Many users change the same resource at once.
loading…
The hard part is usually the interaction between rules, systems, and failure paths.
Engineering Complexity ≈
Entities × State Transitions × Integrations × Failure Modes × Production ConstraintsUse the formula below as a checklist, not arithmetic. A large content site can be routine. A small booking flow can be difficult when it combines capacity, payment, permissions, concurrency, and a real room that must actually be available.
| Dimension | Content platform | Booking platform |
|---|---|---|
| Core state | Viewed, completed, subscribed | Available, reserved, paid, cancelled, refunded, waitlisted |
| Real-time need | Low | High when capacity is limited |
| Integrations | Auth, billing, content | Payments, calendar, notifications, access, maps |
| Failure cost | Progress may not save | Double charge, overbooking, locked door |
| Operations | Manage content | Schedule classes, refunds, waitlists, no-shows, support |
Many users change the same resource at once.
Errors require reconciliation, refunds, and auditability.
Different roles may see or change different data.
Software must match a room, shipment, device, or human schedule.
Your workflow depends on APIs you do not control.
Some mistakes cannot be silently corrected.
A polished number is not a senior estimate. A senior estimate says where the team has evidence, where it is guessing, and which unknown could move the date.
What not to say “This should take about three weeks.”
What I would say “The core workflow is likely three to four weeks. The largest risk is payment reconciliation. I would run a two-day technical spike before committing to the final date.”
What matters A project is difficult when many states, systems, and failure paths interact—not simply when it has many pages.