The five inefficiencies I find in almost every business

I have spent a lot of time now sitting with the people who actually do the work in a business, rather than the people who describe it. The gap between those two accounts is where most of the money goes. Five patterns come up so consistently that I now look for them before I look for anything else.

1. The reconciliation spreadsheet

Somewhere in almost every business there is a spreadsheet whose entire purpose is to make two systems agree. Someone exports from one, exports from the other, and matches them by hand — weekly, sometimes daily. It is never in anyone's job description. It is usually done by the most capable person in the department, because they are the only one who understands both systems.

This is the single most reliable indicator I know of a missing integration. The spreadsheet is not the problem; it is the symptom, and it tells you exactly which two systems were never connected.

2. The approval that outlived its incident

Ask why a particular sign-off exists and you will often get a story about something that went wrong years ago. The control was added, the person who added it has left, and the circumstances that justified it no longer apply. Meanwhile every transaction still queues behind it.

Controls should be costed like anything else. If an approval step adds two days to a cycle and has not caught an error in three years, it is not risk management — it is a habit with a governance label on it.

3. The report nobody reads

Every business produces a monthly pack that takes someone days to assemble. When I ask who acts on each page, the honest answer is frequently nobody. The report survives because stopping it feels riskier than continuing it.

The test I use is simple: stop producing it for one cycle and see who asks. It is remarkable how often the answer is no one, and how much capacity that releases.

4. The number nobody trusts

This one is the most expensive and the hardest to see, because it does not show up as wasted hours — it shows up as decisions not made. When a figure has been wrong often enough, people stop using it. They build a private version instead, and now the business has two answers to the same question and no way to reconcile them.

The fix is rarely a better report. It is finding out why the original number was wrong, which is nearly always a process problem upstream rather than a reporting problem downstream.

5. The manual step at the boundary

Work flows smoothly within a team and stalls at every handover. Sales to operations, operations to finance, finance to the client — each boundary has someone re-keying, chasing or reformatting. The cost is invisible on any single transaction and enormous across a year.

Boundaries are where I look first now, because they belong to nobody. Each side assumes the other owns the gap.

What this means in practice

None of these are technology problems, which is why buying software rarely fixes them on its own. They are process problems that technology can remove once you have found them and costed them. That order matters: diagnose, cost, then build — and be willing to conclude that a particular problem is cheaper to live with than to solve.