A request is approved on Monday. The work itself takes two hours. It is delivered on Friday of the following week. Nobody was lazy. The request simply spent most of its life waiting: in an inbox, in a queue, in someone's "I'll get to it" list.
This is the normal state of many companies. The time inside each team is often reasonable; the time between teams is where days disappear. And because nobody owns the gap, nobody measures it.
This article gives a method you can run in a month with a spreadsheet and the people you already have: find the handoffs, measure the waiting, fix the three worst. It does not need a new system.
What a handoff is
A handoff is any moment when work leaves one person or team and waits for another. Sales passes a signed deal to operations. Operations passes an order to purchasing. A developer passes a change to a reviewer. Finance passes a query back to a manager.
Every handoff has three parts: the send (someone says "this is ready"), the wait (nothing happens), and the start (someone actually begins). The wait is the part you will measure.
If your processes already break as they grow, why your workflows break when you scale explains the general pattern. Here we stay narrow: how long does each handoff wait, and why?
Week 1: draw the handoffs on one page
Pick one process that people complain about, for example "from signed order to first delivery". Do not map the whole company.
- Write the steps in a line, from the first trigger to the result the customer sees.
- Mark every place where work moves from one team or person to another. That is a handoff.
- For each one write down: who sends, who receives, and how the work travels (email, chat, ticket, shared file, spoken).
Most processes of this size have between six and twelve handoffs. Do this with the people who do the work, not from a manual. The real route is often different from the documented one. If you find the same information typed into several places on the way, the spreadsheet problem in manufacturing shows how that becomes a hidden cost.
Weeks 1 and 2: measure the wait
For two weeks, record two timestamps for every item that passes through each handoff:
- Sent: when the sender considered it ready.
- Started: when the receiver first worked on it.
The wait is the difference. You do not need software for this. A shared sheet with five columns is enough: item, handoff, sent, started, and a free-text note when the wait was long. If the work already lives in a ticket or CRM, the timestamps may exist; export them instead of typing.
Two rules keep the data honest:
- Count calendar time and working time separately. A handoff on Friday afternoon that starts Monday morning is not a problem; a handoff that waits three working days is.
- Ask people to log the number, not to explain themselves. The goal is to see the process, not to rank the people. If the log is used to blame, it will stop being filled in.
If you have never measured the cost of manual work, measuring the cost of a manual process in one week uses the same light approach.
Week 3 starts: pick the three slowest
When the two weeks are over, calculate two numbers for each handoff:
- The median wait: the typical case.
- The worst wait: the slowest items, for example the slowest one in ten.
Rank the handoffs by the total waiting time they cause, which is the number of items multiplied by the average wait. Take the top three.
Look at the median and the worst case separately, because they point to different causes:
- A high median means the handoff is structurally slow: the receiver is overloaded, the work arrives at the wrong time, or nobody knows it is waiting.
- A low median but a long tail means most items flow, but some get stuck: missing information, an absent person, or an unclear owner for unusual cases.
A worked example (illustrative, not a client case)
The numbers below are invented to show the arithmetic. Your values will differ.
Imagine a company processes 40 orders in the two weeks of measurement. The log shows three handoffs with the longest waiting:
| Handoff | Items | Median wait | Slowest items | Total waiting |
|---|---|---|---|---|
| Sales to operations (order details) | 40 | 1.5 days | 5 days | about 70 days |
| Operations to purchasing | 25 | 2 days | 4 days | about 55 days |
| Purchasing to warehouse | 25 | 0.5 days | 6 days | about 20 days |
Total waiting is rounded from the items and the average wait. The first two handoffs have a high median, so they are structural. The third has a short median and a long tail, so it is about exceptions, not volume.
Now the notes. In the first handoff, most waiting is because details were incomplete and operations had to ask back. In the second, purchasing checks requests once a day in a shared inbox. In the third, long waits happen when the usual person is away and nobody covers.
Weeks 3 and 4: fix the cause, not the symptom
For each of the three, change one thing and keep it small. The usual causes have usual fixes.
The request is incomplete. Replace the free-form email with a short form or template that lists what the receiver needs. Make the sender responsible for completeness. In the example, a one-page order sheet with required fields removes most of the questions back.
The receiver does not see it. A shared inbox that is checked "when possible" is a queue without a clock. Give the handoff one named owner per day, one place where requests arrive, and a visible count of what is waiting. A fixed slot, for example a short review at 10:00 and 15:00, is often better than constant interruptions.
Nobody covers absences. Name a deputy for each handoff and write down what they need to know. This is the cheapest fix in the list and removes the long tail.
The receiver is overloaded. This one is not solved by a form. Either the capacity has to change, or something with lower value has to stop. Say it openly instead of hiding it in a tool.
The same data is retyped. If the information is entered twice, the delay and the errors multiply. Only then is it worth asking whether a small shared tool or an integration pays off. If it does, moving from a spreadsheet to a small internal tool in 30 days is a realistic path, and the measurement you made in this month is your evidence for the decision.
Agree one expected response time for each handoff, such as "started within one working day". A target turns a feeling of slowness into something you can check.
Week 5: measure again and keep what worked
After four weeks of the new way of working, measure again for one week using the same log. Compare the median and the worst case with the first measurement.
- If the wait fell, write the new rule down so that it survives holidays and staff changes.
- If it did not, the cause was probably something else. Return to the notes and look at the longest items.
- If a handoff is now fast, move on to the next three.
You do not need to remove every wait. A handoff that waits half a day because the receiver finishes a task first is healthy. You are looking for the waits that are long, unexplained and repeated.
What to do this week
Choose one process people complain about and draw its handoffs on a single page. Create the five-column log and ask the receivers to fill it in for two weeks. Do not change anything yet. When the numbers come in, you will know which three handoffs deserve a month of attention, and you will have the evidence to show that the change worked.