By Oliver Niemandt, general manager and head of sales, Cloud On Demand

Fragmentation is expensive. It is also unowned and unmeasured, which is why it never has to justify itself.

For South African organisations heading into next year’s budgets, the cloud conversation is mostly about spend: what it costs, and where it can come down. Both are reasonable questions. Neither reaches the cost of the complexity itself, and that cost is never assembled into a single number, which is precisely why it survives every budget round.

Cost scrutiny is good at finding visible waste. What it cannot see is a cost with no line item. A licence overlap between two platforms bought by different teams in different years. Engineering time spent moving information between consoles that were never designed to speak to each other. A control enforced properly on one platform and assumed on another. A decision delayed because no single person has the full picture. Each is real, and each is booked somewhere else: under software, salaries or risk, or not recognised as a distinct cost at all.

It also compounds. Adding a platform does not add one more thing to manage. It adds another set of connections to maintain, and another place a control has to be built and proved. The environment grows in cost faster than it grows in size, and it grows in exposure the same way. Fragmentation rarely creates one large hole. It creates smaller gaps at the joins between platforms, where each is somebody’s second priority and nobody’s first.

Much of this has a perfectly rational origin. Organisations build multi-vendor environments to retain flexibility, reduce dependency on any single provider and pick the strongest option for a particular requirement. Those were sound decisions on their own terms, and the argument is not that multi-vendor strategies were a mistake. It is that the standing cost of holding all the pieces together is far harder to account for, and it continues whether or not the flexibility that justified it is ever used.

Which is why I would resist jumping straight to consolidation. Simplification is often the right destination, but it is the second decision. Measurement is the first, and skipping it is how sound consolidation cases get turned down. Without a view of what the complexity is costing, the case rests on preference rather than evidence.

Measurement is more achievable than it sounds. It starts with one person owning the number for the environment as a whole, not for the platforms inside it. After that it is a matter of counting what is genuinely duplicated in function rather than in name, the engineering time spent on integration and reconciliation rather than delivery, and the number of times the same control has to be built and evidenced separately. None of that requires new tooling. It requires a decision that the number is worth producing. It will not be precise, and it does not need to be. A defensible estimate changes a conversation that platform-level reporting never will.

For partners, this is the more useful conversation to have while clients plan next year. Anyone can arrive with a saving. Arriving with a cost the client has carried for years and never seen written down is a different conversation entirely. The value is in making it visible, not in having the answer ready before it is.

Complexity was never free. The organisations that spend well next year will be the ones prepared to put a number on it this year, before deciding what to do about it.