I make developer tutorials on the side, and for a long time my topic pipeline was "whatever bug I fought last." That produces the occasional hit and a lot of misses, because the bugs I happen to hit aren't a representative sample of what my audience is stuck on.
About a year ago I replaced the vibes with a ritual. Last Sunday of the month, coffee, thirty minutes, one page in my notes app. Here's the whole thing, phase by phase, in case it's useful to steal.
Phase 1 — Dump the candidates (5 minutes)
I open a note and brain-dump every topic I could plausibly make a video about that month. No filtering. Bugs I fixed, questions from my Discord, a library I finally understood, something a coworker asked twice.
The goal here is volume, not quality. I usually land on 12 to 20 candidates. Most will die in the next phase, and that's the point — this is the cheapest place to kill a bad idea, before it costs me a weekend of filming.
Phase 2 — Pull real demand (10 minutes)
This is the phase that changed my numbers, so I'll spend the words here.
I take each candidate and check whether people actually search for it, and in what words. I use AllyHub's free tool to look up the YouTube most searched keywords around a topic — I type in a candidate like "react useEffect cleanup" and it shows me what people search near it, which phrasings pull more interest, and which videos currently surface for those queries.
Two columns go in my note for each candidate: rough relative demand, and how strong the existing coverage looks. I'm not chasing precise numbers — the tool gives relative signal, not a view-count promise, and I treat it that way. I just want to know: is anyone looking for this, and is it already answered to death?
A candidate I was sure about — "clean architecture in Node" — turned out crowded and well-served. One I nearly cut — "prisma migrate drift resolve" — showed steady interest with only scattered forum answers behind it. Column data flipped my instinct completely.
Phase 3 — Score and rank (5 minutes)
Now I sort. My scoring is deliberately dumb: I want the intersection of real demand and weak existing coverage. High demand plus ten strong videos already there? Skip — I won't win that. Real demand plus stale or thin coverage? That goes to the top.
I also flag anything with essentially zero search volume that I still want to make. More on those in a second. They don't get scored against the demand topics — they're a separate bucket.
By the end of this phase I have a ranked shortlist of three or four topics, and I pick the top one or two to actually produce that month.
Phase 4 — Reality-check with my own judgment (5 minutes)
The data tells me what's wanted. It does not tell me whether I can make a good video about it, and I don't let it pretend otherwise. So the last five minutes are a gut check on each shortlisted topic: do I actually understand this well enough to teach it clearly, do I have a decent example on hand, will I still care about it once the camera's rolling.
A topic can have great demand and still be a no for me because I'd make a shallow video. Demand is necessary. It's nowhere near sufficient. The quality is entirely on me once I hit record — no keyword tool is going to save a tutorial where I don't know the material.
The exemption I keep on purpose
That separate bucket from Phase 3 matters. Some videos I make regardless of search demand, and I refuse to route them through this ritual.
When I release an open-source tool, I make a walkthrough — nobody's searching for a thing that launched yesterday. When I have an opinion about where a framework is heading, I record it. When something breaks in a way that taught me something, I do a post-mortem. These are bets on trust and voice, not discovery, and running them through a demand filter would just talk me out of the work worth doing.
So the ritual isn't "only make what people search for." It's "know which bucket a video is in." Discovery videos earn their spot through demand. Authority videos earn it through having something to say. The whole point of thirty minutes on the last Sunday is to stop confusing the two — which I did, expensively, for about two years before I wrote the ritual down.
Top comments (0)