box is a mass habit, and a page that cannot be quoted loses the click twice.
TL;DR
- 57% saw AI-generated answers or summaries in search in the previous week, across nine countries.
- Weekly use of standalone AI tools rose from 34% to 51% in the six countries with trend data.
- 38% now use AI to get information, up from 11% in 2024, and speed is the main reason.
- Only 14% used an agent for a real task, so search still matters more than agents.
- Build answer-first pages, measure AI referrals and keep an owned channel.
Disclosure: ZeroRaised publishes this analysis and is a launch site for bootstrapped products. Facts come from the linked sources as checked on 1 Nov 2026; ZeroRaised has no commercial relationship with the companies named.
Answer engine optimisation is the work of writing and structuring a page so that an AI answer system can retrieve, understand and quote it, in addition to whatever ordinary search optimisation the page already does.
How large is the shift in the report?#
The report is based on a YouGov survey in nine countries: Argentina, Australia, Denmark, Finland, France, Japan, Norway, the UK and the USA, with about 2,000 respondents per country, fielded from 14 August to 28 September 2026 (the Reuters Institute report).
Weekly use of standalone generative AI tools rose from 34% to 51% across the six countries with trend data. Use at least once a day went from 7% in 2024 to 16% in 2025 and 28% in 2026. The habit is no longer early adopter behaviour.
What does the 57% figure actually measure?#
It measures exposure, not preference. Across the nine countries, 57% had seen AI-generated answers or summaries in search engines in the last week. Separately, 25% had seen them on devices, 21% on social or messaging services, and 20% had not seen them anywhere (the Reuters Institute report).
The report also cites Chartbeat data showing that Google search referrals to more than 2,500 news sites fell by a third between November 2024 and 2025 (the Reuters Institute report). That does not prove AI summaries caused the whole decline. It does show that a referral channel can shrink faster than a content team can rewrite.
Why do people use AI instead of searching?#
Using AI for getting information rose from 11% in 2024 to 24% in 2025 and 38% in 2026. By task, 24% used it for researching a topic, 20% for advice, 18% for factual questions and 15% for recommendations (the Reuters Institute report).
The reasons are not mainly about trust. People cited a faster and easier answer at 64%, interactivity at 56% and aggregated information at 54%. Only 21% said AI gives more reliable information (the Reuters Institute report).
That last number is the opening: write the short answer, then show the evidence an aggregator would otherwise flatten.
Do AI users actually search less?#
The report finds that people who use standalone AI are more likely to say they use search engines less than more, with a difference of 13 percentage points. For news websites the difference is 3 percentage points, and the largest group says their use is about the same (the Reuters Institute report).
Read that carefully. The direction is real, but the largest group has not changed. Search is not disappearing for most people; it is sharing attention with an assistant that answers first. A content plan that abandons search entirely would be as wrong as one that ignores the answer box.
Are AI agents about to replace search?#
Not yet. In the last week, 67% had not used an AI agent, 13% had only experimented and 14% had used one for a real task. In the UK only 9% had used an agent for a real task. Age matters: 26% of 18 to 24 year olds did, against 7% of those aged 55 and over (the Reuters Institute report).
The report notes that Meta released a personal agent called Muse and OpenAI released a product called Dots. Distribution through agents is coming, but the current numbers do not justify rebuilding a business around it. Make the product easy to understand and act on, then wait for behaviour to catch up.
What should a content plan do differently?#
Start with questions, not keywords. Pick the ten questions a buyer asks before they trust the product, then give each one a page with a direct answer in the first two sentences. The guide to optimising for AI search covers the structure in detail.
Add the pieces an answer needs to be assembled from: a plain definition, a table or short list, a dated fact and a source. Keep the page crawlable, because a blocked page cannot be cited.
Then stop treating search as the only front door. Publish the same analysis in a newsletter, a community post and a launch listing, so a referral decline does not remove your only route to a buyer. The free marketing channels guide lists routes that cost time rather than money.
Measure new referrals, but do not expect clean attribution: some AI traffic looks like direct traffic. The Google AI Mode analysis and the Reddit and logged-in AI search piece cover adjacent shifts in the same market.
What this means for bootstrapped founders#
- Rewrite your three most important pages to answer the buyer's question in the first two sentences.
- Add a quotable definition and one dated fact to each page, with the source linked.
- Track referrals from AI products, branded search and direct traffic together.
- Put the same material in an email list you own, because the answer box is rented ground.
- Do not build an agent-only strategy yet: 14% real use is a beachhead, not a market.
- Recheck the numbers each quarter and change the plan when the survey does.
The strategic point is not that content stopped working. It is that content now has two jobs: persuade a person who lands, and survive being summarised for a person who does not. The first 100 users guide covers the rest of the distribution problem.
Where ZeroRaised fits#
ZeroRaised is a launch and discovery site for products built without institutional venture capital or accelerator equity. If you have a product people can try, submit it here and describe it in plain language. The pricing page explains the current options.
For more analysis like this, join the ZeroRaised newsletter. We make no promise of traffic or results, only a place to present the product and a reason to keep the facts consistent.
Sources#
Frequently asked questions
What did the Reuters Institute find about AI summaries in search?
Its Generative AI and News Report 2026 found that 57% of respondents across nine countries had seen AI-generated answers or summaries in search engines such as Google and Bing in the previous week. The same survey found that weekly use of standalone generative AI tools rose from 34% to 51% in the six countries with trend data.
Does this mean SEO is dead for small products?
No. Traditional search still sends buyers, and the report also shows that the largest group of AI users say their search use is about the same. What changes is the shape of the work: pages need a direct answer, clear structure and sources so they can be quoted above the link, not only ranked below it.
Should a bootstrapped founder build for AI agents now?
Not as the main channel. In the report, 67% had not used an AI agent in the last week, 13% only experimented and 14% used one for a real task. The UK figure for real tasks was 9%. Agents are worth watching and worth making your product easy to use from, but they are not yet a substitute for search or email.
What is the fastest content change to make?
Put the answer in the first two sentences of the page, then support it with a definition, concrete evidence and a short list of takeaways. That structure serves a person in a hurry and gives an AI summariser a passage it can lift without guessing.
How do I measure whether this work pays off?
Check referrals from AI assistants and answer features, branded search, citations of your pages and conversations that mention them. Some AI traffic arrives with no clean referrer, so treat direct traffic and branded search as part of the signal rather than relying on one report.