YouTube SEO: The Playbook for Ranking on YouTube, Google, and AI

Three months ago I launched a YouTube-first SEO strategy for Humble&Brag's own site. We went from basically zero to over a quarter million impressions every week. On YouTube, the same trajectory: growing week on week, even when we didn't publish.

And something else started happening. AI tools began surfacing our content in their answers. Claude, ChatGPT, Perplexity, all citing our articles and videos directly.

On the keywords that matter most to us, the search results page looks like this: our content appears in the AI Overview, in the standard organic results, in the video carousel, and in the image results. Most of the links on the page are ours. And we built that position in 90 days, without any paid media.

The system behind this result is what I'm going to walk you through. It works in three steps, each one stacking on the last. The companies that scale YouTube and SEO in unison outperform everyone else in their niche, and the reason most businesses don't do it is that they're missing the foundational knowledge that makes it all work.

The Three Engines Driving YouTube Views

Every view on YouTube arrives through one of three mechanisms, and I want to name them in a way that actually describes what they do, because what YouTube calls them in Analytics doesn't quite capture the behaviour. We covered the detailed mechanics in our traffic sources guide, but here's the strategic picture.

The recommendation engine (Browse Features). Someone opens YouTube and the algorithm surfaces your video to them, typically on the home page. They weren't looking for you. The algorithm brought you to them, based on their watch history. This is where the viral moments happen: if a video takes off through Browse, you can see thousands of views in a day and a sudden peak.

The session engine (Suggested Videos). The user is already watching. They're in the rabbit hole. YouTube serves them videos that are thematically close to keep them on the platform. The user isn't starting a new journey; they're extending one already in motion, whether through the sidebar or through end screens.

The search engine (YouTube Search + Google Search). For me, both sit together even though they're separate in YouTube Analytics. Why? Because for the last decade, I've found that if you rank well in YouTube Search, you'll be found through Google Search too, and often this Google traffic becomes one of your most significant and most predictable view sources.

Here's what the traffic breakdown looks like on our own long-form content over the past year. Browse Features and Suggested Videos spike together when a video takes off, then fall back when the momentum runs out. They reward recency and consistent posting. But if you go quiet, views peter out.

Now look at the search line. It doesn't spike and crash. It just grows, slowly and reliably. That quiet persistent line is what we're going to build, in three steps.

Step 1: Rank in YouTube Search

Most companies think YouTube SEO works like Google SEO: build domain authority, get backlinks, earn your ranking over time. But YouTube rewards relevance much more than Google does.

Relevance comes from one signal more than any other: whether your title contains the exact phrase the viewer just typed. That's the whole mechanism. If your title matches the search query, you're immediately more relevant than any video that doesn't, regardless of channel size, view count, or how long you've been posting.

Your video still needs to deliver on the promise in the title (YouTube measures that through retention, which determines whether you stay ranked). But this first step of choosing keyword and title is critical, and it's why a channel with 500 subscribers can outrank one with 500,000, just by being a more relevant result for that specific query.

So the question isn't "how do I build authority?" It's "which keywords can I claim right now, today, with the channel I have?"

The Keyword Research Process

A good keyword has three properties: low competition, reasonable volume, and what I call "platform proof." Here's the process.

Start with a broad, relevant keyword for your industry. Use a tool like VidIQ or Ahrefs to check volume and competition level. Anything that's not low competition is going to be difficult for a smaller channel.

Once you've found something with good volume and low competition, copy the phrase and search it on YouTube itself. This is the platform proof: review the videos that appear. Are they highly relevant? From established channels? Does the existing content already meet the viewer's needs thoroughly? If so, dig deeper into a more specific variant. The goal is a keyword where the existing results leave room for a better answer.

Finally, check the growth curves and views-per-hour of similar videos in the results. Are they spiking on release and then plateauing (which suggests views are coming from Browse and subscribers), or getting consistent views every day (which suggests search traffic)? If it's the second, you've found a good keyword.

Alternatively, rather than starting with platform metrics, you can start with your customers' interests and pain points. Often, these terms have lower volumes but also lower competition and much more specificity to your business. Remember, as a business, you care about conversion, not just views. This is exactly the approach we took when we targeted "youtube strategist": we knew we wanted to own the term, and now we do.

The Five-Point Upload Checklist

Once you've filmed a video on a validated keyword, cover five bases on upload.

Title: Make sure it's clear what the viewer is going to get. Your keyword should be in the title. Don't ambush the viewer's curiosity here; recognise their intent. Keep it under 60 characters and consider adding a short descriptor at the end to set expectations.

Description: Keyword in the first line, adjacent terms woven naturally throughout. Write for humans, but keep the machines in mind.

Timestamps: Add chapters with descriptive, keyword-aware labels. These can be surfaced as snippets across Google and YouTube Search independently.

Thumbnail: Create a thumbnail that spikes curiosity and stands out from the competing results.

Tags: Generate tags for your video. Tags aren't critical, but the more metadata YouTube and Google have, the better, particularly when the platform is first indexing your content.

The goal of Step 1 is to build that search baseline I described earlier: the quiet line underneath the peaks and crashes. It takes a while to get going, but once it does, it's hard to stop.

Step 2: Turn Every Video into a Google SEO Asset

When Calum and I were at CareerFoundry, we did something embarrassingly simple but effective. We took a blog post that was already ranking for "what is data analytics," with strong traffic and a solid position on page one. And we created a YouTube video on the same topic and embedded it directly into the post. The video was basically the blog post in video form.

Pretty much overnight, the article moved up on page one and the video had views from a traffic source it had never seen before. We hadn't changed a word of content, hadn't built a single backlink, hadn't touched the metadata. We just connected two things that were already working, and they boosted each other from day one.

That's the flywheel. And most companies never build it because they treat YouTube as a standalone channel.

What's Actually Happening

Three things happen simultaneously when you embed a video in a blog post.

First, the embedded video keeps people on the page longer, which Google reads as a strong satisfaction signal. Google internally calls this a "good click" or "last long click," where the user clicks your result, stays on the page, and never goes back to search. A page with an embedded video achieves that in a way text alone almost never does.

Second, your content surface area expands. The video title, description, chapters, and full transcript all become indexable signals. Google can understand and surface your content across queries your article alone would never have ranked for.

Third, it compounds. Search traffic that finds your blog post watches the embedded video, which feeds back into YouTube's signals for your channel. SEO drives YouTube, YouTube drives SEO, and once the flywheel is turning, each channel makes the other stronger with no extra production.

The practical loop: produce a video, write one or more corresponding blog posts, embed the video near the top of the post, and link to the post in the video description. That's the full cycle.

The Schema Decision

This is where most teams stop, and where one small technical step makes a large difference to visibility. I brought in Nadia Mohamed, a technical SEO and GEO specialist, for this advice, and the first thing she flagged is the mistake she sees most often: teams forget the schema.

The question is: which role does this video play on the page? If the video is the main content, VideoObject schema must be added. That's what allows Google to index the video independently and enable features like Key Moments. If it's not the main content (for example, it's a supporting video in a longer article), adding VideoObject creates confusion because you'll have two schemas conflicting. Instead, add the video as a property to the main BlogPosting schema. The video won't be indexed separately, but it will still provide value to the page.

Google's video SEO documentation covers the full structured data requirements. If you're on WordPress, plugins like RankMath or Yoast handle this. If you're on a headless CMS like Framer, it's a small component build. Either way, it's 30 minutes of development time that changes how Google understands your content.

Step 3: Appear in AI-Generated Answers

Here's how most companies build their content. They pick a topic: "What is data analytics," "Why you need YouTube," "Our approach to SEO." They write the article, make the video, post it, and hope it ranks. And it might rank, on Google, even on YouTube. But it gets cited by AI tools in almost nobody's answers, because AI tools don't summarise topics. They answer questions.

That distinction, between a topic and a question, is often what separates content that shows up in AI Overviews from content that doesn't.

What Gets Cited

Three characteristics determine whether AI tools cite your content. First, authority: genuine expertise rather than generic overviews, demonstrated through experience, proof, and original research. Second, specificity: directly answering a defined question. And third, structure: organised into logical sections an AI can extract on its own.

Most companies have the authority part internally but fail to show it externally, and very few are systematic about releasing specific, structured content.

How AI Tools Actually Read Your Content

Nadia Mohamed explained the mechanism: "AI tools don't read content the way we do. They don't go paragraph by paragraph. They look for very short chunks of text with the answer in the very first paragraphs. If they don't find the answer straight away, they move forward. So having question-based headings and giving the answer right at the beginning of each section is super important. FAQs work really well for this, both in the content itself and as structured data."

It's basically the inverted pyramid principle from journalism. Answer first, expand after. And this is exactly how LLMs parse content.

You can give your video a structural advantage by revisiting your chapters. Name every chapter as a question. Not "Introduction" or "Section 2," but "How to validate keyword demand on YouTube" or "Why embedding video lifts your Google rankings." Those chapters become independently indexable, and can surface in AI answers and Google snippets as standalone results.

Why AI Traffic Converts Better

Nadia's data reveals a secondary benefit: "AI traffic converts way better. The user clicking on a citation has already done their research through the tool. They're at a point in the funnel where they're ready to buy."

Pew Research found that when Google shows an AI summary, users click a result only 8 per cent of the time, compared to 15 per cent without a summary. But those 8 per cent are high-intent clicks. The traffic is smaller, it converts better, and there's not much competition for it yet.

Watch Pages and SeekToAction Schema

For companies that want to go further, watch pages and SeekToAction schema on your website are the next step. Watch pages are posts where the video is the primary content, presented above the fold and accompanied by a formatted transcript with headings that mirror the chapter titles. This gives AI tools a textual map of your content, and it's one of the ways LLMs identify you as a primary source.

SeekToAction schema properties let your website, not just your YouTube channel, claim chapter timestamps as rich results in Google search. It's the difference between Google surfacing your YouTube URL and surfacing your own domain with your specific chapter as the answer. Google's structured data documentation covers the full implementation.

The System in Summary

Step 1: rank in YouTube Search. Match the keyword, earn the relevance signal, lead with informational content, and build the search baseline that compounds.

Step 2: turn every video into a Google SEO asset. Write the blog post, embed the video, get the schema right, and let the flywheel run.

Step 3: optimise for AI. Structure your content around questions, answer first then expand, publish a structured transcript.

SEO hasn't died. It's just moved. And YouTube is now one of the most powerful tools in your arsenal for appearing across all three search surfaces simultaneously.

If you want help building a YouTube SEO system that gets your content found across YouTube, Google, and AI tools, book a call with us.