People often ask what tools I work with. The interesting answer is not the list of tools: it is which seven numbers are in front of my eyes, and in what order I read them.
Seven numbers, not a dashboard
A dashboard with thirty metrics is a dashboard you never look at. I have tried. It works for two weeks and then becomes wallpaper.
Mine are seven, and they answer three questions: are we late, are we inefficient, or are we simply stuck.
- Deliveries on time, as a share of the commitment made. Not of the sprint: of the commitment.
- Cycle time, median, from request to delivery. Median, not mean, because one enormous project drags the mean.
- Change failure rate, the share of deliveries that come back. Above fifteen percent, the flow is the problem, not the people.
- Time to restore, in hours. How long we take when something breaks.
- Hours of waiting per person, not hours worked. This is the metric that has changed the most decisions in my career.
- Processes with a single approver. A rising number is a risk nobody is watching.
- Decisions open more than a week. My favourite metric: if it rises, today you made one fewer decision. That happens more often than it looks.
The order I read them in
Monday morning I look at two numbers only: deliveries on time and time to restore. Ten minutes.
If deliveries are fine and time to restore is climbing, nothing else interests me: that is the quarter. Every other number is noise next to that one.
Then I look at open decisions. It is the only number I look at daily, and it is not a KPI: it is a list. A number up by one means you made one fewer decision today.
The three questions I ask every Monday
After the numbers, three questions, always in this order.
What happened last week that was not planned? Not "what went wrong": what happened. A supply problem, someone ill, a supplier. Almost everything that costs you a quarter is something unforeseen, and it is in no report.
How many times did we wait for someone? Not worked. Waited. If last week you spent more time waiting than working, your problem is not productivity.
What would you do differently starting over? I use that one to work out what I am tolerating out of habit.
Why seven and not five
I have tried five, and ten. Five was too few: there was always a number that explained the anomaly, and I ended up asking someone for data every Monday.
Ten was too many: I started interpreting them in order of importance instead of order of urgency, and by the end of the month I no longer knew which of the two I had used to decide.
Seven is the number where I can hold all of it in my head and ask nobody anything. If it holds for four straight weeks without you ever needing to look up a number that is not there, you have found yours.
One honest note: these numbers work for a company that delivers something. If your output has no delivery date, you cannot measure cycle time, and the problem is upstream of you.