1. When is a customer actually gone?

    A thirty day lapsed window is wrong for nearly every customer, and wrong in both directions at once. What to use instead, and what it cannot tell you.

    3 min read

  2. What an accelerator is actually for

    One of nine from about 1,700 applicants, ten weeks in, company wound down. What the programme was genuinely worth, and what it could never have fixed.

    3 min read

  3. Why we wound Presterly down

    34 Shopify stores and €19M of order history in ten weeks, then we stopped. It was not the product and it was not the market. A founder's post-mortem.

    4 min read

  4. Building a CRT that behaves like a CRT

    Most CRT effects are a scanline overlay and some glow. Deriving everything from one premise, a beam painting phosphor, gets you somewhere better.

    4 min read

  5. Multi-tenant Shopify apps: what I'd get right first

    Building multi-tenant Shopify apps: the decisions that were expensive to change and the ones that turned out not to matter at all.

    3 min read

RSS feed