YouTube stopped being a place people visit just to kill time a while ago. It's the second-largest search engine on the internet, right behind Google itself. If you're putting out content there and simply hoping the algorithm notices, that approach runs out of steam fast.
What actually moves the needle is YouTube SEO — a deliberate set of practices that help your videos surface in search, get pulled into recommendations, and earn real visibility over time. This guide breaks down exactly what that means in practice and how to apply it step by step, whether you're starting from zero or trying to revive a channel that's gone quiet.
What AI Overviews Actually Are
An AI Overview is a generated summary Google places above the traditional results for a query. It reads like a written answer rather than a list of links, and it carries inline citations to a handful of the pages it drew on.
The important detail is that it is generated, not retrieved. Google is not picking the best page and showing you a snippet of it. It is reading several indexed pages, synthesising an answer, and attributing some of them. That distinction changes what you are competing for: not a position, but inclusion in the set of sources the answer gets built from.
They do not appear on every query. Navigational searches, transactional searches and anything where Google judges the risk of a wrong answer too high tend not to trigger them. Informational queries — the how, what and why searches that most blog content targets — trigger them most often.
How Sources Get Chosen
Citation does not map neatly onto ranking position. Pages ranking fourth or seventh are cited regularly while the page at position one is left out. The selection runs through a retrieval step that looks for passages which answer the specific question, then a generation step that assembles them.
Three things appear to matter most. The first is extractability: whether a passage makes sense lifted out of its page. A paragraph that depends on the three paragraphs above it to be comprehensible is a poor candidate. The second is factual clarity — explicit statements beat implication. The third is corroboration: whether the claim is supported elsewhere on the web, because a system that has to stand behind its answer prefers sources it can verify.
None of this is exotic. It is a stricter version of what has always made a page useful. The difference is that a human reader forgives vagueness and a retrieval system does not.
What This Does to Your Traffic
The pattern to watch for is impressions holding steady while clicks fall. You are still being shown. Fewer people need to click, because the answer arrived before they got to you.
This hits informational content hardest and commercial content least. A page explaining what a term means is far more exposed than a pricing page or a service page, because the first can be summarised in three sentences and the second cannot.
It is worth being honest about the implication: some of your traffic is not coming back. The realistic response is to shift what you measure on those pages — from sessions to impressions and brand visibility — and to make sure the content that can still earn a click is doing so.
Writing Content That Can Be Extracted
The single highest-leverage change is to answer each question completely in one self-contained passage, immediately under a heading phrased as the question.
That means resisting the build-up. If the heading asks how long SEO takes, the first sentence beneath it should say how long SEO takes. The context, caveats and detail can follow. Writing that defers the answer for four paragraphs reads fine to a human skimming and is nearly useless to a retrieval system deciding whether this passage answers the query.
Be explicit with facts. Write the number, the date, the name. Systems that must attribute a claim prefer sources that state it plainly over sources that gesture at it.
Keep passages independent. If a sentence only makes sense after the one before it, it will not survive being lifted out. This is the opposite of how most long-form content is written, which is why most long-form content is not cited.
Structure and Schema
Clear heading hierarchy does real work here. It is how content gets segmented into retrievable chunks, and a page with one H1 and eight well-titled H2 sections segments far more cleanly than a page with decorative headings that do not describe what follows.
FAQ markup is the most direct signal available, because it hands the system a pre-formatted question and answer pair. It has to match what is visible on the page — markup describing content the user cannot see is a policy violation, and it is more common than most site owners realise.
Organisation and Article markup matter for a different reason: they establish who is speaking. A system deciding whether to cite you benefits from being able to identify the publisher as a recognised entity rather than an anonymous domain.
Authority Signals Still Decide
Structure gets you eligible. Authority decides whether you are chosen over the other eligible sources.
Named authors with verifiable expertise, citations to primary sources, consistent naming across the web, and coverage from publications the system already trusts all feed the same judgement: is this a source worth standing behind?
This is also where most sites lose. An anonymous post on an unattributed domain can be perfectly accurate and still lose to a worse-written page from a publication with an established identity. That is not unfair — it is the system managing risk in exactly the way you would if you had to put your name to the answer.
Measuring What Is Happening
Start in Search Console. Compare impressions against clicks for your top informational queries over the last few months. A widening gap between the two is the clearest available signal that answers are being served above you.
Check the queries themselves. Search your priority terms and note which trigger an overview, which sources get cited, and whether you are among them. This is manual and tedious and there is no reliable shortcut.
Then segment. Informational pages and commercial pages are now behaving differently enough that a site-wide traffic number hides more than it reveals.
What to Do About It
Do not panic-rewrite the whole site. Start with the pages that already rank well on queries that now trigger overviews, because those are where you are closest to being cited and have the most to lose.
On those pages, restructure so each question is answered directly beneath a question-phrased heading, add FAQ markup that matches the visible content, and make sure the page states who wrote it and why they are worth believing.
Then accept the shift in what success looks like. On genuinely informational queries, being the cited source with fewer clicks may be the best outcome available — and it is considerably better than being neither cited nor clicked.
Frequently Asked Questions
An AI Overview is an AI-generated summary Google displays above the traditional search results. It synthesises information from several indexed pages and cites some of them inline, so it answers the query directly rather than returning a list of links.
For informational queries, usually yes. The pattern to look for is impressions staying flat while clicks decline, which means you are still being shown but fewer people need to visit. Commercial and transactional pages are far less affected.
Answer each question completely in a self-contained passage directly beneath a question-phrased heading, state facts explicitly rather than by implication, add FAQ schema that matches the visible content, and build the authority signals that make a source worth citing.
No. Citation does not map neatly onto ranking position. Pages ranking fourth or seventh are cited regularly while the top-ranked page is left out, because the selection depends on passage relevance and source trustworthiness rather than position alone.