◆
No platform team needed
The platform does not require a team to operate. If you have one, they work on your product rather than on keeping the platform alive.
Internal platforms usually begin as an attempt to reduce choices engineers must make and end as another system they must learn. The difference is whether the platform does the work or delegates it back.
Used as the foundation other teams build on, with or without a platform function.
What this looks like in practice
The platform does not require a team to operate. If you have one, they work on your product rather than on keeping the platform alive.
Teams get what they need without filing a ticket and waiting. That is the whole promise and it is usually unmet.
Every team gets the same arrangement because it is the arrangement, not because a standard was circulated.
Who spends what, so conversations about efficiency have numbers in them.
A platform team that sits between product engineers and their infrastructure becomes a queue. The goal is a foundation that needs no intermediary.
Ask product engineers what slows them down and the answer is rarely the technology. It is waiting for somebody else to do a thing that could have been self-service. A platform’s success is best measured by the queue it does not have.
The recommended way being the easiest way, so it gets followed without enforcement.
Internal APIs and common capabilities that teams build on rather than reimplement.
Permissions that are clear enough to audit and simple enough to be correct.
Numbers that make efficiency a conversation rather than an assertion.
We disbanded the platform team and velocity went up. They are all building product now, which is what they wanted in the first place.
No. If you have one, the useful question becomes what they should build instead of maintaining a platform.
Consistency is a property of there being one arrangement rather than of a standard being circulated and hopefully followed.
Yes, within boundaries you set. The alternative — a ticket queue — is where most platform initiatives quietly fail.
Yes, attributable per team, which tends to change the tone of efficiency discussions considerably.
Start on something that will still be right at a hundred times your size, and pay now what a startup should pay.
Move in stages with the old arrangement still running, and cut over only once the replacement has carried real traffic.
Get capacity when the experiment is ready, keep what produced a result, and put models in front of users without a separate platform.
Keep everything at full fidelity, query it without a project, and stop deciding today what you will want to ask next year.
Ship the application your users touch, on an address that stays put, with every version one click behind you.
Build the tools your own teams need, at a cost that makes small improvements worth doing.
Begin with one workload. Keep what you have running. Decide whether to continue once you have seen it carry something real.
No migration required to begin. Nothing to tear down if you stop.