Waiting for an engineer, then learning to build
Crowley spent years with notebooks full of ideas and no one to build them. The fix was to learn enough to build them badly himself.
Stop sketching and start building.
Before Dodgeball, the mobile check-in service he later sold to Google, Dennis Crowley spent several years drawing ideas for products and waiting to meet an engineer who would build them. The engineer never arrived. So he taught himself enough programming, first with Microsoft’s ASP and Access and later with PHP, to build rough versions on his own. He told the story in an interview in the Founder Stories series this March, and this is the advice he drew from it. Foursquare, the check-in service he started in 2009 after Google closed Dodgeball, began the same way: a crude working version, used by friends, improved in public.
The line targets a specific trap. Sketches, wireframes and slide decks feel like progress, and they cost almost nothing to make. They also teach almost nothing, because a drawing of a product cannot be used. A working prototype, however ugly, can be handed to someone and watched. The questions it answers, whether people come back, what confuses them, what they try to do that it does not support, are the ones that decide whether an idea is any good. Dodgeball itself was crude: users texted their location and the service replied with friends nearby. It was ugly and limited, and people used it, which no drawing could have shown.
Sketching still has its place, and the line is too absolute about that. Where building is expensive or slow, a drawing is the cheap test. Hardware, buildings and anything involving safety get sketched and modeled for good reason. Even in software, an hour of drawing screens can save a week of code written in the wrong direction.
The real target is the sketch that never ends: the notebook that fills up for years while nothing gets made. Crowley’s own fix is the most useful part. Waiting for a builder is a choice, and learning enough to make a rough version is usually within reach. For a founder with no technical skill at all, the rough version might be a spreadsheet, a form and a manual process behind it, run by hand until it proves worth automating.