Put the design system in a markdown file the model can read
Chief builder at Bergside, creator of Flowbite and TypeUI
TypeUI's design skills are markdown: buttons, navigation, spacing, accessibility, written by a co-founder with fifteen years in UI and UX. Zoltan would rather give a coding model that file than start in Figma. A Figma board is a picture the model has to guess at. A markdown skill is an instruction it can follow. If you sell taste to people who generate code, write the taste down in the format the model already reads. Then charge for the file, the motion, the sound, and the audit, not for another component zip.
Related advice
Let one big spike tell you where the market is
Bergside sold one Tailwind and Figma kit for 3,000 euros in a single Black Friday night, and that one spike redirected the entire company toward Tailwind components. A sudden, sharp sales response is the market telling you where the demand actually lives, in a way no survey or roadmap meeting ever will. When something you ship spikes far beyond the rest, do not treat it as a lucky one-off. Treat it as a map. Drop what is flat and pour your effort into the thing the market just voted for with its wallet.
A color change is not a test. Watch the path.
TypeUI could keep two Airbnb-style sites feeling like the same product after a color change, and still lose 79.82% of signups by sending people to pricing before login. The palette held. The path did not. Heatmaps, dead clicks, rage clicks, and scroll depth are there so a traffic number has to explain itself. If you ship a redesign, read the step where people stop before you congratulate the visual system. A consistent feeling is not a conversion.
Use search demand to decide what features to build
Before building a new feature or component, check what people are actually searching for. Use tools like Semrush to find keywords with real demand. When Flowbite saw search volume for "avatar tailwind" and "datepicker tailwind," they built those components and captured the traffic. Let search data guide your roadmap instead of guessing what users want.
Set up analytics before you start scaling traffic
Do not invest in scaling traffic until you have analytics running. Use Microsoft Clarity (free) or Hotjar for heatmaps, rage click detection, and session replays. Watch your bounce rate and average time on site (one minute is a reasonable benchmark to start). If users leave immediately, more traffic just means more people leaving. Fix the experience first, then scale.
Extracted from
When the Code Got Cheap