Research Guide
YouTube Search Demand: Validate a Topic Before You Film
Volume estimates are modelled guesses. Demand you can verify comes from what people ask, what already exists, and what your own audience did.
Validate a topic by gathering evidence you can check: the questions people actually type, whether useful answers already exist, how comparable channels performed with it, and what your own audience did the last time you covered something adjacent. Search-volume numbers for YouTube are vendor estimates rather than published figures, so treat them as a weak signal among several rather than the decision.
In this article
Before you begin
What you’ll take away
- YouTube does not publish search volume. Every number you see is a vendor model, so never let one carry the decision alone.
- Autocomplete shows real phrasing people use, which is often more useful than an estimated number.
- A topic with no existing videos is more often no demand than an untapped opportunity.
- Your own adjacent uploads are the strongest evidence you have, because they already tested your audience.
- Write the audience question as a sentence. If you cannot, the topic is not ready to film.
Why the volume number deserves less weight than it gets
YouTube does not publish search volume. Tools that show a monthly number for a YouTube query are modelling it, usually from a mix of Google search data, their own panel, and inference. Different tools produce different numbers for the same query, which is the clearest sign that none of them is a measurement.
That does not make them useless. It makes them one weak signal among several, useful for rough ordering rather than for deciding. The failure mode is treating a precise-looking figure as though it were a fact about the platform.
The evidence below is weaker in appearance and stronger in practice, because you can verify every piece of it yourself.
Start with the words people actually type
Type the beginning of your topic into YouTube search and read the autocomplete suggestions without clicking. Those are derived from real queries. They tell you the phrasing, the qualifiers people add, and the adjacent things they want, which is often more actionable than a volume estimate.
Pay attention to the shape of the suggestions. Questions beginning with how, why or what indicate a viewer expecting an explanation. Comparison phrasings such as "versus" or "or" indicate someone deciding between options.
Words like "not working", "error" or "problem" indicate someone already stuck, which is usually the most motivated audience there is.
Collect five to ten real phrasings before going further. You are building a picture of intent, not a keyword list.
Read the supply side honestly
Search the phrasings and look at what already exists. Three patterns matter, and only one of them is an opportunity.
If strong, recent videos already answer the question well, the demand is real but the bar is high, and you need a genuine angle rather than another entry. If the results are dated, thin, or answer a slightly different question, that is the opportunity: real demand with an unsatisfied version of it. If almost nothing exists, the honest default assumption is that there is no audience, not that you found a gap.
Absence of supply is usually evidence of absence of demand.
Note the channel sizes in the results too. If every result is from a channel far larger than yours, the topic may be real but not yet winnable for you, which is a scheduling decision rather than a rejection.
Your own uploads are the best evidence you have
Every video you have already published was a test of your specific audience. That makes your own history stronger evidence than any external estimate, because it controls for the one variable nobody else can model: who subscribes to you.
Look for the closest adjacent topic you have covered. Compare it against your own median rather than your best video. If an adjacent topic performed above your median, that is a real signal from the only audience that matters here.
If it performed below, a bigger version of the same idea is unlikely to rescue it.
This is the comparison Kolliq is built to make: recent uploads against your own baseline, with the sample size visible so a promising pattern backed by two videos is not mistaken for a proven one.
Write the audience question as one sentence
Before committing, write the question your video answers, phrased the way a viewer would ask it. Not the title, and not the topic. The question.
If the sentence comes out vague, that is the finding. "Something about analytics" is not a question anyone types. "Why did my views drop when my subscriber count went up?"
is. The second tells you what the first thirty seconds must do, and the first does not.
This step catches more bad ideas than any research tool, and it costs a minute.
Decide, publish, and let the result update the picture
Validation reduces uncertainty; it does not remove it. At some point the only remaining test is publishing, and the useful discipline is deciding in advance what the result will mean.
Record what you expected before you publish, then review against your own median at a consistent video age. A topic that beat your median once is a candidate for a second attempt, not a proven vein. Two or three consistent results make a pattern worth building on.
Over time this replaces estimated volume with something better: a record of what your particular audience actually turned up for.
Your next review checklist
- Collect five to ten real autocomplete phrasings for the topic.
- Classify the intent behind them: explanation, comparison, or stuck.
- Search those phrasings and read the supply honestly.
- Note whether results are strong, thin, absent, or only from much larger channels.
- Find your closest adjacent upload and compare it with your own median.
- Write the viewer question as one sentence in their words.
- Record the expected result before publishing.
- Review against your median at a consistent video age.
Questions creators ask
Does YouTube publish search volume?
No. YouTube does not release search volume figures. Tools that display a monthly number for a YouTube query are producing a model, typically informed by Google search data and their own sampling. Different tools disagree on the same query, which is why the number should inform rather than decide.
Is a topic with no competing videos a good opportunity?
Usually not. Absence of supply is more often evidence that nobody is searching for it than evidence of an untapped gap. Before treating it as an opportunity, find the question being asked somewhere you can verify, such as autocomplete, comments, or a community you participate in.
Can Kolliq tell me the search volume for a topic?
No, and it deliberately does not estimate one. Kolliq compares public performance signals and your own uploads against your baseline. For demand, it can show you how adjacent topics performed for your channel, which is evidence rather than a model.
How many videos does it take to know a topic works?
One result above your median is a candidate, not a conclusion. Two or three consistent results across comparable uploads make a pattern worth building a cluster around. Fewer than that and you are reading normal variation.
Should I use a keyword tool at all?
They are useful for discovering phrasings and for rough ordering between options. The mistake is treating a modelled number as a measurement, or letting it outweigh what your own audience already demonstrated.
From reading to doing
Check the topic against your own baseline
Your adjacent uploads already tested this audience. Compare them against your median before committing to the next one.
Open topic analysis