Practice · 03

Optimisation, or the quiet cost of accretion.

Instances grow faster than the teams running them. Customisations accrete, performance drifts, releases get skipped, licence counts creep. Nothing breaks in a way you can point to. But somewhere along the way, the platform stopped feeling like an asset and started feeling like a tax.

What I notice

Two or three releases behind, with the upgrade quietly deferred each quarter. Form load times that everyone complains about and nobody can fully trace. Licence counts climbing faster than headcount. Heavily customised instances inherited by teams who don't know which customisations are load-bearing and which are leftovers from a project that ended in 2019.

And underneath: the absence of platform governance. Not in a heavy-handed, change-control-board sense — in the much simpler sense of "is anyone deciding what should and shouldn't get built?" Without that, every well-intentioned customisation eventually compounds into the next person's problem.

What I've learned

Technical debt isn't a developer problem — it's a budgeting problem. Customisation costs are paid every release, by everyone who touches the instance afterwards. The senior architect's most underrated skill is being the person willing to say "we should remove this" in a room full of people who built it.

Upgrade-readiness is a discipline, not an event. The teams that upgrade smoothly aren't the ones with magic tooling — they're the ones who never let themselves get more than one release behind in the first place. Compounding cost, compounding cure.