Insights

The MVP trap

The mistake is not the MVP. The mistake is treating the MVP as a foundation instead of a proof of concept.

I have seen this pattern more times than I can count, most often in logistics tech.

A founder builds an MVP. It handles 10 shipments a day. Investors like it. The team celebrates. Six months later they are doing 500 shipments a day, and everything is on fire.

Not because the product does not work. Because it was never built to work at this scale. It was built to prove the idea. Those are two completely different things.

The moment the old way of building stops working

There is a moment in every product where the old way of building stops working. You do not always see it coming, but you feel it:

  • Deploys take longer every month.
  • A change in one place breaks something unrelated.
  • Onboarding a new engineer takes weeks.
  • There is a growing list of code nobody wants to touch.
  • Nobody is quite sure what the system does any more.

This is not a people problem. It is a transition problem. The product stopped being a startup and became a system, and nobody changed how they were building it.

What the MVP trap looks like in practice

In logistics, the cracks tend to appear in the same places:

  • The data model was designed around a single warehouse. Now there are four, and nothing maps correctly.
  • The driver app talks directly to the database because that was the fastest way to build it. Now every new feature touches three systems at once.
  • Status updates are hard-coded for a linear flow. Real logistics is never linear: exceptions are the rule, and the system has no way to handle them gracefully.
  • Third-party integrations were bolted on one at a time. Nobody owns the data contract between them, so when one changes the whole pipeline breaks.

Swap shipments for orders, patients or payments and the same pattern appears in SaaS, healthcare and fintech.

Startup mode and system mode

In startup mode, speed is everything: fast decisions, accepted shortcuts, ship. That is the right way to build an MVP. It works until the shortcuts start talking to each other.

In system mode the rules change. Every decision has downstream consequences. The architecture has to be intentional, because the cost of fixing it later compounds fast.

None of this is the founder's fault. An MVP is supposed to be fast. The mistake is not the MVP. The mistake is treating the MVP as a foundation instead of a proof of concept.

How to get out of the trap

The transition from MVP to a production-grade system is not a feature sprint. It is an architecture conversation. The questions to answer before you scale are not about what to build next. They are about what the current system cannot support, and what has to change before it can.

In practice that usually means:

  • Mapping the system as it really is, including the integrations nobody owns.
  • Putting tests, a release pipeline and monitoring around it before changing anything.
  • Re-architecting the weak parts one at a time, not rewriting everything.
  • Making the data model and the data contracts explicit.

Most teams skip that conversation because it feels like going backwards. It is not. It is the work that makes everything after it faster.

If your product is holding up at low volume but showing cracks as you grow, the cracks are telling you something. Listen before they become fractures.

Where I can help

Architecture Clarity Sprint

4 to 6 weeks, fixed scope

An architecture review and technical debt assessment: a clear picture of what your system supports today, where it will break next, and what to fix first. The best place to start, and the groundwork any future CTO will need.

  • Map of your current system and its real constraints
  • Risk list ranked by business impact
  • Scale roadmap for the next 12 to 18 months
  • A prioritised plan your team can act on
Architecture Clarity Sprint: details

MVP to Production Rescue

Project based

For products that proved the idea but cannot carry the growth. I turn the MVP into a system that is safe to build on.

  • Re-architecture of the weak parts, not a rewrite
  • CI/CD, release standards and observability
  • Database and performance tuning
  • Cloud cost reduction on AWS and GCP
MVP to Production Rescue: details

More essays

Hire the CTO second

A CTO who joins a system they did not build spends the first 90 days discovering what exists. Get architecture clarity first, so they arrive with a map and start executing.

Agents need permissions, not instructions

A sentence in a system prompt is a suggestion, not a constraint. If your agent can reach it, your agent can break it. Build guardrails that make the wrong action impossible.

Tell me what is breaking.

I will tell you honestly whether I can fix it.