How to Do YouTube Keyword Research in 5 Steps (Free, 15 Minutes)

YouTube keyword research is less mysterious than the tools industry makes it sound. The whole thing — from blank page to a videos-to-make list — takes about 15 minutes once you've done it a few times. This post walks the 5-step workflow using free tools (our YouTube Keyword Research tool plus YouTube itself). No paid subscriptions, no Chrome extensions, no "keyword difficulty" scoring that may or may not be accurate.

Step 1: Pick 3-5 tight seed keywords

A seed is a 1-3 word phrase your channel could legitimately make 20+ videos about. Not a single video idea — a topic cluster.

Good seeds (specific, ownable):

• "home gym setup" (fitness channel with a home gym focus)
• "sourdough bread" (baking channel)
• "oil painting tutorial" (art channel)
• "budget travel italy" (travel channel with a geographic focus)

Bad seeds (too broad, too narrow, too ambitious):

• "fitness" — too broad; you can't own this.
• "how to do a 225lb deadlift with knee sleeves" — too narrow; this is a single video, not a seed.
• "mukbang" — too competitive if you're a new channel; you'll rank nowhere.

Aim for seeds where a mid-size channel (10K-500K subs) is ranking, not where the top results are channels with 10M+ subs. The competition at that tier is usually beatable with good content; the top tier is not.

Step 2: Expand each seed with autocomplete

Open the free YouTube Keyword Research tool. Enter your seed. Leave all three expansion options checked (questions, modifiers, alphabet). Click Find Keywords.

You'll get 100-300 queries grouped by type:

• Base — YouTube's top suggestions for the seed alone.
• Questions — how/what/why/when/where/who prefixed to your seed.
• Modifiers — best/free/top/vs/for/with/without etc. plus your seed.
• Alphabet — your seed followed by each letter a-z, showing what YouTube autocompletes.

This is the "what are people actually typing" layer. Every query in the list is one YouTube is suggesting to real users in real-time — not a guess, not a scrape, the same endpoint powering YouTube's search bar dropdown.

Download the CSV. Repeat for each of your 3-5 seeds. You'll have 500-1,500 queries total.

Step 3: Group by search intent

Open the combined CSV in Google Sheets. Skim through. Most queries fall into one of five intent buckets:

1. Tutorial intent — "how to X," "X tutorial," "X for beginners." Video format: step-by-step walkthroughs.
2. Comparison intent — "X vs Y," "best X 2026," "X or Y." Video format: comparison or ranking.
3. Review intent — "X review," "is X worth it," "X honest review." Video format: product/tool review.
4. Problem-solving intent — "X not working," "fix X," "why does X do Y." Video format: troubleshooting.
5. Entertainment/curiosity intent — "X explained," "X reaction," "weird X." Video format: commentary or storytelling.

Tag each promising query with one of these. Delete queries that don't match your channel (a fitness channel probably isn't making "home gym setup reaction" content). You should end with 30-80 tagged queries across your seeds.

Sell Custom Apparel — We Handle Printing & Free Shipping

Step 4: Eyeball competition on YouTube itself

Paid tools give you a "competition score" number. You can get roughly the same information by just searching the query on YouTube and looking at the top 5 results. Things to check:

• Subscriber counts of the top 5 channels. If they're all 5M+ subs, your video probably won't rank. If there are mid-size channels (50K-500K) in the top 5, the query is beatable.
• Upload dates. If the top 3 videos are from 2022 with 500K+ views each, the query is evergreen and ripe for a 2026 update (which Google prefers for freshness).
• Thumbnails. If every thumbnail is low-effort (plain text, generic stock photo), your better thumbnail can out-click them. If every thumbnail is polished MrBeast-style, you're competing with production budgets.
• Video length. If top results are 3-minute shorts-ish videos and you want to make 20-minute deep dives, there might be a gap in the market.

This takes about 60 seconds per query. For 30 queries, it's a 30-minute job. Out the other end you'll have 10-15 queries where you can realistically rank.

Step 5: Craft the title around the query

The title should contain the exact query (or something very close) — this matters more for YouTube than for Google. YouTube's algorithm heavily weights title-to-query match when deciding to surface your video to searchers.

Good titles follow a pattern: the query + a hook.

• Query: "how to set up a home gym in a small apartment." Title: "How to Set Up a Home Gym in a Small Apartment (500 Sq Ft Tested)."
• Query: "sourdough bread recipe for beginners." Title: "The Sourdough Bread Recipe I Wish I'd Had as a Beginner."
• Query: "best budget travel italy." Title: "Budget Travel in Italy: Under $50/Day (2026, After-Inflation Prices)."

The hook differentiates from competitors at the same query. "500 Sq Ft Tested" tells viewers your video is concrete, not theoretical. "After-Inflation Prices" addresses the fact that older videos have stale prices. Your hook is what gets the click when the search result shows your video next to five others with the same query in their titles.

Avoid cramming multiple queries into one title. One query = one video. YouTube's algorithm doesn't reward keyword-stuffed titles — it rewards titles that match a single searcher's intent cleanly.

Pitfalls and common mistakes

Mistake: chasing high-volume queries. "How to lose weight" has massive volume and zero chance of ranking against Athlean-X or Jeff Nippard. "How to lose weight without gym" is more specific, lower volume, and beatable.

Mistake: ignoring low-volume queries because they have low volume. Long-tail queries (5-8 words) each get maybe 100 searches/month but convert at 10x the rate of head terms. Five long-tails at 100 views each beats one head term at 500 views every time — and the long-tail videos compound over years.

Mistake: optimizing for a query your content doesn't match. You rank for "easy sourdough bread recipe" with a 40-minute video on fermentation science. Viewers bounce in 20 seconds. YouTube downranks you. You're back to nowhere. Match content to intent.

Mistake: over-researching, under-shipping. 15 minutes of research per video is the right budget. Two hours of research per video is procrastination. The best title in the world doesn't beat a published video with a decent title.

Mistake: never revisiting. Your keyword list from January is stale by April. Rerun research quarterly. Viewer search terms drift.

Frequently Asked Questions

Do I need a paid tool for this?

No. Free tools hit the same YouTube autocomplete data. Paid tools add a scoring layer which is useful but not essential for most creators.

How many videos per month can I research this way?

One 30-minute session yields 10-15 researched queries — roughly a month of weekly uploads. Do it monthly, not per-video.

What about tags and descriptions?

Title match is the biggest lever. Tags matter slightly; descriptions matter less than people think. Focus on title-query match first.

Should I use trending queries?

Trending queries have big upside and big downside (short tail, high competition, fleeting relevance). For a new channel, evergreen queries compound more reliably than trending ones.

How do I know if a query is evergreen vs seasonal?

Check YouTube Trends or Google Trends for the query — if search volume is stable across 12 months, it's evergreen. If it spikes predictably ("Halloween costume," "tax filing 2026"), it's seasonal.