Pick one process that feels slow. Write down the total time it takes end to end. Now write down how many minutes of actual work happen inside it. For most service businesses the working time is under 15% of the elapsed time. The other 85% is waiting: for a reply, for approval, for someone to notice a message. Automating the work makes almost no difference. Removing one wait does. Which single wait, if you deleted it, would cut the most elapsed time?

Quick one for the weekend. Think about the last two pieces of software you bought for your business. For each one, answer honestly: what did you stop doing after you bought it? Not what it added. What it replaced. If the answer is "nothing, it's just there now," you didn't solve a problem. You added a place where work can go and not come back. This is the most common pattern I see. A tool gets bought to fix something, the old way never actually stops, and now there are two versions of the same process running at once. The spreadsheet is still open. The group chat is still where decisions actually get made. The CRM has the official record and the real record lives somewhere else. Nobody decides to run two systems. It happens because deleting the old one requires a decision nobody makes. So the weekend question: which of your tools didn't replace anything? That's usually the first place to look.

Every owner has some version of this: they bought the CRM, set up the pipeline, trained the team, and within six weeks it was stale. The diagnosis is always "the team isn't disciplined." It almost never is People maintain a system when the system returns something to them. Your CRM currently takes five minutes of typing and returns nothing to the person typing. It returns something to you: visibility. So you're asking someone to do admin for your benefit and calling it a process problem when they stop. Three structural fixes before you touch a setting. 1. Make the record the work, not a report about the work. If the proposal gets sent from the CRM, the CRM stays current. If the proposal gets sent from email and then logged, it won't. Anything that requires a person to describe what they just did somewhere else is a losing design. 2. Cut the required fields to the ones you'd act on. Go through your form and ask, for each field: what decision changes based on this? If nothing, delete it. Most CRMs die of field bloat. 3. Decide who the system is for. If it's for you, own the data entry yourself or automate the capture. Don't distribute the cost and centralize the benefit. Only after those three does the tool question matter. The Field Audit List every required field in your CRM. Next to each one, write the decision that changes based on it. Delete every field with a blank next to it. Most people cut between a third and half. What's left is what people will actually fill in.

When you post a broken process here, this is what you get back. Five fields in, a completed map out. It's also the format to use once you're mapping processes on your own. Everything in it is something I'll ask you about anyway. The header Process name: Owner (one person, not a team): Trigger. What starts it: Done. What has to be true for this to be finished: The steps One block per step. Repeat as many times as you need. Step 1 Who: Tool: Work time: Wait time: H / A / S: The H / A / S tag. Mark every step: H = human, and should stay human A = already automated S = should be automated, once the process is stable The rest Where it breaks. Be specific. Which step, and what actually happens. Exception paths. The rush job, the awkward client, the referral who skips intake. Note roughly how often each happens. Then ask two questions Which step has no owner? Which handoff has the longest wait time? Those two answers are usually the whole diagnosis. Most people expect the problem to be a missing tool and find it's a missing name. Two notes on filling it in Work time and wait time are separate fields for a reason. Work time is someone actually doing something. Wait time is the gap before the next person picks it up. In most service businesses the waiting is 80% of the elapsed time, which is why automating the work barely moves anything. And if a step genuinely has no owner, write nobody. That's the most useful thing you can put on the page.

This week, do one thing: every time someone interrupts you to ask how something works, or to get a decision unstuck, write down the question. Don't change anything. Just log it. By Friday you'll have a list. Read it back and mark each one: D — a decision only you can make I — information that exists somewhere but they couldn't find P — no process exists, so they came to you The I's and P's are your bottleneck. Not the D's. Owners usually try to delegate decisions and hold on to the information it's the wrong way round. Post your count on Thursday. Most people are surprised by the total.