Four phases, milestone-based, transparent. No surprise invoices, no silent scope creep, no "we'll figure it out at launch" energy.
Most builds land at eight weeks typical, twelve in the worst case. The four phases below describe the structure — actual time depends on what got decided in week one.
The same pair of hands runs Scope through Support. No handoffs, no account managers, no telephone-game between phases.
One call, three sharp questions, one architecture proposal. I qualify the fit before anyone commits.
Discovery sprint (1-2 weeks) produces: problem framing + user map, stack + architecture decision, scoped roadmap with estimates, working prototype where it adds value.
Small commits, fast feedback. Production-grade from day one — observability, backups, docs.
Milestone-based. Design + code in the same hands, so no telephone-game loss between surface and system. Staging deploys from commit 0.
Launch with monitoring wired, CI green, rollback plan ready. No 'we'll fix it in the next sprint.'
Vercel preview → production with DNS cutover runbook. Sentry + Vercel Analytics active before first real user touches it.
Retainer or hand-off — your call. Documented either way, so the next person knows what they're looking at.
Retainer: reserved hours/week, priority response, quarterly roadmap reviews. Hand-off: onboarding doc + architecture diagrams + 2 weeks of overlap.
One call, three sharp questions, one architecture proposal.I qualify the fit before anyone commits.
If discovery surfaces that this isn't a good fit (wrong scale, wrong tech, wrong moment), I say so and we don't proceed to Build. You keep the discovery output — architecture, roadmap, prototype — with no lock-in.
Small commits, fast feedback.Production-grade from day one — observability, backups, docs.
Mid-build scope expansion: if the brief grows mid-engagement, we re-quote the delta as a separate scope. I won't absorb it silently — that's the path to dishonest estimates and burnout.
Launch with monitoring wired, CI green, rollback plan ready.No 'we'll fix it in the next sprint.'
Production isn't done at the deploy commit. It's done when monitoring runs green for 7 days under real load. Budget that week separately if you can.
Retainer or hand-off — your call. Documented either way,so the next person knows what they're looking at.
I cap maintenance retainers at 4 clients (separate from the 2-3 active build engagements above). If I'm full, I'll say so and refer to vetted alternatives — not pretend I have capacity I don't.
Ready to scope? Book discovery or send a brief —
milestones once the shape is clear.