YouTube keyword research: choosing a query for your video

1. Collect real search phrases

Start with the viewer’s question, product or desired outcome. Use YouTube search suggestions, audience comments, competitor topics and customer language. Record specific long-tail queries because they often reveal clearer intent and a more focused result set.

2. Validate relevance and competition

Review each candidate’s results and check whether the leading videos match your format. Consider titles, publication dates, views and channel strength. If the result page answers a different need, the keyword is a poor fit even when its volume looks attractive.

• relevance to the video;
• clear user intent;
• an opportunity to deliver a better, more precise answer.

3. Record a baseline ranking

Before changing a title or description, record the current ranking for the query and country. Recheck at sensible intervals after optimization. This reveals movement without mixing different regions or personalized results.

4. Build a query map

Group phrases by shared intent, not merely by repeated words. For example, “how to clean a coffee machine” and “what to use to clean a coffee machine” may fit one tutorial, while “buy coffee machine cleaner” implies a commercial choice. Assign one video to each cluster when it can provide the most complete answer.

Avoid making two similar videos compete for the same primary query without a reason. When topics overlap, distinguish the angle: one video can be a beginner tutorial, another can diagnose common mistakes and a third can compare products. Different promises create distinct relevant queries.

5. Prioritise candidates

Exact YouTube search volume is often unavailable, so do not present third-party estimates as fact. For an initial decision, score every candidate consistently across four criteria and record why it was selected.

• fit with the video’s actual content;
• clarity of intent and specificity;
• strength and format of the current competing results;
• value of the target viewer to the channel or business goal.

6. Test a hypothesis instead of guessing

Write the hypothesis before the edit, such as “a more specific title will improve relevance for query X.” Record the baseline rank, date, title and key YouTube Studio metrics. Do not combine several unrelated experiments at once or you will not know what affected the outcome.

A flat ranking does not always mean the keyword is wrong. Check whether the video earns impressions, clicks and adequate retention; compare its answer with the leaders; and verify that the query’s intent has not shifted. Then decide whether to improve the video, change the target phrase or create a separate video.