Most advice about YouTube SEO still reads like it was written in 2016: pick a keyword, put it in the title, the description and the tags, then wait. This guide takes a different route. It goes through what YouTube says in its own documentation, what independent data shows, and where the honest answer is that nobody outside YouTube knows.
A note on how to read it. Where something comes from YouTube Help, I say so. Where it comes from a study run by a tool vendor or agency, I say that too, because those studies are mostly correlations and the companies behind them often sell something. And where I'm drawing a conclusion the data doesn't prove, I mark it as my reading, so you can test it on your own channel instead of taking it on faith.
YouTube SEO Has Changed
YouTube SEO in 2026 is less about telling YouTube which keyword you want to rank for, and more about helping it work out who a video is for and whether those people were glad they watched it. Keywords still matter. They just stopped being the whole job.
From Exact Keywords to Topic Understanding
YouTube Help describes search ranking in two parts: how well the title, description and video content match what the viewer searched for, and which videos drive the most engagement for that search. The phrase to notice is “video content.” The match isn't limited to the text boxes you fill in.
In practice, that means a video doesn't have to contain the exact phrase a viewer typed to be a candidate for it. A video that clearly explains how to get your first 1,000 subscribers can plausibly show up for searches worded differently from your title. That's my inference, but it's easy to check: open the YouTube search traffic source in YouTube Analytics and read the search terms. If you see queries that don't appear anywhere in your title or description, YouTube is already matching on topic rather than exact wording.
The practical consequence is that you write for the topic and the viewer's question first, and treat the exact keyword as the clearest way to name it, not as a magic string.
Why Metadata Alone Cannot Rescue a Weak Video
Metadata gets a video into the pool of possible results. What happens after that depends on viewers.
YouTube Help is direct about this: search results are not simply a list of the most-viewed videos for a query, and the recommendation system learns from what people watch, what they don't watch, how much of a video they watch, and whether they seem satisfied. If your title wins the click and the video loses the viewer in the first thirty seconds, better keywords won't fix it. The reverse also holds. A video that satisfies people but is packaged vaguely will have trouble getting found in the first place.
So think of title, description and tags as the entry ticket, and the video itself as what decides how long you stay in the room.
Search vs Suggested vs Browse
This distinction is essential, because “YouTube SEO” advice often treats YouTube as one system when it's at least three, and they reward different things. You can see how your own video splits across them in the traffic source report in YouTube Analytics.
| Surface | What YouTube says it looks at | What that means for you |
|---|---|---|
| Search | How well the title, description and video content match the query, plus which videos drive the most engagement for that query | Clear wording matters most here. The viewer already knows what they want. |
| Suggested (Up Next) | Which videos the viewer is most likely to watch next. Often related to the video they're watching, but can be personalized by watch history | Topic fit with neighboring videos, and a package that still makes sense at small size next to other titles. |
| Browse (Home) | How well the video has interested and satisfied similar viewers, plus the viewer's watch and search history | Packaging and audience match. The viewer didn't ask for anything, so the title has to earn attention. |
This is also the honest answer to the “YouTube search vs suggested videos” question people type into search. They aren't competing for the same traffic. A keyword-led title helps most where someone is typing a query. Where someone is scrolling, a keyword-first title can be a weaker hook than a promise or a curiosity gap. Some of the title data later in this guide points the same way.
What YouTube Actually Learns From a Video
YouTube doesn't publish a list of ranking signals or their weights, and anyone who gives you precise percentages is guessing. What follows combines what YouTube Help states with reasonable inference, and I'll flag which is which.
Title
Stated by YouTube: the title is part of how a video is matched to a search. YouTube also tells creators to use titles that accurately represent the content.
The title does two jobs at once. It tells the system what the video is about, and it tells a person whether to click. Those jobs sometimes pull in different directions, which is why a lot of title advice contradicts itself.
Description
Stated by YouTube: the description is also part of the match. Beyond that, YouTube doesn't say how much weight it carries compared to the title, so treat claims about its exact importance with suspicion. What is clear is that it's the one field where you can explain the video properly, link to related material and add chapters.
Transcript and Spoken Content
YouTube Help includes “video content” in the match, which fits with what you'd expect from a platform that transcribes speech. How exactly spoken words are used isn't documented in detail, so this is the least certain item on the list.
There are still sensible things to do with it. Say what the video is about plainly and early, using the words a viewer would use. Check the auto-generated captions for names, product terms and jargon, which are the places automatic transcription tends to slip. Vendors that sell transcription put auto-caption accuracy somewhere around 60 to 80 percent depending on audio quality, but they have a reason to make auto-captions look weak, so treat that range as rough.
Viewer Behavior
Stated by YouTube: the recommendation system pays attention to what viewers watch, what they don't watch, what they search for, likes and dislikes, and “not interested” feedback. It also learns from how much of a video someone watches.
Translated into creator terms: whether people click when your video is shown, whether they stay, and whether they react negatively. A title that overpromises produces exactly the behavior that the system is built to notice.
Engagement and Satisfaction
YouTube Help says likes and dislikes are some of hundreds of signals, and that the system tries to learn whether viewers were satisfied by what they watched.
You will see a lot of blog posts claiming that YouTube “confirmed” in 2025 or 2026 that satisfaction has replaced watch time as the main ranking signal. When I searched for it, I found those claims on vendor blogs and not in a primary source from YouTube, so I'd call it unverified. What YouTube Help supports is the narrower claim: satisfaction is something the system tries to measure, and watch time is one input among several.
Do YouTube Tags Still Matter?
The short answer is barely, and YouTube says so itself. The longer answer is where the old advice came from and the one situation where tags still earn a minute of your time.
What Tags Were Originally Designed For
Tags are a hidden metadata field in YouTube Studio. The commonly told story is that in the early years of YouTube they were the main way you told search what a video was about, which is why a whole genre of tag generators and “copy your competitor's tags” tutorials grew up. That history is mostly retold by SEO vendors rather than documented by YouTube, so I'd treat it as folklore with a grain of truth.
What Tags Are Useful For Today
YouTube Help answers the question “How important are tags?” with “Not important,” and says they are mainly used to correct for common spelling mistakes, using the example of YouTube versus U Tube versus You-tube.
That's a long way from the advice still found on some tag-generator pages, which tell you to fill the full 500-character field with 20 to 30 tags and put the best ones first. That advice conflicts with what YouTube publishes, and I couldn't find a primary source supporting it. An older Ahrefs study from 2019 found that only about a third of top-ranking videos used the target keyword in their tags, which points the same way: tags were never the lever people thought they were.
So if you're asking “do YouTube tags still matter” for ranking, the evidence says they matter very little.
When Tags Still Help
There are two cases worth a minute.
- Commonly misspelled terms. If your channel name, a product name or a technical term is often typed wrong, add the common misspellings as tags. This is the use YouTube itself describes.
- Ambiguous terms. If a word means different things (a game with the same name as a film, say), vendors suggest tags can help disambiguate. YouTube doesn't confirm this, so treat it as a low-cost hedge, not a tactic.
A few things to leave out: unrelated popular tags, a competitor's entire tag list, and hashtags dressed up as tags. Hashtags are a separate feature. They're visible and clickable, while tags sit in a separate field that viewers never see. Spend two minutes on tags and move on to the title.
What Makes a YouTube Title Clickable in 2026?
The best-known recent dataset on this comes from AIR Media-Tech, published in June 2026. They analyzed 18,080 English-language channels across 11 niches and four size tiers (by monthly views), after removing channels where 80 percent or more of uploads were Shorts. Their main metric was median engagement rate per video (likes plus comments divided by views), with 75th-percentile views as a second measure.
Two caveats matter before using any of it. First, this is correlational, and AIR says so: they couldn't separate the title from the thumbnail, and they can't prove titles caused the results. Second, engagement rate isn't click-through rate. The study tells you which title patterns show up on videos whose viewers engage more, not which titles get clicked most. I'll use “clickable” in this section as a loose proxy, and I'll bring in two within-channel studies that found much smaller effects, because they change how you should read the headline numbers.
Title Length
AIR's headline finding is that 30 to 50 characters performed best across most niches and size tiers. Under 30 characters did better in gaming, entertainment and food. Titles of 90 characters or more tended to get fewer views, though the people who did click on them engaged more, which AIR reads as a search-driven audience that found an exact match.
Now the counterweight. A separate study from YTGrowth looked at 28,947 videos across 707 channels, normalized for channel size and video age, and found a correlation between title length and performance of 0.0046, which is essentially zero. It also lists four other published studies that each recommend a different length, from roughly 20 to 40 characters up to 65 to 70, with one unverified claim that 90 to 100 characters is best. Another analysis from OverseerOS compared 277 breakout videos against 2,345 normal uploads from 59 channels and found the same median title length for both: 58 characters. In that sample, about three quarters of breakouts used a title within 25 percent of their channel's usual length.
My reading of the gap: AIR compares titles across different channels, so length may be partly standing in for niche and audience. The other two studies compare videos against the same channel's own baseline, where the length effect mostly disappears. Both can be true at once.
So how long should a YouTube title be? Long enough to be unambiguous, short enough that the important part survives truncation (AIR puts mobile truncation at roughly 60 to 70 characters), and consistent with what already works on your channel. Treat length as a readability check, not a growth lever.
Search Intent vs Curiosity
A title can do one of two jobs, and mixing them up is the commonest mistake.
Confirm what the viewer already wants. AIR's strongest example is food. Short recipe-name titles like “Perfect Pasta” or “Crispy Roast Chicken” had the highest engagement rate in the whole dataset at the smaller size tier, about 10 percent. The viewer searched for a crispy roast chicken, and the title says they found it. Anything extra just gets in the way. AIR's rule of thumb is a good one: if the viewer knows what they want, name it. If they need convincing, pitch it.
Create a reason to click. In gaming and entertainment, AIR found that very short, withholding titles can work because the viewer can't resolve them without watching. They also found that curiosity language, the “what nobody tells you” and “the real reason” family, was the only positive signal that carried across every niche and size tier.
Here's how that plays out for a YouTube SEO video, using titles I wrote as illustrations (not measured ones):
- “YouTube Keyword Research for Small Channels (2026)” names the topic, the audience and the year. It's built for someone typing a query.
- “Why Your Best Video Isn't Showing in YouTube Search” names a problem the viewer already has and withholds the cause. It's built for someone browsing.
Neither is better in the abstract. The first is the safer choice when your traffic is mostly search. The second needs the video to deliver a real answer, or you've written a curiosity gap you can't close.
Specificity
A specific title gives the viewer something they can judge. AIR's own illustrations make the point well:
- “Make money online” versus “How I make $8,000/month from one YouTube channel.” The vague one could be about anything. The specific one makes a claim the viewer can weigh and the creator has to back up.
- “How I spent 7 days in Vietnam” versus “Vietnam in 7 days.” In their travel data, leading with the destination did better, because the destination is the thing the viewer is searching or browsing for.
- “30-day squat workout plan” versus “30-day squat challenge results.” AIR's fitness data favored the second, because “results” tells you the video resolves a question.
The cost of specificity is accountability. A specific promise that the video doesn't deliver is worse than a vague one, because it creates the disappointed viewer YouTube's systems are designed to notice.
Numbers and List Titles
AIR found numbers helped in tech, business, fitness and food, and did little elsewhere. The dividing line is whether the number carries information.
Numbers that work are constraints and proof: “3-ingredient pasta,” “$50/day in Tokyo,” “5-minute breakfast.” Each one tells you how hard it will be or what it will cost. Numbers that don't work are decoration: AIR's example is a title like “7 tips for your first BG3 run,” where the seven doesn't change what the viewer expects.
A quick test: take the number out. If the title means the same thing without it, it's decoration.
Question Titles
AIR found question marks positive in education, business and science and tech, and weak in gaming, food and kids. Their explanation is that a question works when the viewer already has it in their head.
OverseerOS tested this from another angle. Among 7,935 million-view titles, only 6.8 percent contained a question mark. When they matched 891 pairs of question and statement titles from the same channels published close together, the question version won 49.9 percent of the time and the statement version 50.1 percent, with a median view ratio of 1.000. A question mark, on its own, gives no advantage.
Put those together and the practical rule is simple. “Do YouTube tags still matter?” works because it's the exact question people type. “Is this the worst video on YouTube?” works only if the video answers it. Use a question when it's the viewer's question.
Why Clickbait Can Backfire
YouTube Help tells creators to use titles and thumbnails that accurately represent the video, and lists what to avoid: misleading or sensational packaging, shocking or disgusting imagery, and what it calls “loud” titles that lean on ALL CAPS or strings of exclamation marks.
That sits awkwardly next to AIR's finding that ALL CAPS correlated with better results in gaming and entertainment, and worse in education. My reading: capitals can read as enthusiasm to an audience that expects it, but YouTube's published guidance doesn't endorse them, so I'd test it carefully and not build a strategy on it.
The cleanest example in the AIR report is a faceless narration channel that toned down horror-style thumbnails and buzzword titles. They report CTR rising from 5.6 to 6.4 percent, and engaged views going from roughly 357,000 to 983,000. Two warnings: it's a single case from a company that sells channel audits, and their explanation (that quality classifiers suppressed the old packaging) is their theory, not something YouTube has published.
The part you can defend from YouTube's own description is the mechanism. The system learns from what viewers don't watch, from dislikes and from “not interested” clicks. A title that promises more than the video delivers generates exactly those. YouTube also says changing a title or thumbnail can help, but only because viewers react differently to the new version, and it adds that you shouldn't change what's already working.
YouTube Keyword Research in 2026
Keyword research on YouTube is less about finding a magic phrase and more about learning the words your audience uses and checking whether you can realistically compete for them. None of the tools below see YouTube's real search data, so treat their volume numbers as estimates.
Start With the Topic, Not the Keyword
Begin with a viewer problem you can solve better than what already exists, and let the phrasing come second. YouTube's own advice on performance is to focus on knowing your audience, and keywords are simply the language that audience uses.
Good sources for a first list: questions in your own comments, questions in the comments of the biggest videos in your niche, and the sections of competitor videos that people ask follow-up questions about.
YouTube Autocomplete
The search bar is the simplest tool for how to find keywords for YouTube. Type your topic and read the suggestions. Then add a letter, or words like “for beginners,” “vs,” “without” and “2026,” and read again. The suggestions reflect what people are typing on YouTube itself, although YouTube doesn't explain how they are chosen. It's free, and it gets you the real phrasing of long-tail queries that a tool might miss.
Competitor Analysis
Search your candidate phrase and open the top results. Don't just count them. Look at:
- Channel size, and whether small channels appear on page one.
- Views relative to subscriber count, which shows whether the video is pulling its weight.
- How long ago it was published, since old results can be easier to beat.
- Format: tutorial, list, review, essay, how long.
- What the comments say is missing.
One caution: your results are personalized by your own history, so what you see isn't exactly what a new viewer sees. Check from a clean browser session when it matters.
Search Intent
Before you commit to a phrase, work out what the person typing it wants. A viewer searching “YouTube SEO” may want a definition, a checklist or a full course. The top results show you what YouTube has learned satisfies people for that query. If every top result is a ten-minute tutorial, a 45-minute documentary is making a different promise from the one the search implies. YouTube says search rewards videos that drive engagement for the query, so matching the expected format matters as much as matching the words.
Long-Tail Queries
Long-tail queries are longer, more specific phrases, such as “YouTube SEO for small channels” instead of “YouTube SEO.” Vendors argue they have less competition and clearer intent, which is plausible but I couldn't find independent data proving it for YouTube. The trade-off is real either way: lower volume, a more precise audience, and a better chance of the video matching what the viewer wanted.
For a sense of where behavior is heading more broadly, Google reports that the average AI Mode search is about three times as long as a traditional Google query. That's about Google, not YouTube, but it points the same way: people are phrasing things as full questions.
Matching the Keyword to Your Channel Size
Here's a free heuristic from vendor guides, which I'd treat as a rule of thumb: search the phrase and see who ranks. If the first page includes channels under about 10,000 subscribers with decent view counts, the term is probably reachable for a small channel. If it's only large channels with big view counts, pick a narrower version.
YouTube doesn't say channel size is a direct ranking factor. But larger channels start with more viewers and more history, so they usually collect early engagement faster. That's my inference, and it's the reason to chase YouTube SEO for small channels through narrower topics, not the head terms.
How Much Does the Description Actually Matter?
A moderate amount. Officially, the description is part of how YouTube matches a video to a search. Beyond that it's a place to explain the video to humans, add chapters and link to related material. It's rarely what makes or breaks a video, but it's cheap to do well.
The First Two Lines
Only the opening of the description is visible in most places before someone taps to expand. How much appears varies by device and placement, and the figures vendors quote disagree: vidIQ reports roughly 90 characters on a search results page and about 160 below the video, while other guides quote 100, 125 or 150. Since nobody agrees, treat the first 100 characters as your working window and check how yours looks on your phone.
That window should say what the video is and who it's for, in a normal sentence, with the main phrase in it. For example: “YouTube SEO for small channels: how to pick keywords, write titles and test changes in YouTube Studio, with real examples.”
That's how to optimize a YouTube description for the part people actually read.
Supporting Context
After the opening, add two to four sentences covering what the video walks through, who it suits and what it assumes. Use the same terms the video itself uses, so the description and the spoken content describe the same thing. Mention related videos or playlists a viewer would want next.
Keywords Without Keyword Stuffing
Use the main phrase once or twice in natural sentences, add the variations a person would actually say, and stop. If you read it aloud and it sounds like a list of search terms, cut it. YouTube's spam and deceptive-practices policies cover misleading metadata, so stuffing is a risk without being a benefit. Check the current policy page for the exact wording.
Links, Chapters and Resources
Chapters are the most useful addition. Per the vendor guides I checked, a video needs at least three timestamps to get chapters, and they're less useful on videos under about five minutes. Write chapter titles in plain words someone might search (“metadata and SEO audit,” not “the part about tags”). Google has described how timestamps in a YouTube description can make a video eligible for key moments in Google Search, which gives each section a chance to appear as its own result. YouTube Studio also has an option to allow automatic chapters and key moments.
Put your most important link, usually the next video to watch, high up, and keep promotional links below the useful content.
YouTube Search vs Google Search
This is the interesting part, because the two search engines are more connected than most creators assume, and less connected than some vendors claim.
Why One Video Can Rank in Both
YouTube Help itself compares its search to Google's: both try to surface the most relevant results for a query. A video that answers a specific question clearly, with a title that names the question, a description that explains it, and chapters that label the parts, gives both systems the same kind of clean signal. That's an inference, but it's a reasonable one.
The difference is what happens afterward. YouTube measures how viewers behave on YouTube: clicks, watch time, satisfaction. Google measures its own results page and the pages behind it. A video can do well in one and poorly in the other.
How Google Uses YouTube Content
The documented part is key moments. Google's Search Central guidance explains that key moments show timestamped segments of a video directly in search results, and a Google product manager has described how, for YouTube videos, timestamps in the description make a video eligible for them. Anything beyond that, such as how much Google uses YouTube transcripts or engagement to rank a video in web results, isn't documented that I could find, so I won't pretend otherwise.
You can see how much Google traffic you get yourself: in YouTube Analytics, external traffic sources show whether Google Search is sending people.
What Changes When Search Becomes AI-Assisted
Google says AI Mode passed one billion monthly users about a year after launch, and that AI Mode queries have more than doubled every quarter since launch. It also reports that the average AI Mode search is roughly three times the length of a traditional search, that planning-related queries have grown 80 percent faster than AI Mode queries overall over the past six months, and that searches starting with “which” have grown 40 percent faster than the overall rate. Google has also said overall search queries are at an all-time high and that these AI features are adding searches, not replacing them.
What that does and doesn't tell you:
- It doesn't tell you AI search replaces YouTube. Nothing in Google's numbers says that, and I'd be wary of any article that claims it. I also found no published data on how often YouTube videos are used or cited in AI Mode answers, so that's currently unknown.
- It does tell you queries are getting longer and more specific. People ask comparison, planning and “which one should I choose” questions in full sentences.
- The sensible response is the one you'd already make. A video that answers one specific question, says so in its title and first lines, and labels its sections with chapters is easier to match, in YouTube search, in Google results and in whatever answer layer sits on top.
A Better YouTube SEO Workflow for 2026
The order matters here. Most creators do metadata last and then try to fix a weak idea with it. This sequence puts the decisions that matter first.
- Topic. Choose a problem your channel is credible on. Skip topics where you have nothing to add to what already ranks.
- Audience. Decide who it's for in one sentence: beginners, a small channel, a specific role. This makes every later choice easier.
- Search intent. Search your idea and look at what's on page one. Note the format, length and promise of the top results.
- Keyword validation. Use autocomplete and the competitor checks above. Check that small channels can appear for the phrase. Pick one primary phrase and one or two variations.
- Title concept. Write three titles: one that names the topic plainly, one that makes a specific promise, and one that raises a question the viewer already has. Choose the one that matches how most of your traffic arrives.
- Video structure. Plan the video so the first 30 seconds confirm the title's promise, and so each section can become a chapter with a clear name.
- Metadata. Write the title, the first line of the description, the supporting paragraph and the chapters. Add tags only for misspellings. Make sure the thumbnail and title tell the same story.
- Publish. YouTube says publish time isn't known to affect long-term performance, so don't spend effort on finding the perfect hour.
- Analyze. After a few weeks, read the search terms report, impressions click-through rate, average view duration and the traffic source split. Look for queries you didn't plan for and for a gap between CTR and retention.
- Revise. If CTR is weak and retention is fine, the packaging is the problem. If CTR is fine and retention is weak, the video isn't delivering on its title. Change one thing at a time and leave it long enough to see the effect. YouTube cautions against changing what's working, and says a change helps only to the extent that viewers react to it.
What We Would Stop Doing in 2026
This is the short list of habits that cost time and have little or no support in YouTube's own guidance.
Keyword Stuffing
Repeating the same phrase in the title, description and tags doesn't add information. YouTube's systems are built to read the video's content and viewer behavior, and stuffing can cross into misleading metadata. State the topic clearly once or twice and move on.
Writing Descriptions for Robots
A description written as a block of keywords helps nobody. Viewers see the first lines and decide. Write for the person, and let the search terms appear naturally because you're describing the same video.
Adding Random Tags
YouTube says tags are not important and mainly cover spelling mistakes. Twenty-five loosely related tags is a ritual, not a strategy. If a tag isn't a misspelling or a true disambiguation, leave it out.
Copying Viral Titles
A title that worked for a channel with a huge existing audience often relies on that audience. The within-channel studies above suggest that what separates a breakout from a normal upload is mostly not a copyable title pattern. Borrow the structure if it fits your topic, but write the promise your video can keep.
Chasing Search Volume Without Considering Competition
A phrase with high estimated volume and ten large channels already serving it is a worse target than a narrower phrase where a small channel is already getting views. Check who ranks before you commit, and remember that volume numbers from tools are estimates.
The 2026 YouTube SEO Checklist
Use this before you publish, and again when you review a video a few weeks later.
Before you film
- I can state who the video is for in one sentence.
- I searched the topic and looked at the format and length of the top results.
- I picked one primary phrase and checked that small channels can appear for it.
- The video's promise is something I can actually deliver in the first 30 seconds.
Title and thumbnail
- The title names the topic or the problem in plain words, with the key phrase near the front when search is my main traffic.
- Any number in the title carries information (a constraint, a cost, a time frame).
- A question title asks something my viewer is already asking.
- The title and thumbnail make the same promise and neither overstates the video.
- No ALL CAPS or strings of exclamation marks.
Description
- The first line says what the video is and who it's for, in a normal sentence.
- A short paragraph explains what's covered, using the same words as the video.
- Chapters have plain, searchable names (at least three).
- Links to the next video or playlist are near the top.
Spoken content and captions
- I said the topic clearly near the start.
- I checked the captions for names, product terms and jargon.
Tags
- I added tags only for common misspellings or ambiguous terms.
After publishing
- I checked the search terms report for queries I didn't plan for.
- I compared CTR with average view duration to see whether the problem is packaging or content.
- If I changed the title or thumbnail, I changed one thing and waited before judging.
Conclusion
YouTube SEO in 2026 isn't a metadata game. Keywords still help YouTube understand the subject, but titles, descriptions and tags operate inside a larger system that includes viewer intent, content relevance, satisfaction and performance. The most useful SEO strategy is therefore not to optimize every field separately, but to make the entire video package answer the same viewer need.
If a viewer searches a question, sees your title, reads the first line of your description, and gets the answer in the first minute of the video, every signal is pointing the same way. That's about as close to a reliable YouTube SEO strategy as the evidence supports. Everything else is a hypothesis, and the good news is that your own analytics can test most of them.
Sources
- YouTube performance FAQ and troubleshooting (YouTube Help)
- How to write a YouTube title that gets clicked: research across 11 niches (AIR Media-Tech, June 2026)
- YouTube title length doesn't predict performance: a 28,947-video study (YTGrowth)
- We compared 277 breakout YouTube titles with 2,345 normal titles (OverseerOS)
- Should YouTube titles be questions? 7,935 million-view videos (OverseerOS)
- How AI Mode is changing the way people search in the U.S. (Google)
- Google Q3 report: AI Mode and AI Overviews lift total search usage (Search Engine Journal)
- A new way to enable video key moments in Search (Google Search Central)
- Google's instruction to activate key moments of videos in Search (SEOZoom)
- YouTube video descriptions: how to write them better in 2026 (vidIQ)
- YouTube keyword research: free tools and methods that work (TubeMilestone)
- YouTube chapters (Humble and Brag)
- Ahrefs on YouTube tags