Key findings
YouTube is cited in one in four AI Overviews about New York car accidents.
What the 40 most cited videos have in common.
One in four AI Overviews cites a YouTube video
When someone searches Google for help after a car accident in New York, the AI Overview at the top of the page cites a YouTube video in one answer out of four. No other source comes close on that engine, and the other engines barely use video at all.
We ran 111 questions that people ask after a crash in New York, every day for 30 days to 7 November 2026, across 14 competing firms. Google AI Overviews produced about 2,390 answers. YouTube was cited in 25.6% of them, 1,081 citations in all, and when an Overview read a video it cited it 1.75 times on average. On Gemini the figure was 0.3%. On ChatGPT it was under 0.1%.
That makes video the one source group that moves Google’s AI answers and almost nothing else. Its share of voice is 55% on ChatGPT and 21% on AI Overviews.
What the 40 most cited videos have in common
We pulled the 40 videos Google cited most often. Together they account for 591 of the 1,081 citations, 55% of the total. Four things stand out.
Thirty of the 40 answer one question, and the question is the title. Is New York a no-fault insurance state. Should I accept the first settlement offer from the insurance company. How long do you have to report a car accident in New York. The title reads like the search.
Twenty-four name New York or a New York city in the title. Buffalo, the Bronx, Rochester, Albany, New York City. The other sixteen are general questions, and Google cites videos from firms in Florida, Texas and Arizona for those. A New York firm that has not made the video loses the citation to another state.
Seven are Shorts. Under a minute, vertical, one question each. Google cites them alongside ten-minute explainers.
The firm name is usually in the title. When it is, the Overview has the firm’s name next to the answer it just gave.
What is missing from the list is production. These are attorneys answering a question to a camera. Nothing about them suggests a budget.
Which questions bring up a video
The cited videos cluster around a small set of situations.
What to do next. After a crash, after a hit and run, when the other driver lies, when the insurer calls.
How the rule works. No-fault, comparative fault, how long a settlement takes, how long you have to report.
Whether to settle. Five of the 40 are some version of should I accept the first offer.
Choosing a lawyer. What to ask, how to check an attorney’s bar status, how to choose in a named city.
Case types. Pedestrian, motorcycle, truck, rear-end, back and neck injuries.
Three videos from one of the tracked firms are in the top 40, with 40 citations between them. A Buffalo car accident FAQ, a Rochester Short on whether the rear driver is always at fault, and an Albany Short on back and neck injuries. Each one names the city and answers one question. That is the pattern to repeat across the rest of the 111 questions.
How to make the video Google cites, and sources
One question per video, taken from what clients ask. Our list of 111 questions is a starting set. Use the client’s wording as the title.
Say the state and the city. In the title, in the first sentence and in the description. A general answer competes with every firm in the country. A New York answer competes with a handful.
Put the firm name in the title and answer in the first minute. The description should restate the answer in two sentences and match the page on the firm’s site that answers the same question, so the two sources agree.
Record the short version too. A Short of the same answer is a second source for the same question.
Measure AI Overviews on their own line. A blended visibility number would have hidden that this firm’s weakest engine is the one that reads video.
The full method for this engine is in How a law firm gets named in Google AI Overviews and AI Mode. How an engagement runs is on our law firms page, and every figure we publish is on the numbers page.
Sources
SourceNode tracking: 111 questions, 14 New York motor vehicle accident firms, ChatGPT, Gemini and Google AI Overviews, daily, 9 October to 7 November 2026. About 2,390 Google AI Overview answers, 1,081 YouTube citations, and the 40 most cited videos by citation count.
Share of voice by engine from the same tracker, 30 days to 17 October 2026.
The researcher
Odin Lund is co-founder and COO of SourceNode and leads its data analytics and research, including the daily tracking behind this work across ChatGPT, Gemini and Google AI Overviews. Odin founded First Query, the AI search practice that became SourceNode, and co-founded The Pixel Paper.
Meet the research team
About this research
Published by SourceNode
SourceNode is a data and research-first AI search firm. We record what AI models say about a business every day, trace each claim to its source, and use that data to get businesses named in AI answers. Every figure we publish carries its method and its sample.
Share Blog
Related research
AI search
AI search statistics for 2026: the numbers we trust, with sources
Read the research
AI search
What McKinsey’s AI search research says, and what to do about it
Read the research
