How to show up in AI search results with B2B video

Short answer: Take the verbatim questions your buyers already ask, answer each one deeply in a long-form video, publish it on YouTube with the transcript and schema on your own site, and then measure three things: watch duration, AI citation, and how many buyer questions you have answered. When I tested 150 buyer questions across engines, Google AI mode cited video in 92% of answers and Perplexity in 94%.

Why this matters now

The buyer journey has changed. Buyers now want to get about 80% of the way through the funnel before they are willing to talk to a salesperson or take a sales call.

That means they are educating themselves. They are asking AI. They are searching Reddit, Quora, LinkedIn, and YouTube. They are asking their friends.

So the question is simple. When they search, are you being found? Are you the one with the content they watch and read? Or are you getting summarized without credit, left out of the answer entirely?

I was talking to a sales leader in B2B SaaS recently who said his outbound process felt like crawling through glass, and they had essentially no inbound. That is the position this process is built to fix.

Why video specifically

Most tech companies already have content. Blogs, podcasts, webinars, sometimes video. The problem is that it usually sits outside the funnel buyers are actually searching, and it answers the things the company wants to say rather than the questions buyers are asking in their own words.

Text answers have a particular weakness right now. When a buyer asks a question, the AI overview summarizes the text and often takes the idea without crediting the source.

Video behaves differently. In my test of 150 buyer questions across engines, video was cited in 92% of Google AI mode answers and 94% on Perplexity.

There is a straightforward reason. Google owns YouTube, and people watching YouTube videos is how Google earns ad revenue. Pushing video into AI search results is in their interest, and it only works if people actually watch.

ChatGPT is the exception. It does not really cite video. The workaround is to put the video on a page on your own website along with the transcript, the text, and schema markup. That gives ChatGPT a text surface to cite. Every video I post also goes out as a blog post for exactly this reason.

Where the content comes from

Start with your transcripts.

If you have sales call transcripts, you have a goldmine of literal quotes from buyers about the problems and questions they have. That is the raw material.

Two things matter about how you use it:

Use their words, not yours. Buyers search in their own language. When my buyers talk about this work, they say "video partner," "boutique video partner," "production capability on site." That is how they think about it, so that is the language the content has to use.
Catalog the questions verbatim. Real examples from my own list: "How do typical structured engagements work?" and "Is long-form video actually the right first move, or should I be doing something else given how early I am?"

Those questions become your to-do list. Each one is a video.

The five long-form rules

These are the Digital Accomplice YouTube long-form rules:

Always use buyer language.
Always skew deeper.
Always show and tell.
Always use examples.
Always play the YouTube engagement game well.

LLMs and people both want answers, but they do not want bad answers. If you create the best answer in video format, you have a much higher chance of being cited. If you go deeper than anyone else, better still. And if you post frequently, you signal to both people and LLMs that you have the latest information, which matters because this space goes stale fast.

The process, step by step

1. Research

Pull the verbatim questions out of sales transcripts and calls. Some are direct quotes. Some are inferred from context. Keep them in one list.

2. Plan

Decide which questions this video answers. Then decide:

What steps do you walk the viewer through? If you want to play the engagement game well, you cannot just ramble. Walk them through a process.
What format fits? Show and tell, checklist, Q&A.
Is there a call to action, and where does it go?
What examples will you use?
Who does what: research, planning, recording on camera, posting, measuring, and deciding what to implement based on the measurement. That can be one person or six. Naming it is part of the plan.

3. Build and record

Perfect is the enemy of progress. This is not cinema. It is educational content, essentially a seminar or a training. Getting caught up in lights, cameras, and graphics slows you down, and consistency and depth matter more than polish.

Here is my actual setup. I use a Miro board as a guide through the topics, arranged the way you would take a buyer from early questions to late questions. I am not reading off a script. I record in one long take using Riverside with screen share. I chop the head and the tail off the edit, add captions, and export. That is it.

4. Post and optimize

The content has to live on YouTube and be optimized to get cited, found, and clicked. Each surface does a different job:

Thumbnails get you discovered.
Titles earn the click.
Descriptions are citable and crawlable by AI.
Transcripts are material for getting found.

Then run the YouTube A/B test. You can test three different thumbnails and three different titles at once. The test seems to take between two days and two weeks, and the winner is selected automatically and stays live.

5. Measure

Three measurements tell you almost everything:

Watch duration. Are they watching it through? Where do you lose them? I am getting about three minutes of retention right now, which is not ideal but is better than it was.
AI citation. Run tests to find out whether you are showing up. If you are not, make content to fix it.
Buyer questions answered. How many items on the list are checked off? That tells you how much content you actually have working for you.

Measure on two clocks. Check the next day for lessons you can apply to the next video. Check again after two weeks, once the A/B test has finished running.

6. Iterate

Identify the improvements and implement them. Every stage is a place to iterate: planning, building, recording, posting. Each round, ask what you learned, what you want to change, how to increase watch time, and what new buyer questions have surfaced.

This is about reps and learning, not about building the perfect thing and launching it.

What a finished pipeline looks like

Every question you answer plugs a hole in the funnel. The leaks stop. Buyers can get themselves to that 80% mark on their own before they take your call, so by the time they reach you they have already decided you are probably the right fit and only have a few specific pricing or implementation questions left.

Once you have answered the whole list, the machine for inbound exists. The leads at the end are better qualified because they have already seen the content, and YouTube and the LLMs are pushing that content toward buyers on your behalf.

Who this is for

This is for tech, cybersecurity, and B2B SaaS companies that need AI visibility to drive pipeline growth, especially teams whose outbound is grinding and whose inbound is thin.

Want help?

I am Dane Frederiksen at Digital Accomplice. This stuff is complicated and fast changing, and it has not been a thing for very long, but it is what my team and I do every day.

If you want help building this, reach out:

https://www.digitalaccomplice.com

Frequently asked questions

Does video actually get cited in AI search?

In my test of 150 buyer questions across engines, video was cited in 92% of Google AI mode answers and 94% on Perplexity. ChatGPT is the exception and does not really cite video.

How do I get cited by ChatGPT if it does not cite video?

Put the video on a page on your own website with the transcript, the text, and schema markup. That gives ChatGPT a text surface to cite.

Where do the video topics come from?

Sales call transcripts. They contain literal quotes from buyers about the problems and questions they have, in the language buyers actually use.

What should I measure?

Watch duration, AI citation, and how many buyer questions you have answered. Check the next day for immediate lessons, then again after two weeks once the YouTube A/B test has run.

How long does the YouTube A/B test take?

Between about two days and two weeks. You can test three thumbnails and three titles, and the winner is selected automatically.

Do I need high production value?

No. This is educational content, closer to a seminar than to cinema. Consistency and depth beat polish, and getting caught up in production slows you down.