Ask a business what its systems cost and you will get the licence fees. That number is usually accurate and almost always irrelevant. The real cost of a software estate is not what you pay for it — it is what your team does to compensate for the gaps between the parts.
The tax nobody invoices for
Each disconnection creates a small, permanent overhead. A number re-typed from one screen into another. A weekly export reconciled by hand. A question that requires opening three systems to answer. None of these is worth a project on its own, which is precisely why they survive for years.
They compound in a specific and predictable way:
- Time — the same information handled repeatedly by people paid to exercise judgement
- Errors — every manual transfer is an opportunity for one, usually discovered late
- Latency — decisions wait on a report that has to be assembled before it can be read
- Disagreement — two departments with two sources arguing about which number is right
- Ceiling — growth requires more administrators, so scale stops being profitable
Why it persists
Because no single instance is expensive enough to escalate. The person re-keying orders has been doing it for three years and no longer notices. The finance team has built a spreadsheet that works. Everyone has adapted, and adaptation makes the cost invisible.
The most expensive processes in a business are usually the ones everybody has stopped noticing.
Making it visible
The exercise we run first is simple and does not require any technology. Take one transaction — an order, a client onboarding, a work order — and follow it from origin to completion. Count every system it touches, every point a human moves information between two of them, and how long it waits at each stage.
It is common for a process people describe as taking an afternoon to turn out to span four days, with about ninety minutes of actual work in it. The rest is waiting and re-entry. Once that is on a page, the business case stops needing to be argued.
Integration usually beats replacement
The instinct when this becomes visible is to replace everything with one platform. Sometimes that is right. More often it is an expensive over-correction that disrupts a working business to solve a problem integration would have solved.
Most estates contain at least one system worth keeping — usually the one finance or operations knows deeply. The better question is rarely what to replace. It is what to connect, and in what order.
Start with whichever connection removes the most manual handoffs. That is almost never the most technically interesting one, and it is almost always the one that changes how the business feels to work in.