Walk one recent job through one process with the people who did the work, and the hours show up on the map: in handoffs, in information typed twice, in jobs waiting for someone to notice them and in checks that exist because an earlier step cannot be trusted. The five steps below take three and a half hours. With a break, plan on an afternoon of about four.
A good first process runs every week and has a clear start and end
Pick one process, not a department. Good candidates run at least weekly and have an obvious trigger and an obvious finish: client onboarding, quote to order, or job completion to invoice. If you are unsure which one to choose, ask the team which process they would fix first, and map that one. They live with it every day.
Bring three things into the room:
- The people who do the work. Two or three of them, not only their manager.
- One real job from the last two weeks. Recent enough that people remember what happened, and specific enough that nobody can fall back on the typical case.
- A whiteboard or a shared online canvas, with notes or markers in four colors, one for each kind of waste in Step 3.
Walking one real job shows the work as it is done
A map drawn from memory shows how the work should flow. A map of one real job shows how it does flow, including the workarounds that never made it into the manual.
Step 1: Agree where the process starts and ends (15 minutes)
Write down the event that starts the process, such as “client signs the proposal”, and the event that ends it, such as “first invoice paid”. If the people in the room disagree about either one, write the disagreement down. It is your first finding: a process without an agreed end has nobody who owns all of it.
Step 2: Walk the job from trigger to end (90 minutes)
Use the specific job, not the typical case. Ask “what happened next?” until you reach the end. Write each step as a verb and an object: “create client folder”, “send welcome email”, “copy price into invoice”. Next to each step, note who did it, which tool they used and how long the job sat before the step began.
When someone says “normally” or “it depends”, ask what happened on this job. Exceptions are where the workarounds live.
Step 3: Mark the four kinds of waste (45 minutes)
Use one color for each:
- Handoffs. The work passes to another person or team.
- Re-entry. Someone types information that already exists somewhere else.
- Waits. The job sits until someone notices it, signs it or replies.
- Checks. Someone reviews the work because the previous step cannot be trusted.
A step can carry more than one mark. A contract that goes to the owner for a signature is both a handoff and a wait.
The count shows which two or three fixes are worth pricing
Counting turns the colored marks into hours, and the hours rank the fixes.
Step 4: Count the marks and estimate the hours (30 minutes)
Count each kind of mark. Then estimate two kinds of time:
- Staff time for re-entry and checks: the minutes per job, multiplied by the jobs per month.
- Elapsed time for waits: the working hours the job sat still.
An estimate within 50% is good enough. The aim is to see which steps hold most of the hours, not to produce an exact figure. Keep the two kinds of time apart. Staff time shows up in payroll. Elapsed time shows up as a client waiting longer for kickoff and an invoice that goes out later.
Step 5: Sort each marked step with three questions (30 minutes)
Ask the questions in this order:
- Should this step exist, and does it need the person who does it now? If not, remove it or move it before you think about tools.
- Is the problem how people work, or what the tool can do? If two people do the step differently, it is a process problem. If they agree but the tool cannot record it, it is a tool problem.
- Would connecting two tools remove the step? If so, note which tool should hold the information and which should read it. Where several tools hold the same information, Run a keep, connect or replace review settles which one is the record.
Then pick the two or three changes with the most hours behind them. Those are the fixes worth pricing. A fix pays for itself when the hours it removes in a year are worth more than it costs to build and run.
Illustration: two waits hold 36 of the 46 hours before kickoff
An agency maps client onboarding, from signed proposal to kickoff call, using a client who signed nine days earlier. The map has 14 steps and 5 handoffs, and the client’s details are typed into 4 tools. The job took 46 working hours from signature to kickoff. Typing, checking and the other steps took 6 of them.
Two waits hold most of the time. The contract sits for 22 working hours until the owner countersigns it, and the delivery lead then waits 14 more to hear that the deal has closed. Sorting the marked steps with the three questions in Step 5 produces three fixes.
| Fix | Kind of fix | Hours removed |
|---|---|---|
| Account lead signs standard contracts | Process | 20 |
| Notice to delivery lead on signature | Connection | 13 |
| One client record for all 4 tools | Connection | 3 |
Source: Verticulate illustration
Note: The first two fixes remove elapsed time. The third removes staff time spent typing and checking.
The largest fix comes from the first question. The owner signs every contract, including those on standard terms and list prices. Letting the account lead sign those, and sending only exceptions to the owner, removes most of the wait without any software. The second fix is a notice from the e-signature tool to the delivery lead the moment a contract is signed. The third gives the client one record that the other three tools read from, which ends the retyping and the check that follows it. Together, the three would bring the time from signature to kickoff down from 46 working hours to about 10.
Four mistakes send the map to the wrong fix
Each of them produces a map that looks complete and points at the wrong fix.
Mistake 1: Mapping with managers alone
Managers know how the process should work. The people who do it know how it does, including the workarounds nobody mentioned to the manager.
Mistake 2: Timing the work and ignoring the waits
Timing each task feels like the rigorous part. A job can spend far longer waiting between tasks than being worked on, as in the illustration, and a map that times only the tasks misses that time. Mark every wait and estimate it, even when nobody is idle during it.
Mistake 3: Fixing as you go
Finish the map first. A fix made halfway through tends to move the problem to the next step, because nobody has yet seen the whole chain.
Mistake 4: Buying software the same week
Share the map with the team, let it settle for a week, and decide afterwards. The largest fix in the illustration needed no software at all, and that only became clear once the map was complete.