YouTube Is Now a GEO Signal: How Contractor Videos Get Cited by AI Engines
When a homeowner asks Perplexity “How do I know if my furnace heat exchanger is cracked?” the AI does not just scan contractor websites. It watches videos. According to OtterlyAI’s 2026 YouTube Citation Study, which analyzed over 100 million AI citation instances across ChatGPT, Google AI Overviews, Perplexity, and Microsoft Copilot, YouTube is the second most-cited social platform in AI search, trailing only Reddit. On Perplexity specifically, YouTube accounts for 22.7% of all social platform citations.
Most home service contractors have no YouTube presence. No how-to videos, no diagnostic explainers, no job walkthroughs. That absence costs them citations from AI engines that are actively looking for video content to support their answers about exactly the trades and services contractors offer every day.
What the Data Shows
Perplexity drives 38.7% of total YouTube citations across AI platforms. Google AI Overviews drives 36.6%. ChatGPT contributes only 4.4%. The practical implication: if you want YouTube citations from AI engines, Perplexity and Google AI Overviews are the surfaces that matter. Both are pulling from YouTube at scale, and both answer home improvement and service queries constantly.
Two platform behaviors that directly shape how you should structure your content:
- Google AI Overviews primarily cite timestamped, chapter-structured videos. Of all timestamped YouTube citations in the dataset, 73% appeared in Google AI Overviews. Videos with chapters were cited 78% of the time across multiple segments, meaning one structured video generates several distinct citation opportunities.
- Perplexity cites broader video content without requiring chapter structure. It pulls from video titles, descriptions, and transcripts when generating local and informational answers, which makes description quality especially important for Perplexity visibility.
Why Popularity Does Not Drive Citations
The OtterlyAI study found something most contractors would not expect: views, likes, and subscriber count have near-zero correlation with AI citation frequency. A video with 200 views and clear chapter structure gets cited more often than a video with 20,000 views and no structural markers. AI engines are not looking for popular content. They are looking for well-organized, reference-quality content that answers a specific question clearly.
Citation volume by video length from the study:
| Video Length | Share of AI Citations |
|---|---|
| Under 5 minutes (including Shorts) | ~6% |
| 5 to 10 minutes | 26.1% |
| 10 to 20 minutes | 32.1% |
| Over 20 minutes | ~36% |
Long-form videos account for 94% of all AI citations. Shorts rarely appear. For contractors, a 10 to 15 minute video explaining how to identify AC failure signs before calling a technician is a stronger GEO asset than a 45-second clip of a finished job. AI engines cite reference content, not promotional highlights.
What Home Service Contractors Should Film
The highest-value content for contractor GEO falls into two categories: diagnostic explainers and seasonal preparation guides. Both match the exact queries that drive AI Overview responses and Perplexity citations for home services.
Diagnostic explainers walk a homeowner through how to identify a problem before calling a professional. A plumber explaining five signs of a failing water heater, an electrician walking through what a breaker that keeps tripping usually means, a roofer explaining the difference between storm damage and normal wear: each of these answers a specific question homeowners type into AI engines. The video does not need production quality. It needs clarity and structure.
Seasonal preparation guides have a predictable, recurring search cycle. Pre-winter furnace prep, pre-summer AC inspection, fall gutter cleaning before freeze, spring roof assessment after winter. An HVAC contractor who publishes one solid furnace prep video in early October has an asset that gets cited by Google AI Overviews every time the seasonal query peaks, compounding in value each year it stays live.
Additional high-citation video formats for contractors:
- Cost breakdowns: "What does AC replacement actually cost in 2026" is a question homeowners ask AI engines constantly. A contractor who answers it on video owns that citation space.
- DIY boundary guides: What a homeowner can safely check themselves vs. what requires a licensed technician. This content builds trust and filters leads before the first call.
- Material comparisons: Tankless vs. traditional water heaters, asphalt vs. metal roofing, spray foam vs. batt insulation. Decision-stage content that AI engines pull from when a homeowner is evaluating options.
- Process walkthroughs: What happens during a furnace inspection, step by step. Demystifying the service visit reduces friction for homeowners who have never hired a contractor for that service.
Structuring Videos for Maximum AI Citations
Only 31% of the videos cited in the OtterlyAI study had chapter or timestamp structure. That means 69% of cited videos were earning citations without being optimized at all. Contractors who add structure sit in a minority that gets disproportionately rewarded by Google AI Overviews specifically.
Three structural changes that increase citation surface area:
- Add YouTube chapters via timestamps in the description. For a water heater video: "0:00 Introduction, 1:20 Sign 1: Inconsistent Hot Water, 3:45 Sign 2: Rumbling or Popping Sounds, 6:10 Sign 3: Visible Rust or Corrosion, 8:30 When to Call vs. When to Wait." Each chapter becomes a separately citable segment in Google AI Overviews. Among timestamped videos in the OtterlyAI dataset, 78% were cited across two to five chapters from a single video, which multiplies the citation surface from one recording session.
- Write descriptions that function as metadata, not marketing copy. Start with a one-sentence answer to the video’s core question. Then list what the video covers, point by point. AI engines parse the description when deciding what the video is about and whether it answers the query. A description that says "Licensed HVAC technician explains five warning signs that your furnace heat exchanger may be cracked, including what sounds to listen for, which visual indicators matter, and when the problem requires emergency attention" gives the AI a citable summary it can use independently of the video itself.
- Enable captions and review them for accuracy. Perplexity pulls from video transcripts when generating local and informational answers. YouTube auto-generates captions; review them in the first week after upload to catch trade-specific terms the auto-transcription gets wrong. A clean transcript makes the video’s spoken content fully indexable as text.
The Dual-Asset Play: One Recording, Two Citations
A video you publish on YouTube produces a second citable asset if you repurpose the transcript as a written article on your website. A 12-minute video generates roughly 1,500 to 2,000 words of spoken content. Clean that transcript, add headings that match your YouTube chapters, and publish it as a blog post or service page article.
The result: the same information becomes citable from two distinct sources. Your YouTube video reaches AI surfaces that prefer video citations, primarily Perplexity and Google AI Overviews. Your website text reaches AI surfaces that prefer written citations, primarily ChatGPT. A contractor who runs this for five diagnostic or seasonal topics has ten citation assets from five recording sessions. Most competitors have zero from either source.
Local Signals in Your Video Content
AI engines answering location-specific queries look for geographic signals in your video content. Include your city and service area in your video title when natural ("How to Winterize Your Plumbing in [City]: What to Check Before the First Freeze"), in your description, and in your spoken content. A homeowner asking "Who should I call for furnace repair in [city]?" triggers a query where AI engines weight local authority signals. A video where a contractor says "we serve the [metro area] and see this problem frequently in older homes in [neighborhood]" is a local citation in video form, independent from anything on your website or Google Business Profile.
Three Steps to Start This Week
- Film one diagnostic video. Pick the question homeowners ask most often before calling you. Film a 10 to 15 minute answer with your phone. The production bar is low: steady camera, decent lighting, clear audio. Do not script it. Answer the question the way you would explain it to a homeowner standing in front of the problem.
- Structure it with chapters before uploading. Write out timestamps at one to two minute intervals. Each label should be a standalone phrase describing exactly what that segment covers. Add these to the description on day one. Do not upload first and optimize later: AI platforms index new content quickly and the structure should be in place from the first crawl.
- Publish the transcript as a blog post. Download the auto-generated captions from YouTube Studio after the video processes. Clean the text, add paragraph breaks and headings that match your chapter timestamps, and publish it to your site. You now have two citable assets from one recording session, covering two different AI surfaces from the same afternoon of work.
YouTube does not require a following to generate AI citations. The OtterlyAI data is clear on that: views and subscribers have no meaningful correlation with citation frequency. A contractor who records three structured explainer videos in May has more AI citation surface area than a competitor with zero videos and a perfect Google Business Profile. The investment is one afternoon of filming and a few hours of cleanup. The citation value runs indefinitely, and it compounds each time an AI engine reaches for a video source to answer a home service question in your market.