Aaron Francis - how to make videos developers want to watch
Aaron Francis turned a video about changes to PHP into a 300,000-view hit. His method is to hide a genuinely useful technical lesson inside a package people actually want to click — without letting YouTube dictate his life.
Aaron Francis built PlanetScale's acclaimed YouTube channel and now publishes on his own channel and runs Try Hard Studios, helping developer-tool companies make better videos.
We break down his process from finding an idea to recording in short chunks, choosing an audience and angle, learning from comments, and making a company channel feel like it is run by recognizable human beings rather than a corporate blob.
The Q&A below is experimental — extracted from the episode transcript with AI. If you spot a mistake, let us know at jack@scalingdevtools.com.
How do you come up with ideas for developer videos?
Step zero is caring about the subject. If you don't care, it is hard to make somebody else care. Once you start publishing in an area, ideas stop being the problem: you prime your brain to notice relevant documentation, conversations and examples everywhere. Before you start, you wonder how you will find an idea; afterward, you wonder how you will find time for all of them.
How do you make a dry technical subject entertaining without turning it into clickbait?
Start with a real educational nugget, then wrap it in personality, excitement, humor and a clear angle. The title and thumbnail should pique curiosity, but the video has to fulfil their promise. I made a video about changes to PHP over the previous decade, but packaged it as 'PHP doesn't suck (anymore).' It reached about 300,000 views; a plain 'What's new in PHP' list would not have earned the same attention.
Did the entertaining style come naturally, or did you develop it deliberately?
Be yourself, because playing a character is exhausting, but turn yourself up a notch or two on camera. People tend to appear more downbeat on video, partly because it feels strange to sit alone in a room gesturing enthusiastically at a camera. Find creators whose energy, pacing or delivery you admire, work out what aligns with you, and borrow those pieces. Before recording a course lesson, I watch the previous few videos so I can match their energy and tone.
How much do you plan before recording?
I use bullet points rather than a full script. I know the signposts — set up the project, install the dependency, show the demo, sign off — but improvise the connective tissue. That is where the jokes and liveliness appear. I record in chunks of two or three minutes and use natural visual changes, such as moving from the editor to a talking head and then to the browser, as edit points. I may take a 30-second introduction five or ten times, and sometimes finish a whole video, realize I now understand the best route through it, and record it again in half the time.
How long should those videos be, and how much should the algorithm influence that?
For YouTube I usually aim for roughly six to fifteen minutes, and I would split a 45-minute subject into separate videos. I want to give the platform what helps it promote the work, but I do not want to make clickbait thumbnails or pretend to be surprised. The useful question is where what YouTube wants overlaps with what you want to do with your life.
Who were the PlanetScale database videos for?
The audience was the 'meaty middle': application developers who do not want to be DBAs but do want to become expert users of a database. That choice shaped the packaging. Instead of leading with virtual generated columns, I could lead with stopping free-tier abuse through Gmail plus-addresses, then teach the database technique inside that application problem. The audience recognizes the problem before they know they need the underlying concept.
What exactly is an angle or story for an educational video?
It is the hook, narrative, packaging or point of view that explains why this audience should care. In a paid course, somebody has already chosen to learn the subject, so you can teach it head-on. On YouTube, people do not click on things they do not know they need to know. You put an esoteric lesson in terms they already understand and make a promise the lesson can genuinely deliver.
How should a DevTools company go broad on YouTube?
Teach the broad problem your tool is a solution to, not only narrow integration tutorials. A video called 'How to integrate Next.js with Clerk' is useful for a specific middle- or bottom-of-funnel audience. Something like 'Eight authentication mistakes you're probably making' can reach far more people, teach them something useful, and let discovering that Clerk handles those problems be the second-order effect. At PlanetScale, I could teach databases broadly and let viewers discover where they might host one afterward.
What do you learn from YouTube analytics after publishing?
I am mostly vibes-based. I once compared the first 30 seconds across videos and learned to get to the point faster, and I compare view counts across my last few releases. But I care more about qualitative feedback. I read every comment: if hundreds of people repeatedly mention the humor, pace or examples, that explains something about why the larger audience responded too. It also lets me talk with the people who spent time with the work.
Where should developer creators look for inspiration?
Look outside DevTools at fields that have understood YouTube for years — construction, woodworking, furniture making or even lawn mowing. Watching only videos about how to make YouTube videos leads to the same MrBeast advice. Develop your own taste by noticing what makes work in other niches funny, sincere, detailed or compelling, then bring even a small part of that craft back to a space still full of hour-long webinars with no custom thumbnail.
Links & references
- Aaron Francis's YouTube channel ↗
- Aaron Francis's X ↗
- Mostly Technical Podcast ↗
- Try Hard Studios ↗
- Aaron's handwriting robots ↗
This episode is brought to you by WorkOS.