How YouTube fits an expert practice
Ask what YouTube does for an expert firm and the answer usually arrives in the vocabulary of reach. Subscribers, watch time, posting cadence, the algorithm. That vocabulary belongs to businesses that sell attention and rent it back to advertisers. Your business sells judgment to a small number of people who pay well for it. The useful question is narrower. What does a video keep doing in the months after the week it went up?
For an expert, the answer is reference. A recorded talk answers a question your buyers ask. A person finds it later, and a machine quotes it when assembling an answer. Views tell you it is happening. Reference is the thing you were paid in.
The return on YouTube is being someone's source
Ahrefs released its Q1 2026 AI Search Benchmark Report in May. It draws on 146 million search result pages and 730,000 AI responses. One study inside it covered 75,000 brands. YouTube mentions tracked with AI brand visibility more closely than any other measure tested. A site's link volume and page count barely moved with it.
Marketing spent 15 years treating links as the currency of authority. That ordering is a change in what counts.
Mentions and uploads are different things. A founder gets mentioned on YouTube by appearing on someone else's show. A reviewer names the firm. A client walks through the work. Owning a channel is one route in. Guesting is another, and for many experts it is the faster one, so that fork gets settled first.
The second-largest-search-engine line comes from 2008
You have heard that YouTube is the world's second-largest search engine. The claim traces to a 2008 comScore reading that put YouTube's on-site searches ahead of Yahoo's. Two decades later the line still gets repeated, with no fresh measurement underneath it. Read it as a slogan.
What survives the correction is smaller and more useful. People do search YouTube, mostly for things easier to watch than to read. They look for procedures, walkthroughs, and comparisons. They also want to see whether they like the person talking.
Search rarely supplies most of a channel's views. The home feed and the suggested rail carry more of that load. Search views arrive slowly and keep arriving. A channel's total climbs while single uploads look quiet for months.
Small channels get cited, and that is the useful finding
OtterlyAI examined more than 100 million AI citations across six answer engines. The researchers looked at which YouTube videos were being quoted. Views, likes, and subscriber counts showed no meaningful relationship with how often a video was cited. 41% of the cited videos had fewer than 1,000 views.
A video with 200 views can be the source an answer engine reaches for. It only has to hold the clearest answer to the question.
That finding matters most to an expert practice. It removes the barrier that keeps accomplished people off the platform. You do not have to win a category to be quoted inside it.
Three caveats belong with the finding. The study looked only at videos already cited, so it explains repeat citation better than first selection. Correlation is not cause, and the researchers say so themselves. And the scale is modest. Social and video sources made up about 5% of all citations in that dataset. YouTube took roughly a third of that slice. Most of what an answer engine quotes is still ordinary web pages. The firms publishing these numbers also sell AI-visibility software. Trust the direction more than the decimal.
Google and the answer engines are moving in opposite directions
Google is showing fewer videos. Semrush's tracking had the video carousel appearing on just over 30% of queries, and it has fallen under 20%. One tracker following 1,200 large-brand searches watched the feature drop to around 5% after August 12, 2026.
Google also narrowed video thumbnails in 2023 to pages where the video is the main content. That ended an old trick. Embedding a clip inside an article no longer earns a video result.
Answer engines went the other way, and they disagree with each other about video. Google's AI answers and Perplexity lean on YouTube heavily. ChatGPT accounted for 4.4% of the YouTube citations in the Otterly dataset, and Gemini and Copilot for almost none.
So the payoff depends on where your buyers ask their questions. A buyer who lives inside ChatGPT will probably not meet your videos there this year. A buyer using Google's AI answers or Perplexity might meet them this month.
What earns the reference
Length does more work here than polish. In that same dataset, 94% of AI citations pointed at long-form videos and roughly 6% at Shorts. Half of the cited videos ran between 5 and 20 minutes. Chaptered videos were cited repeatedly, often across several chapters at once. A chapter behaves like a heading on a page.
The shape that follows is plain. One question per video, phrased the way a buyer would ask it, and answered completely. Mark the sections, so a reader or a machine can jump to the part it needs. Whether your channel is set up to earn that is a separate question. The answer turns on titles and format rather than equipment.
Recording is the cheap part for a practice that already captures its founder's thinking. An hour of recorded interview produces the talk and the transcript in one sitting. Your site, your newsletter, and the answer engines all read that transcript.
Where YouTube does not pay for an expert practice
Some buyers do not watch. Search the three questions your clients ask you most. Look at what Google and your buyer's assistant return. See whether any video appears at all. When no video appears anywhere, a channel sits a long way from your revenue.
Review costs decide the cadence in regulated work. Clinicians, attorneys, and advisors carry an approval step on everything published. Filming takes an hour. The recurring cost is the review, and that cost decides whether a monthly rhythm survives its first quarter.
Some expertise cannot be said out loud at a useful level of detail. Sometimes the specifics that would make a video worth watching are the ones you cannot disclose. What is left is thin, and thin does not get quoted.
A channel confirms interest more often than it manufactures interest from nothing. That limit runs through every other piece of a public presence. Expect a video to make a warm buyer more certain. Expect much less of it as a cold-start machine.
A channel that stops at six videos leaves a shelf too thin to be anyone's reference. The decision worth making is about the twenty-fourth video.
Take the question you have answered so many times you could give it from memory. Search it on YouTube and watch who answers it now. Compare the best answer there with the one you give in a first meeting. If yours is better, that gap is what a channel is for. Closing it does not require an audience.
