Ryan Coughlan

side quests

Products

Things I build for myself, usually because I wanted them to exist. Each one is a chance to practise the whole loop — idea, prototype, users, iteration.

Crunchtime

live

Football predictions with your mates. Pick the scores, share a link, settle it in a mini-league.

Read the story →

YourBookPal

live

Track what you're reading and talk to it — the themes, the easter eggs, the bits you missed.

write-up soon

Celery

beta

FODMAP and macro tracking that doesn't make eating feel like admin.

write-up soon

Mimesis

wip

Language learning by imitation — shadow real speech until it's yours.

write-up soon

how I take a product from idea to traction

The process

The same loop runs on everything here, from a side project to a funded startup. It isn't a straight line: each stage can send me back a step, and each pass cuts away more of what doesn't matter.

  1. 01before any code

    Validate the problem

    Talk to the people who live with the problem: what they chase, where the admin burns them out, why existing tools get abandoned. Write down the riskiest assumption and decide what evidence would kill it.

    • interviews
    • desk research
    • Miro
    • Notion

    Crunchtime: weeks with league organisers before a line of code.

  2. 02days, not sprints

    Prototype to learn

    Build the cheapest thing that tests that assumption, not the product. Working prototypes beat specs: something people can click, and engineers can build from directly.

    • Claude Code
    • Figma
    • Next.js
    • LLM role-play testing

    Crunchtime: Quick Predict — no login, picks live in the URL.

  3. 03measure, cut, repeat

    Iterate with evidence

    Put it in front of real users and watch what they do, not what they say. Cut what doesn't earn its place. Each pass should be simpler and hit harder.

    • PostHog
    • Mixpanel
    • Hotjar
    • A/B tests
    • SQL

    Athena Blue: 150 questions → a 20-minute adaptive intake.

  4. 04small, safe, often

    Ship like it matters

    One roadmap with stable IDs so nothing gets quietly re-argued. Staging that reproduces the interesting states. A security pass before anyone else's data goes in.

    • Vercel
    • Supabase
    • Postgres RLS
    • regression suites

    Crunchtime: two privilege escalations found and pinned before launch.

  5. 05habit, not sign-ups

    Grow and keep them

    Design the growth loop into the product rather than bolting it on. Pick the funnel that really predicts success, and one health metric that tells you when it's quietly dying.

    • funnels
    • retention cohorts
    • behavioural nudges

    Orbit: 20% MoM growth; +36% referrals from behavioural nudges.

and then→ back to 01 with what the users taught me.

also

Automations & websites for small businesses

Moving to its own studio site. Ask me about it in the meantime.

studio · soon