YouTube is testing AI-generated answers at the top of its standard search results: for some users, a search no longer starts with a list of videos but with an Ask YouTube response. Social Media Today reported the test on 18 October 2026, citing YouTube's own page of test features and experiments. It covers select signed-in users on desktop in the United States, searching in English.
Until now, Ask YouTube was a separate mode: a user chose to ask a question and received a conversational answer. In this test, the answer comes to the user. Someone who types a practical question into YouTube's ordinary search bar may see an AI response first, with the usual videos below it.
For brands that use video, this changes what being visible on YouTube means. A search can now produce two kinds of presence: appearing among the videos, and contributing to the answer above them.
What is YouTube testing?
An Ask YouTube response placed at the top of ordinary search results, aimed at open-ended, how-to and skill-based questions that have no single right answer. YouTube's example is a question about removing marks from painted walls. In its words, "if you are part of this experiment, you may see an Ask YouTube response at the top of your search results".
The user can select "Show more" to see the full response and ask follow-up questions, or scroll past it to the standard video results. The test is limited to "select signed-in desktop users in the U.S. searching in English". YouTube has given no figure for its scale or duration, and has not said whether mobile or other languages will follow.
How did Ask YouTube get here?
In six months, Ask YouTube went from an opt-in test for paying subscribers to a feature open to all US users, and now to a test inside standard search. The steps, as reported by the trade press:
| Date | Step | Who could use it |
|---|---|---|
| 27–28 April 2026 | Ask YouTube launched as an experiment in YouTube Labs | US Premium subscribers aged 18+, opt-in, desktop, English |
| 6 July 2026 | Wider desktop availability, as reported by seo-day.de | Signed-in US users aged 13+, desktop, English |
| 13 October 2026 | Available to all US users, including the mobile app | All US users |
| 18 October 2026 | Ask YouTube responses tested at the top of standard search results | Select signed-in US desktop users searching in English |
When the April test opened, Search Engine Journal described the answer format: a generated text summary, a primary cited video linked to a timestamped section, and galleries of long-form videos and Shorts. In its test, a follow-up question produced a comparison table linking back to the source videos. YouTube warned that "quality and accuracy may vary" and asked for feedback.
What does this change for visibility on YouTube?
A YouTube search can now produce two distinct forms of visibility, and a brand can hold one without the other. KAIS separates them as follows.
Video Discovery is the classic form: a brand's video appears in YouTube's search results. AI Answer Inclusion is the new one: a video, its creator or its content contributes to an AI answer displayed on YouTube, as the primary cited video, a timestamped segment, or a video in a supporting gallery.
The two do not have to coincide. A video that ranks well in the list may not be the one the answer relies on. A video further down may supply the segment the answer quotes. And when the answer sits above the list, it takes the first place in the user's attention, much as AI Overviews do in Google Search.
This is a direct extension of Search Everywhere Optimization, the KAIS view that visibility is built across every place where people search, not only Google. YouTube is one of the largest of those places, and it is now adding its own answer layer.
Can AI Answer Inclusion be measured yet?
Not reliably: the test is limited, its selection rules are unpublished, and it is not visible from most markets. YouTube has not said how it chooses the videos an answer relies on. The search-results test reaches only some signed-in desktop users in the United States, searching in English. A measurement taken from Bangkok, Paris or a signed-out account would not see it at all.
The lesson of the past weeks applies here too. A study of Google results collected by rank trackers showed how much the conditions of collection shape what is measured, as discussed in Your rank tracker may be seeing a different Google. For a feature shown only to some accounts, in one country, on one device, the account, location and device are part of the measurement. A citation indicator built without recording them would not be reliable.
What can be done now is observation. On an eligible account, for a fixed set of how-to questions, one can record whether an answer appears, which video is cited first, which segment is linked, and which videos appear in the galleries. That is a log of observations, not yet a score.
How can a brand prepare?
By making its videos easy to quote for the practical questions its buyers ask. YouTube has not published how Ask YouTube selects videos, so what follows is the KAIS reading, based on how the answers are built: a summary, a primary video linked to a timestamped section, and supporting galleries.
First, start from the questions. The test targets open-ended, how-to and skill-based questions. A brand should list the practical questions its buyers ask and check which of them its videos actually answer.
Second, make the answer easy to locate inside the video. Since answers link to timestamped sections, a video organised in clear chapters, each answering one question, gives the system a segment to point to. This follows the same logic as page structure for text, where cutting at real sections helped AI retrieval most, as covered in Page structure helps AI find your content.
Third, say the answer plainly. A segment that states the answer in its first sentences, with the specific names, figures and terms the question uses, is easier to cite than one that builds up to it. The same failure modes that leave web pages uncited, as in Why AI doesn't cite your page, are likely to apply to video: answering a neighbouring question, or omitting the entity the buyer named.
Fourth, write what the video says. Accurate titles, descriptions and captions describe the content in words a system can read. They help classic YouTube search today, whatever the answer layer becomes.
Four steps to prepare videos for YouTube answers
- Questions first: list the how-to and practical questions buyers ask, and map them to videos.
- Chapters: one question per chapter, so an answer can link to a precise segment.
- Plain answers: state the answer early, with the names and figures the question uses.
- Words on the page: accurate titles, descriptions and captions for every video.
How solid is the information?
The test itself is confirmed by YouTube's own wording, but almost everything that would matter for a strategy is still unknown. YouTube has not published how many users see the answers, how videos are selected, whether views from answers are reported separately, or when other countries and languages might follow. The timeline relies on trade press reports of YouTube's test pages. Some sources describe further details, such as how views from Ask YouTube count in analytics, without linking to an official YouTube page, so they are left out here.
Supported: YouTube is testing Ask YouTube responses at the top of standard search results for some signed-in US desktop users searching in English, aimed at open-ended and how-to questions, with "Show more" and follow-up questions. Reported, not officially detailed: answers cite a primary video linked to a timestamped section, with galleries of videos and Shorts. Not known: selection rules, scale, and any date for other markets.
Where does this sit in the KAIS AI Visibility Chain?
The test extends the last stages of the chain, Be Cited and Influence the Answer, to video.
Be Discovered → Be Retrieved → Be Understood → Be Trusted → Be Cited → Influence the Answer → Be Recommended.
Video Discovery corresponds to Be Discovered and Be Retrieved on YouTube. AI Answer Inclusion corresponds to Be Cited and, when the answer is built around a brand's segment, Influence the Answer. The two need separate observation, for the same reason that being cited and being recommended need separate measures in AI engines.
On YouTube too, the answer now comes before the list.
For years, visibility on YouTube meant a place in the list of videos. In this test, the list starts below an answer. A brand's video can still rank well and be skipped, or sit lower and supply the segment the answer quotes.
The measurement is not ready yet. The preparation is: questions mapped, chapters clear, answers said plainly.
Sources
- YouTube Community — YouTube test features and experiments. YouTube's page of current tests, cited by Social Media Today for the search-results experiment.
- Social Media Today — YouTube Tests AI Replies in Searches, Andrew Hutchinson, 18 October 2026.
- Social Media Today — YouTube rolls out AI search chatbot to all US users, 13 October 2026.
- Search Engine Journal — Google Tests 'Ask YouTube' Conversational Search Experiment, Matt G. Southern, 27 April 2026. Answer format and initial eligibility.
- MacRumors — YouTube Tests AI-Powered 'Ask YouTube' Conversational Search Feature, 28 April 2026.
- seo-day.de — Ask YouTube: AI search expands to US desktop, July 2026. Used only for the July eligibility change.