There is a stage in a service business where the work is still good, the people are still good, and everything takes slightly longer than it should. Nothing is broken. Nobody can point at the problem. The founder is busier than ever and cannot say what with.
We see the same six signs almost every time. Each one looks like a people problem or a tooling problem. Each one is an operating-flow problem: work has outgrown the way it moves between people and systems, and the gap is being closed by memory.
1. The same details are typed into three systems before work can move
An enquiry arrives by email. Someone copies the name into the CRM, again into the case system, again into the invoice. Nobody decided that this should happen. It happened because each system was chosen for one job and nobody chose what connects them.
What it is telling you: no record has been named as the source. When two systems disagree, nobody knows which one is right, so people keep both up to date by hand and trust neither.
What changes: repeated data entry becomes connected records and bounded synchronisation. Decide which record is the source before connecting systems, and automate only predictable movement with agreed permissions.
2. Where things stand is pieced together from messages and memory
Ask where a job is and you get a reconstruction: a search of the inbox, a look at the shared drive, a message to the person who was last seen with it. The answer is usually right. It takes twenty minutes.
What it is telling you: the state of the work lives in the conversation about the work. An inbox shows messages, not the status behind them. Nothing holds "where this is now" as a fact anyone can read.
What changes: inbox-driven work becomes structured queues and current state. Put work in a queue that shows its owner, its status and its next action, so waiting work is visible without asking.
3. Routine decisions wait for the one person who remembers the rule
A discount above a threshold, a request that is slightly unusual, a client who has asked for something twice. Each goes to the founder or the senior operator, not because the decision is hard, but because nobody wrote down who else may make it.
What it is telling you: authority is undocumented. Ordinary routing shares a queue with the decisions that genuinely need judgement, and the person with the judgement spends their week on the routing.
What changes: founder-routed decisions becomes explicit ownership and authority. Name who can decide, approve and escalate at each step, so routine routing stops sharing a queue with material decisions.
4. Missing items are chased from memory, on the day they are needed
Somebody knows the client still owes a signed form. Somebody else knows the accountant has not sent the figures. Neither is written down anywhere the other could see, so both are chased on the day they are remembered, which is usually the day they were needed.
What it is telling you: there is no agreed list of what is outstanding, no owner for chasing it, and no place for an item to wait visibly until it arrives.
What changes: chasing from memory becomes an outstanding-items list with named owners. Hold what is outstanding as an agreed list, with an owner for each item and a visible place for it to wait until it arrives.
5. Reporting is rebuilt by hand before every meeting
The weekly view is assembled from exports, each with its own definition of "open" and "closed", by the one person who knows how to reconcile them. It is out of date by the time it is read.
What it is telling you: nobody has agreed which records feed which number. The report is not read from the operation; it is reconstructed from it.
What changes: manual reporting becomes defined operating views on agreed records. Agree the sources, definitions and ownership behind each view, so reporting is read rather than rebuilt.
6. Failed work disappears, and the client finds it first
An automation stops, an integration drops a record, a request lands outside anyone's remit. It is not that nobody cares. It is that nothing made the failure visible to anyone, so it was found by the client.
What it is telling you: the exception route was never designed. Work that falls off the normal path has nowhere to go.
What changes: hidden failure becomes visible exception management. Give failed or ambiguous work a visible exception with an owner, so the next step is a decision rather than a scramble.
What the six have in common
None of them is fixed by more software. Each one is an absence: no named source record, no held status, no written authority, no outstanding-items list, no agreed definitions, no exception route. Add another tool without deciding those six things and you have added a seventh place for work to hide.
They are also cheap to describe. You do not need access to anyone's systems to write down which of the six you recognise, which workflows show it most, and who currently absorbs it. That description is most of what a diagnosis needs.
How to read your own business
Take one workflow that matters, an enquiry, a case, a job, and follow it end to end on paper. At each step ask four questions. Where does this live right now? Who owns the next action? What has to be true before it moves on? Where does it go if that is not true?
Every question you cannot answer from a record is one of the six signs. Write those down. That list, in your own words, is what our free Operating Infrastructure Audit reads. It is also, on its own, the most useful hour you will spend on operations this month.
What all six responses have in common is that they move a decision out of someone’s head and into the design of the work. Routine work follows a pattern; material work gets attention. Whether that changes productivity, cost or throughput is for you to measure.
A comparison of operating mechanisms, not a measured result. No numeric outcome is claimed for any of them. The situations described are composites and name no client.