Your inbox is the content calendar

Most content calendars get built back to front. Someone opens a grid, blocks out four weeks, and then goes looking for something to put in the boxes. The boxes came first and the ideas came second, which means the ideas have to be invented on demand — in one sitting, by one person, usually late on a Friday. That is how you end up with a month of topics nobody asked for, written in a voice nobody uses, published because a square was empty.

There is a list you already have that solves this, and it is not a list of topics. It is the set of questions you keep answering in writing. Support replies. The same paragraph you retype in a sales thread. The thing three different people asked in your community last month. The bit of every onboarding call where you say "so the way this actually works is". Each of those is a subject with a demand of at least one confirmed human, phrased in the words that human chose.

The phrasing is the part people throw away, and it is worth more than the topic. When a customer writes "I don't know if I should pause it or just leave it another week", they have handed you a title, an opening line, and the exact shape of the confusion. Rewrite that into "campaign optimization best practices" and you have kept the subject and lost everything useful about it. Capture questions verbatim, with the date, in one plain file. No tooling, no taxonomy, no folder structure you will abandon in March — one file, paste, done. The same instinct is behind [asking people how they found you](/blog/ask-them-how-they-found-you): the useful information was always in what someone said unprompted, and the only real work is having somewhere to put it.

Not every question deserves a post, and the filter that matters is when it gets asked. Questions from people who have already bought are documentation — write them up, put them in your help pages, keep them out of the blog. Questions from people who have not bought yet are marketing, because someone weighing the same decision is out there typing something close to that sentence into a search box tonight. "How do I export this?" is a docs page. "How do I know if this is worth it for a team of two?" is a post, and it is a post that only ever gets read by people who might buy.

Let us do the arithmetic with invented but plausible numbers, because the volume required is lower than people assume. Say you get thirty inbound messages a week — support, sales, community, DMs, all of it. Say a fifth of them are questions rather than bug reports or scheduling, so six real questions a week, twenty-four a month. Cluster those and you will typically find four or five distinct recurring ones; the rest are one-offs. Four recurring questions a month is one post a week, sourced entirely from people who volunteered the subject. Compare that to the Friday brainstorm, which produces maybe eight ideas, of which three are usable and all of them are guesses.

the topics were never missing. they were sitting in your sent folder.

Writing gets easier too, because the first draft exists already. You answered the question when it arrived, in about four sentences, and that reply is a better opening than anything you will compose from a blank page — it is short, specific, and it was written to be understood by one particular person rather than by everyone. Start from that reply. Widen it: the thing you assumed they knew, the case where your answer would be wrong, the number that makes the tradeoff concrete. This is also the point where AI drafting earns its place, and where it does not. Handing a model your real question and your real four-sentence answer and asking it to expand the structure is a genuinely good use of it. Asking it to invent a topic is how you get the Friday brainstorm again, only faster and blander. And whatever comes back, read every number in it before you approve — a draft that sounds confident about a statistic nobody measured is the one failure mode worth checking for every single time.

The search side of this is a happy accident rather than the point. When you write in the phrasing people actually used, you tend to match the phrasing people actually type, without doing keyword research at all — real questions and search queries come from the same place. That is a reason to keep the wording, not a strategy on its own; what genuinely [moves your search positions](/blog/what-actually-moves-your-search-positions) is slower and less romantic than a well-chosen title. Treat the match as a bonus that costs you nothing, and do not let it talk you into contorting a good sentence.

One honest limit: if nobody is asking you anything yet, this well is dry. A product with four users generates four users' worth of questions, which is not a content calendar. In that case borrow instead of inventing — go where your buyers already talk to each other and read what they ask nobody in particular. It is a weaker signal, because you cannot tell which of those people would ever pay you, and you should treat it as a starting hypothesis rather than confirmed demand. But it beats the blank grid, and it stops being necessary the moment your own inbox starts producing.

The grid is not the problem. Building it before you have anything to put in it is. Fill it only with questions somebody actually asked, and publishing stops being a thing you have to think up and starts being a thing you have to keep up with.