Agentic SEO Tools
All posts
By Agentic SEO Tools Teamyoutubeai-overviewsvideocontent-agentsdata-reports

Video's share of AI Overviews citations from Q1 to July: YouTube script agents

Video's share of Google AI Overviews citations more than doubled between January and July 2026. An agent workflow from prompt gaps to video scripts to citation tracking.

About 2.7% in January. About 5.4% by early April. Over 6% by late July. That is video's share of Google AI Overviews citations across two of Promptwatch's content type reports, and it is one of the steadier climbs in their data this year.

Each figure comes from a specific report and window, so here they are with their sources. The Q1 2026 AI Overviews content type report covers January 1 to April 5, 2026 (its URL says March, its title says Q1). It found video "roughly doubled" over that window, from about 2.7% to about 5.4% of classified citations. The July 2026 report, covering July 1 to 31, says video reached "almost 6.3% by the end of July" and sums the trend as "over 6% by late July, up from ~2.7% in January."

What else moved around it

Video didn't grow in a vacuum. In the Q1 window, listicles were the biggest AI Overviews format at around 26% of daily citations (Q1 report, January 1 to April 5, 2026). By July they averaged about 18%, and from July 28 on, product pages were the most cited format every day, ending the month at 17.9% against 16.2% for listicles (July report, July 1 to 31, 2026). The July page says listicles "lost the top spot in Google AI Overviews for the first time."

Promptwatch line chart titled "What Type of Content Gets Cited by AI Overviews in July 2026?", showing the daily share of each content type, with a cluster of four lines near the top, two of which cross near the end of the month, and a separate lower group of lines Promptwatch's "What Type of Content Gets Cited by AI Overviews in July 2026?" chart for July 1 to 31, 2026, captured November 14, 2026.

The chart shows two bands. The big formats sit at the top and the two leading lines cross in the last days of July. Video sits in the lower band with social posts and landing pages. A few points of share, in other words, but growing.

A note on terms. "Video" in these reports is a content type, the classified format of the cited page. That's not the same as YouTube's share of AI Overviews citations as a domain, which Promptwatch measures in its social media reports. The two numbers answer different questions, so don't add them or compare them directly.

What the cited videos look like

Promptwatch's YouTube citation data for February 2026, covering February 1 to 28, 2026, profiles the videos that ChatGPT, Perplexity, and Google AI Overviews cited. That month YouTube earned 2.25% of AI Overviews citations and beat Reddit (1.65%) as its top social source. ChatGPT barely cited YouTube at all (0.05%).

The profile of a cited video is not what a YouTube growth team would expect. Nearly 80% of cited videos had under 100K views, and the largest view bucket was 10K to 100K, at 36%. About 80% of citations came from channels under 100K subscribers, and channels with 1M or more earned 7% (YouTube February 2026 report, same window).

Age matters even more. 54% of cited videos were over two years old and nearly 72% were over a year old. About 4% were under a month old (YouTube February 2026 report).

Promptwatch bar chart titled "How Fresh Are the YouTube Videos Cited in AI Search?", with the longest bar for videos two or more years old and much shorter bars for every younger age bucket Promptwatch's "How Fresh Are the YouTube Videos Cited in AI Search?" chart for February 1 to 28, 2026, captured November 14, 2026.

Our read: AI Overviews cites videos that answer a specific question and have been around long enough to be found, not videos that went viral. The February page suggests engines match on titles, descriptions, and transcripts, but that's the page's own inference, not something it measured. Either way, it points the same direction: a focused video with the question in its title and the answer near the start.

The agent workflow

A "YouTube script agent" sounds like one tool. In practice it's four steps, and only some of them should be automated.

Step 1: find the prompts where video gets cited

Start with your tracked prompts in Google AI Overviews. Promptwatch's YouTube citation report lists every video cited in answers to those prompts, with views, engagement, and the channel behind it. Citation Trends classifies citations by content type, so you can see which of your prompts lean on video and which don't. A prompt where AI Overviews cites three videos and none of them are yours is a candidate.

The "My channels" filter separates your own videos from everyone else's, so you know which candidates you already cover.

Step 2: write the brief from what's already cited

For each candidate prompt, open the cited videos and note what they answer, how long they take to get to it, and what they leave out. That's the brief. Agent Chat can do the first pass over the YouTube report, grouping cited videos by prompt and summarizing what each covers, and the person who will make the video reads the summaries and picks the angle.

Step 3: draft the script

This is the step where a writing agent earns its keep. Give it the brief, the prompt word for word, and your product facts, and ask for a script that states the question in the first line and answers it within the opening. Promptwatch doesn't produce or publish video. Its Content Creation can pull YouTube transcripts into a brief for written content, which is useful for the article that goes alongside the video, and for checking what competitors' videos actually say.

A person records the video. Generated video and voices are possible, but nothing in Promptwatch's data says engines cite them, and a brand's credibility is easier to lose than to rebuild.

Step 4: track the video like a page

Promptwatch Academy playbook titled "Reddit and YouTube presence strategy", explaining why AI models cite Reddit threads and YouTube videos and how to see which ones feed your category's answers Promptwatch Academy's Reddit and YouTube presence strategy playbook, captured November 14, 2026.

Once the video is live, track it in the YouTube citation report under My channels and add the video URL to Page Tracker. Given the age profile above, expect this to take a while. Only about 4% of cited videos in February were under a month old, so a new video that isn't cited in its first weeks hasn't failed.

One customer example

Promptwatch's own case study for Crisp describes a citation-driven focus on Reddit and YouTube, alongside scaling to five to ten articles a day. It also reports that AI-referred traffic converted at twice the rate of traditional channels for Crisp. That's one named customer's result, not a forecast for anyone else, but it shows the pattern of picking channels from citation data rather than habit.

Where this fits on an agentic stack

Content agents in our catalog, like Outrank, embed YouTube videos in articles, per its catalog entry. That's useful for the written companion piece but it doesn't tell you which videos AI Overviews cites. The decision of what to film needs the citation side, which is where Promptwatch leads, ahead of AthenaHQ and the publishing-only tools. For more on LinkedIn and YouTube as citation sources, see what an agent should do with LinkedIn and YouTube citations.

The loop

The finding: video went from about 2.7% of classified AI Overviews citations in January to about 5.4% by early April (Q1 report, January 1 to April 5, 2026) and over 6% by late July (July report, July 1 to 31, 2026), and the cited videos were mostly small and old. The action: find the prompts where video is cited and you're absent, brief from the cited videos, script with an agent, record with a person, and publish the written companion. The measurement: My channels in the YouTube report, Page Tracker on each video URL, and AI Overviews visibility on those prompts, given a few months.

Start the 7-day free trial on monthly billing and open the YouTube report filtered to Google AI Overviews. The list of cited videos is your first script backlog.