The configurator is the sales floor

Most of what goes wrong with a configurator is decided long before anyone opens a design file. The decision gets made in a procurement call, or in a template, or by whoever owned the site three vendors ago — and then everybody downstream spends two years working around it.

This is a placeholder body. It is here so the reading column, the subheads, the pull-quote and the figure strip can all be judged at real length before the actual writing lands. Replace it.

What the buyer is actually doing

A buyer configuring a boat is not browsing. They are building the thing they intend to own, and they will do it four or five times across two weeks, usually on a phone, usually while comparing against a competitor's PDF. Every second of load time is a chance for them to stop.

Nobody abandons a configurator because it was ugly. They abandon it because it did not answer the question they came with.

Where the weight goes

Geometry, textures, and the same hull exported four times by four different people. The fix is boring: one pipeline, one export budget, and a rule that nothing ships to production without going through it.

Asset payload, before

136 MB

Asset payload, after

4.6 MB

Models in pipeline

18

LCP, mobile

1.9 s

What we'd do differently

Start with the constraint. Decide the payload budget, the deadline, and who owns the export before anyone argues about lighting. Everything after that is production, and production is the easy part.