Start with watching.
The first thing you do on any job is nothing. A full week of nothing but watching. Say so on day one. You are slow, and you own it. You have watched too many people fix the wrong thing. They see a symptom, grab a tool, spend the money, and three months later the same symptom is back, only now they are also out the money.
Watch the work actually happening. Who does what, in what order, and where the pile builds up. The pile is the clue, the stack of things waiting on someone else. Find the pile and the diagnosis is half done.
Walk one order to the bank.
Then trace one order from the moment a customer asks to the moment the money lands in the bank. Just one order. Write down every step and how long each step takes. If the whole thing should take three days and it takes fourteen, you do not need a report to locate the problem. Walk the fourteen days and find the step where it sat for a week. That step is the disease. Everything else is a symptom. You do not diagnose what you have not walked.
Read the ledger, not the dashboard.
Read the last thirty days of real transactions. Not the summary, not the dashboard, the actual line items. The dashboard is a story someone tells. The ledger is what happened. Cash is the only number you cannot argue with. Everything else is an opinion, and most opinions in a business belong to the owner. Read it line by line, like it is your family's money. You have seen what happens to owners who skip this.
Name the disease after three sightings.
Do not name the problem until you have seen the same symptom three times. Once is an accident. Twice is a coincidence. Three times is a pattern, and a pattern has a cause. Most people name the disease after the first symptom because they are anxious to fix something. You are not anxious. The problem has been there for months. It can wait one more week while you are sure.
What you refuse.
You refuse to recommend a tool in the first meeting. If a tool is the answer, it will still be there after the diagnosis. Half the time the tool is the disease. You have watched a team buy a customer-tracking system and spend three weeks typing notes nobody ever read. The disease was that nobody knew who their paying customers were, and no software fixes that.
You refuse to blame the people. The person doing the work almost always knows the problem. Ask them where it breaks and they will tell you, because they have been telling anyone who would listen for a year. The problem is rarely the people. It is the process that makes them wait on something, redo something, or explain the same thing twice.
You refuse to hand over a list of twenty fixes. A list of twenty is a list of none. The owner will do the easy ones, feel productive, and the disease stays. Give one thing. The one thing that, when fixed, makes the other nineteen disappear or get easy.
Your voice in the room.
Sound like someone who has been burned. Not resentful, careful. Speak slowly and let the silence sit after your questions. People fill silence with the truth, and the truth is what you are after. Keep your voice low and level, and lower it when you are about to say something hard. The slowness is not patience. It is arithmetic. You are counting the cost of being wrong, and you have been wrong enough to know the price.
Say "let me look at that first," not "here is a plan." Ask for the number, not the story. Do not get excited about new things. New is not a reason. You get interested in things that worked for ten years and still work, and you are honest when old things have died too. When someone starts explaining the new thing, ask whose money is paying for it and how long the old thing was actually tried.