Process Improvement: Map, Measure, Fix

Somewhere in your business there is a sheet of paper, an email or a sticky note that takes eleven days to travel eleven feet. Nobody is lazy. Nobody did anything wrong. The note simply waits in an inbox, then in a tray, then on a desk while someone is on vacation, and when it finally arrives, three people ask what it is.

That is the normal state of most processes: busy people, slow work. Process improvement is the art of finding out why, without a six-month study and without a consultant who arrives with a deck. You can do the first round yourself, with a whiteboard, a pack of sticky notes, a stopwatch and the willingness to look at your own business the way a visitor would.

This guide shows a simple method in three moves: map, measure, fix. It works for a clinic, a restaurant, a contractor or a three-person startup. I am an operations specialist, so I will say it plainly: this is the part of the business where I spend most of my time, and where small changes pay back fastest.

First, what a process actually is

A process is the sequence of steps that turns a request into a result. A customer asks, something happens, the customer gets what they asked for. Every business has dozens, and most of them were never designed. They grew, one workaround at a time, like a garden nobody planned.

Some examples you already run, whether you call them that or not:

• A new inquiry arrives and someone answers it.
• A quote is requested, prepared, sent and followed up.
• A job is sold, handed to the team, delivered and invoiced.
• A new employee is hired, equipped and trained.
• A complaint is received, solved and learned from.

Notice that the first two are what people call "marketing" and "sales." In my view they are operations processes with a different name. A lead that waits three days for an answer is not a marketing problem. It is a queue. That idea is the core of Project Focus, and it is why this method works on any part of the business.

Step 1: choose one process

The most common mistake is trying to fix everything. Pick one. Good candidates have three traits:

1. It repeats. It happens at least a few times a month, so you can measure it.
2. It touches the customer or the money. A slow quote or a late invoice costs real cash.
3. Someone complains about it. The people who do the work usually know exactly where it hurts. Ask them.

Write the process as "from X to Y." From "request for a quote arrives" to "customer says yes or no." From "job is signed" to "crew shows up." A process with a clear start and a clear end is much easier to improve than one that is "sales" or "operations."

Then name one owner. Not a committee. One person whose Monday is a little worse if the process is slow. That person is also the one who will hold the gain later, which is the most skipped step in the whole method.

Step 2: map it the way it really happens

Do not map the process in the handbook. Map the one that happens on a bad Tuesday. The difference between the two is where the money leaks.

Here is the method I use, and it takes about an hour with the people who do the work:

1. Write each step on a sticky note, starting with a verb: "receive," "check," "send," "approve." One step per note.
2. Put them on the wall from left to right in the order they actually happen.
3. Add who does each step with a mark or color. Count the handoffs, which is the number of times the work changes hands.
4. Add the waits. Between steps, add a different color of note wherever the work sits in an inbox, a tray or a person's head.
5. Add the loops. Any place where the work goes back for a fix is rework, and it deserves a note of its own.
6. Walk the map with a real example. Take one real request from last month and follow it from the first step to the last, using the actual dates in the email or the system.

The last step is the most useful and the most humbling. Almost every time, the map on the wall has fewer steps than the real thing, and the real thing has a mysterious step that exists because of one customer, three years ago.

If you want to keep the map afterwards, a simple list is enough. For a long or complicated process, borrow the swim lane idea: one row per person or team. The RACI chart guide helps if the question "who actually does this?" starts an argument, and it usually does.

Step 3: measure with three numbers

Now you need numbers, and you need very few. Three are enough for a first round:

• Lead time: the elapsed time from the start to the end of the process, for one item. Use calendar days or hours, and say which.
• Hands-on time: how much of that time someone is actually working on it. Ask the people who do each step for an honest estimate, or time a few items.
• Rework rate: the share of items that go back for a fix. Count them over a month.

Add a fourth if you can: the queue, the number of items waiting right now. A queue of 40 requests is a more honest picture than any average.

Write a one-line definition for each number and keep it the same from week to week. "Lead time" can start when the customer sends the message or when someone opens it, and those are not the same number. The habit of defining numbers is the subject of Project Focus, Part 3, and it is the difference between a measurement and an argument. If you want a few more ideas on which numbers earn a place on the page, see dashboards without vanity metrics.

Collect at least ten real items if you can. Five is thin, and a single memorable disaster will mislead you. Do not worry about precision. You are looking for the shape, not the third decimal.

The villain is waiting, not working

Here is the finding that surprises almost everyone. When you compare hands-on time with lead time, the hands-on part is often a small slice. The work takes two hours; the item takes two weeks. The rest is waiting: for an approval, for a missing piece of information, for someone to come back, for the batch to fill up.

Divide hands-on time by lead time and you have a rough "process efficiency." If an item takes 2 hours of real work and 40 hours of elapsed time, the efficiency is 5 percent, and 95 percent of the journey is waiting. I will not tell you what number is normal for your industry, because I do not have a trustworthy benchmark and I distrust anyone who quotes one without a source. Your own number, tracked over time, is the one that matters.

This changes how you improve. If you believe the problem is that people work too slowly, you push them harder. If you see that 95 percent of the time is waiting, you stop pushing people and start removing queues, handoffs and approvals. The first approach makes everybody tired. The second makes the customer happy.

The calculator at the end of this guide does the division for you, and shows how many hours your customers spend waiting each month.

Step 4: find the one step that hurts

With a map and three numbers you can usually spot the culprit in minutes. Look for four patterns:

1. The longest wait. Which gap between steps is the biggest? That is where the days go.
2. The bottleneck. The step with the least capacity decides how much the whole process can deliver, no matter how fast the other steps are. This is the core of the theory of constraints, and the capacity planning guide shows how to find it with a few numbers.
3. The rework loop. A step that sends work back is expensive twice: once to do it wrong, once to do it again.
4. The handoff that loses information. If the person receiving the work always has to ask a question the sender could have answered, the handoff is missing a field.

Pick one. Only one. The temptation here is to fix five things at once, which makes it impossible to know what worked, and also impossible to finish. A process improvement project is still a project, so give it a scope. The project charter guide takes ten minutes to read and saves ten meetings.

Step 5: fix in the right order

When you know the step that hurts, resist the urge to buy software. Work down this ladder, and stop at the first rung that solves the problem.

1. Eliminate. Can you stop doing this step altogether? Many steps exist to check something that no longer needs checking, or to produce a report nobody reads. Ask "what would happen if we just stopped?" and try it for two weeks.
2. Simplify. Can the step be shorter? Fewer fields, a shorter form, a standard answer to the most common question.
3. Combine. Can two steps happen at once, or can one person do both? Every handoff removed is a wait removed.
4. Reorder. Can the slow step start earlier, or run in parallel with something else?
5. Standardize. Write the steps and the checks on one page, so the result does not depend on who is on duty.
6. Automate. Only now, and only the steps that are stable and repeated. A template, a reminder, a form that pushes data to the right place. See business process automation: twelve examples that pay back and the seven workflows to put on autopilot.

The order matters. If you automate first, you freeze a bad process in software and make it harder to change. An old saying in operations is that automating a mess gives you an automated mess. It is not elegant, but it is true.

The smallest test that could work

Before you change the process for everyone, try the fix on a small slice for two weeks: one team, one product, or the items that arrive on Mondays. Measure the same three numbers before and after. If lead time drops and rework does not rise, extend it. If not, you learned something cheaply. This is the spirit of kaizen: small changes, tested quickly, kept when they work.

A worked example (illustrative)

The numbers below are invented to show the method. They are not from a client.

A small design studio receives requests for quotes by email. The owner maps the process on a wall: the request arrives, the owner reads it, asks the designer for an estimate, writes the quote, sends it, and follows up. Seven steps and four handoffs.

She follows ten real requests from the past month. The average lead time from request to quote sent is 6 days. The hands-on time, adding up reading, estimating and writing, is about 2.5 hours. The efficiency is under 5 percent. Two of the ten quotes needed a correction after sending, so the rework rate is 20 percent.

The longest wait turns out to be the gap between "asks the designer" and "gets the estimate": 3 days on average, because the designer is deep in client work and answers questions in batches on Fridays. The bottleneck is not the designer's skill. It is that every quote waits for one person's attention.

The fix does not need software. The studio writes a one-page price list for its five most common jobs, so those quotes need no estimate at all. For unusual jobs, the owner and the designer agree on a 30-minute slot every day at 11:00 for estimates. After two weeks, the average lead time on the tested items falls from 6 days to 2. The rework rate does not change, so she adds a short check before sending and measures again. It is not a miracle. It is a map, three numbers and one box circled.

The hall of fame of process mistakes

I have seen all of these, and I have made a couple myself.

• The heroic fix. One person stays late and makes it work. It does not scale and it ends in resignation.
• The tool first. "We need a new system." Perhaps, but not before you know what the process is.
• Mapping the ideal. The wall shows what the handbook says, and nothing changes.
• Measuring too much. Fifteen numbers on a dashboard that nobody reads.
• No owner. A process without an owner is a process that gets exactly the attention it already gets.
• Blaming people. The same people in a better process do better work. Very often the fix is a wait, a form or a missing field.
• Skipping the follow-up. The first month looks great. Six months later the old habits are back. A fix needs a weekly glance to stay fixed.

Step 6: hold the gain

The last move is what separates a clever afternoon from an improvement. Do four things:

1. Write the new steps on one page, with the checks. Keep it short enough that people will read it.
2. Keep one number on the weekly page. Usually lead time or the queue. Put it next to your other numbers in the weekly review.
3. Name the owner and tell the team. The owner decides when the process needs another look.
4. Write down what you learned. Five questions are enough; the lessons learned template has them.

Then, three months later, repeat the exercise on the next process. Improvement is a habit, not an event. Small teams that do this once a quarter become, over a year, a different company.

A 30-day plan

If you want a schedule, here is a simple one.

• Week 1: choose the process and its owner. Write "from X to Y." Gather the people who do the work.
• Week 2: map the process on the wall, follow ten real items and collect the three numbers.
• Week 3: find the one step that hurts. Try the smallest fix on a small slice.
• Week 4: compare the numbers, extend or drop the fix, write the new steps and put one number on the weekly page.

At the end of the month you will have an improved process, a habit and a little proof. If you want help choosing the process or running the first round, that is the kind of work I do in operations consulting, and it is easier to start with a short, paid diagnostic than a long engagement, as described in how to hire a business consultant.

Try the numbers

The tool below turns your three numbers into one picture. Enter the hands-on time and the elapsed time of one item, how many times the work changes hands and how many items you handle each month. It shows how much of the journey is waiting, how many hours your customers wait each month, and what a 30 percent cut in waiting would do. The sample numbers are invented. Nothing you type leaves your browser.

Interactive tool

How much of the journey is waiting?

Loading the tool…

What to do this week

Choose one process that repeats, touches the customer or the money, and has someone who complains about it. Write "from X to Y," name an owner and book one hour with the people who do the work. Bring a pack of sticky notes. That is the whole first step.

If you would like to see where your business leaks the most, try the two-minute diagnostic, look at the free templates or read about my services and project management consulting services. When you are ready to talk, tell me about your business. In the new series Process teardown I take one common process at a time and map it step by step.

What is the process in your business that you would least like to map on a wall? Tell me in the comments.

General information, not legal, financial or tax advice. Examples are illustrative and the numbers are invented.

Frequently asked questions

What is process improvement in simple terms?

It is the habit of looking at how work actually moves through your business, measuring it with a few honest numbers and fixing the step that hurts most. The usual sequence is to map the process, measure lead time, hands-on time and rework, then remove, simplify or combine steps before you consider automation.

How do I find the bottleneck in a process?

Map the steps and find the one with the least capacity, the longest queue in front of it or the longest wait before it. The bottleneck decides how much the whole process can deliver. Speeding up a step that is not the bottleneck changes nothing the customer can see.

Should I automate first or fix the process first?

Fix the process first. Remove steps that are not needed, simplify the rest and write the steps down. Then automate the stable, repeated parts. Automating a confused process freezes the confusion into software and makes it harder to change.

How many processes should I improve at once?

One. Start with a process that repeats, touches the customer or the money and has someone who complains about it. Give it an owner, three numbers and a 30-day window. When the gain holds, choose the next one.

Join the conversation

Did this help? A quick reaction tells me what to write next.

Useful Made me smile Go deeper on this

Be the first to react.

Comments

Loading comments…

Leave a comment

Name (optional) What is the process in your business that you would least like to map on a wall? Post comment

No email and no sign-up. Every comment is reviewed before it appears. Please keep it respectful and leave links out.