How to Do Keyword Research for YouTube, Step by Step

TLDR

Six steps in a deliberate order. Demand from autocomplete, volume to rank, competition to filter, outliers to prove the angle, packaging lessons from the winners, then a one-table plan. Order beats tooling.

  • Start in YouTube's search bar, not in a paid tool
  • 95% of videos get under 1,000 views, outlier data is how you join the other 5%
  • A cheap stack run in the right order beats an expensive one used backwards

Disclosure. We're affiliates for OutlierKit. We pay for the tool ourselves and use it on the agency side. If you buy through our links we earn a recurring commission at no extra cost to you. Affiliate disclosure.

Most YouTube keyword research starts in the wrong app. Someone opens a paid tool, types in a topic, exports two hundred keywords sorted by volume, and calls it a content plan. Then the videos ship, the views don't come, and the tool gets blamed for a problem that wasn't the tool.

We run the research side of client channels at Market Correct, and after 12+ years across 400+ clients of picking what earns budget, the honest finding is that the order of operations matters more than the tool. The workflow below is six steps in a deliberate sequence. It isn't complicated, it's just strict about order. Demand first, volume second, competition third, outliers fourth, framing fifth, and only then a plan. Run it in this order and a cheap stack beats an expensive one used backwards.

The six-step workflow at a glance

  1. Pull real queries from YouTube autocomplete
  2. Put volume numbers on the shortlist
  3. Read the competition before you commit
  4. Find the outlier videos in the niche
  5. Steal the framing, not the topic
  6. Turn it into a plan a team can run

Each step feeds the next one. Don't skip ahead. That's how teams end up filming high-volume topics that ten massive channels already own.

Step 1. Pull real queries from YouTube autocomplete

Don't start in a paid tool. Start in YouTube's own search bar. Type your topic and write down every completion autocomplete offers, then add a letter after the topic and harvest the next set. These are real phrases people are typing right now, straight from the source, free.

The point of this step is intent, and it's the one teams rush past. Autocomplete phrases are closer to YouTube intent than anything a web tool will seed you with, because they came from YouTube. Twenty minutes here usually produces 30 to 50 candidate queries, which is plenty of raw material for the rest of the workflow.

Step 2. Put volume numbers on the shortlist

Now rank the candidates by demand. Run the shortlist through a tool that reports YouTube search volume so you're ordering by interest instead of by gut. VidIQ's native score, OutlierKit's keyword data, or Keywords Everywhere inline all work. If you already pay for Ahrefs, its free YouTube keyword tool and Keywords Explorer YouTube mode cover this step without a new subscription.

One warning before you trust any of those numbers. YouTube doesn't publish search volume, so every figure is a third-party estimate, and native estimates deserve more trust than web-derived proxies. We wrote a full explainer on where YouTube search volume numbers come from and how much weight they can carry. The one-line version is that volume ranks ideas against each other. It doesn't forecast views.

Step 3. Read the competition before you commit

A high-volume keyword owned by ten massive channels is a worse bet than a mid-volume keyword nobody has served well. That's the single most common mistake we see in creator research, because volume is the number every tool puts in the biggest font.

For each surviving query, look at who currently ranks. Channel size, video age, and production quality tell you whether there's a gap. A results page full of two-year-old videos from mid-sized channels is an invitation. A page of fresh uploads from channels with millions of subscribers is a warning. It doesn't mean never, it means not first. OutlierKit's low-competition keyword finder and niche finder are built for exactly this read, and the niche finder pairs competition level with RPM ranges so you can also see whether the lane pays.

Step 4. Find the outlier videos in the niche

This is the step most keyword research skips entirely, and it's the one that decides what gets filmed on our client work.

95% of YouTube videos get fewer than 1,000 views. The outliers, videos doing 3x to 10x their channel's average, are the closest thing YouTube has to proof that a specific angle works.

An outlier is a video massively outperforming its own channel's baseline. That baseline correction is what makes the signal honest. A big channel getting big views proves the channel is big. A mid-sized channel getting 8x its normal views proves the video itself did something right. Search volume tells you a topic has pull. The outlier tells you which treatment of that topic converts to actual watching.

This is the job we use OutlierKit for, since outlier detection is its core feature rather than a side module. 1of10 is the closest single-purpose alternative if you want a second option.

Run the outlier step on your own niche before your next shoot. The trial is 10 credits, no card. Try OutlierKit free

Step 5. Steal the framing, not the topic

Pull up the outliers you found and study the packaging before you script anything. The hook in the first 15 seconds, the title construction, the thumbnail concept. You're not copying the video. You're learning why that angle beat the channel's baseline, then applying the lesson to your own take.

Web SEO habits don't transfer cleanly here either. A title tag and a meta description are set-and-forget. YouTube packaging is the product. If you come from search, our post on using Ahrefs for YouTube covers exactly where the web toolkit stops and the video-native work begins.

Step 6. Turn it into a plan a team can run

Write the plan down in one table. Each row is a video. The columns are the target query, the volume estimate, the competition read, the outlier reference, and the planned angle. That's a content plan a team can execute without you in the room, instead of a list of vibes.

The table also makes the plan auditable after the fact. When a video overperforms, you can see which research call was right. When one flops, you can see which column lied to you. Over a quarter, that feedback loop sharpens the research more than any tool upgrade. You can't buy that.

Which tools to use for each step

Step Free option Paid option
1. Seed queries YouTube autocomplete Not needed
2. Volume Ahrefs free tool, Google Trends VidIQ, OutlierKit, Keywords Everywhere
3. Competition Manual results-page read OutlierKit niche and keyword finder
4. Outliers None worth using OutlierKit, 1of10
5. Framing Watch the outliers yourself OutlierKit hook analysis
6. The plan A spreadsheet Still a spreadsheet

If you're choosing between the mainstream suites for the workflow jobs, we compared them head to head in VidIQ vs TubeBuddy, and the full field is ranked by job in our best YouTube keyword research tools roundup. Google Trends with the YouTube Search filter lives at trends.google.com, and it doesn't rotate out of the stack as budgets grow.

The bottom line

Verdict

YouTube keyword research is a sequence, not a tool. Seeds from autocomplete, volume to rank them, competition to filter them, outliers to prove the angle, packaging lessons from the winners, and a one-table plan at the end. Most teams do step 2 with an expensive tool and skip steps 3 through 5 entirely, which is why most videos land in the 95% that go nowhere.

You don't need the expensive tool to start. Run the sequence honestly on the free stack, and add paid tools only where a step is slowing you down. That's how we'd spend a client's budget, so it's how we'd spend yours.

About Market Correct. We're a performance marketing agency with 400+ clients managed over 12+ years. We run Google Ads, paid social, and programmatic for B2B and DTC brands, and this workflow is the one we run before any client channel gets a content plan. Our guides are meant to help you decide, not to land a commission.

Agency lens

Want this workflow run on your channel instead of by you?

We run this exact sequence before any client channel gets a content plan. If you'd rather review the plan than build it, talk to us.

Talk to us about your campaigns
FAQ

The YouTube keyword research workflow, common questions

Six steps, in order. Pull real queries from YouTube autocomplete, put volume estimates on the shortlist, read the competition for each survivor, find the outlier videos in the niche, study the winners' hooks and thumbnails, then write it all into a one-table plan. The order matters more than the tool.

Yes, in two big ways. YouTube discovery runs heavily on suggested videos and the home feed, so search demand is only one path to views. And YouTube doesn't publish search volume, so every number is a third-party estimate. Web SEO habits transfer partially, but packaging and outlier data have no Google equivalent.

YouTube autocomplete for real queries, Google Trends with the YouTube Search filter for direction and seasonality, and Ahrefs's free YouTube keyword tool for rough volume. That free stack covers demand, momentum, and magnitude, and it's enough to start filming against.

Look for mid-volume queries where the current results page is weak, old videos, mid-sized channels, or thin treatments of the topic. A results page full of fresh uploads from huge channels is a warning. OutlierKit's low-competition keyword finder and niche finder automate this read and pair it with RPM ranges.

A video massively outperforming its own channel's average, typically 3x to 10x the channel baseline. The baseline correction is what makes it an honest signal. It proves a specific angle worked, rather than proving the channel was already big. It's the strongest single input for deciding what to film.

Barely. Tags are a minor signal next to titles, thumbnails, and hooks, which is why tag-first tools are optimization utilities rather than research tools. Spend research time on the angle and the packaging, and treat tags as five minutes of cleanup at upload.

One primary query, plus its close variants folded into the title, description, and spoken content. A dozen keyword variations usually map to one video, not twelve. Grouping variants under a single video is what turns a keyword dump into a content plan.

No. Keyword Planner measures Google search demand, not YouTube search demand, and the two don't phrase or behave the same way. Use YouTube-native sources for video decisions and keep Keyword Planner for Google Ads work.

Run the full six-step workflow once per planning cycle, quarterly for most channels, then do a quick outlier check before each shoot. Demand shifts slowly, but outliers surface fast, and a pre-shoot check is cheap insurance against filming an angle that's already gone stale.

Start with the free stack, autocomplete plus Google Trends plus Ahrefs's free tool. Add one paid tool when a step starts costing you shoots. For volume and daily triage that's usually VidIQ, and for the outlier and competition steps it's OutlierKit, which has a 10-credit trial with no card.