clouddevopsplatform-engineeringagiledelivery-velocityterraformgithub-actions

Why Moving Fast Matters More Than Getting Everything Perfect

Erik Osterman
byErik OstermanCEO & Founder of Cloud Posse
Feb 08 2025

Let's be blunt:

Perfect is the enemy of done.

And nowhere is this more obvious than in cloud platform engineering.

I've seen too many teams fall into the same pattern:

We're being responsible. We're researching the right tools. We're building prototypes. We're connecting things. We're documenting everything. We're almost ready to build "the platform."

And month after month, every standup sounds the same:

  • The same status updates.
  • The same excuses.
  • The same "almost ready" platform.

Meanwhile, in the rest of the org:

Everyone is still stuck fighting fires.

Developers are twiddling their thumbs waiting for "the platform."

Leadership is questioning where the value is.

Seen this movie 100 times.

Why Time to First Value is Critical for Team Momentum

You are not delivering value until your first production service is running on 'the platform.'

Period.

Every month spent "researching and planning" without shipping erodes trust:

  • Engineers lose faith.
  • Stakeholders lose patience.
  • Leadership starts looking for the next initiative to fund.

But when you hit time to first value quickly:

  • Developers get hands-on.
  • The platform gains credibility.
  • Team momentum builds.
  • Leadership sees progress.

Speed builds trust. And trust keeps the initiative alive.

How "Getting Stuck" on Perfect Architecture Burns 6-12 Months

Here's the trap:

The team thinks they're being productive.

Every status update sounds fine:

  • "Still 'finishing up' the prototype."
  • "Ran into some IAM issues, but on the right track."
  • "Waiting for upstream changes to be merged, and then everything is good to go."

But here's what really happens:

They burn 6 to 12 months:

  • Researching which tools and techniques they'll need.
  • Prototyping — showing they can get a minimal version online.
  • Chasing new, shiny tools — which evolve under them.
  • Writing long design docs to retroactively justify decisions.
  • Drawing beautiful diagrams — instead of shipping.

Meanwhile:

  • The rest of the org is stuck fighting fires.
  • Developers are waiting for "the platform."
  • Leadership starts losing patience.

And because it's taken so long, now they have to keep updating their architecture as tools evolve — and they've underestimated how hard the real integrations will be.

Elegant architectures don't always translate to elegant implementations. The devil's in the details.

Nobody Is an Expert in All of This

Here's another pitfall:

No one engineer is an expert in all the things you need to get a cloud platform off the ground.

And if they are? Congratulations — they are now your single point of failure.

It's a double-edged sword.

Realistically:

  • Most engineers have deep expertise in one or two areas.
  • They might have implemented pieces of this stack.
  • They might be very well-read — book smart — in areas they haven't touched.

But there's a big gap between:

  • Knowing something from reading about it
  • And having hands-on experience shipping it to production — and seeing where it breaks.

And this is why architecture traps happen:

  • Over-designing the parts you know well.
  • Underestimating the gaps in the parts you don't.

Use Reversible Decisions to Move Faster

One of the best ways to avoid this trap:

Make reversible decisions first. Defer irreversible ones until you have real-world feedback.

That means:

  • Stand up a platform that can run production workloads — even if it's not perfect.
  • Get fast feedback from real use.
  • Iterate.

Most early decisions are reversible:

  • CI/CD patterns
  • Observability stack
  • IAM patterns
  • Repo structure

What's not reversible:

  • Spending 12 months "designing" and shipping nothing.
  • Burning trust with leadership and stakeholders.

Why Layered, Incremental Architectures Win (For Real)

This is one of the hardest-won lessons we've baked into our reference architecture:

It's still in use 10 years later — and still outperforms 99% of internal platforms we see.

Why?

Because it was built to be layered and incremental:

  • You're not locked into a rigid architecture.
  • You can swap out pieces as better tools emerge.
  • You don't have to rebuild everything to modernize it.
  • You avoid architecture dead-ends.

Real-world platforms survive because they can evolve.

Don't Over-Invest in Static Perfection

Another common trap:

Don't over-invest in elaborate workarounds for limitations that will likely be solved soon.

Too often, teams:

  • Find a limitation.
  • Spend weeks building a complex workaround.
  • Get stuck maintaining it.
  • Miss out when the tool evolves and the limitation goes away.

Better path:

  • Implement simple patterns that can evolve.
  • Assume many limitations will be solved (they usually are).
  • Don't let today's workaround become tomorrow's technical debt anchor.

Finally, A Platform Your Engineers Can Use — and Understand

One of our core principles:

A platform your engineers can start using now — and actually understand — because it's been documented, battle-tested, and evolved in production.

Too many "perfect" platforms:

  • Never ship.
  • Or ship something so complex that only one engineer understands it.

That's not winning.

Winning is:

  • Delivering something your whole team can use.
  • Solving real business problems today.
  • Evolving it as you go.

If you take one thing from this post, let it be this:

It is more valuable to launch fast and evolve than to delay for perfect platform.

Every month of delay is value you'll never get back.

You don't need perfection on Day One. You need:

  • A platform your engineers can start using now.
  • A platform they can understand.
  • The ability to evolve it continuously.
  • Fast feedback from real use.

That's how platforms win. That's how teams win. That's how businesses win.


Talk to an engineer: If your team is stuck in "design" mode — or burned months chasing perfect architecture — let's chat. We've helped dozens of teams break that cycle and deliver real value in weeks.

Erik Osterman
Erik Osterman
CEO & Founder of Cloud Posse
Founder & CEO of Cloud Posse. DevOps thought leader.

Share This Post