Where demand actually lives (and why your keyword tool cannot see it)

The gap between a query and a question

A keyword tool can tell you that 4,400 people a month search for design system adoption. It cannot tell you that most of them are three weeks into an internal argument about whether their component library is worth maintaining, that the argument always splits along the same two lines, or that the phrase they use with each other is not "adoption" at all. It is "nobody uses it".

That gap matters because the query is the compressed, cleaned-up version of the question. By the time someone types a search term they have already decided what kind of answer they want. The confusion that led them there, the part where a piece of writing could actually help, is gone.

Threads keep the confusion. That is their whole value as a source.

What a thread knows that a query does not

Read fifty threads on a topic and four things fall out that no volume estimate contains.

The real vocabulary. People describe problems using words that never appear in marketing copy. Those words are what they will recognise in a headline.

The failed answers. Threads are full of advice that got tried and did not work, usually with someone explaining why. That is the section your piece needs and your competitors skipped.

The disagreement. Where practitioners split, there is an article. Where everyone agrees, there is a one-line answer and nothing more to say.

The recency. A question asked constantly in 2022 and never since has been answered by something: a release, a default, a tool. Volume tools smear that over a rolling year and hide it.

How to read demand without drowning

The mechanical part is unglamorous: gather the discussion on the topic, strip the noise, and read what is left. A live argument beats a settled one, and a question nobody answered properly beats a question already answered forty times, because the gap is where the useful material is.

Then cluster. Individual threads are anecdotes; clusters are demand. When forty threads collapse into six recurring questions, those six are what the audience is actually asking, and the largest of them is usually not the topic you started with. It is sharper, and it is the one worth writing.

The last step is the one most workflows skip: decide whether the cluster is strong enough to write about at all, and write the number down. If the demand does not converge, the honest output is not an article. It is "nobody is asking this yet".

What this changes about publishing

Once you work from demand rather than from a calendar, three habits change on their own. You stop writing the topic you assumed and start writing the one underneath it. You stop padding, because every section can be traced to something someone asked. And you start being able to answer the question a stakeholder always asks, why this post?, with a figure instead of a feeling.

None of this requires software. You can do all of it by hand with a spreadsheet, a weekend and a high tolerance for reading. What software buys you is the willingness to do it every time, including for the posts that turn out not to be worth writing.