Read Delivery Data Without Averages:Study the Slow Tail
How to read delivery data without averages: sort finished work, circle the slowest few and ask what each waited on. Twenty minutes a month shows your real drag.
To read delivery data without averages, look at the spread: most work finishes in a few evenings, some takes weeks, and that slow tail breaks your plans. Ignore the mean, list the slowest few items each month and ask what happened to each.
How to read delivery data without averages
Quick test. Your average job takes about a week. Can you promise a customer Friday?
No. You can't. Because the average doesn't know that one job in a handful drags on for three weeks, and you don't know which one it is until it's late.
That's the trouble with averages. They answer a question nobody asked. You never plan around "typically". You plan around "what's the worst I should expect, and can I live with it?"
So stop looking at the mean. Look at the shape. Where do most items cluster? How far out does the slow end reach? Which items are out there, and why?
That is the whole method to read delivery data. Everything below is how to do it with a notebook and twenty minutes.
What an average hides
Picture a month of finished work, each item with the time it took from actually starting to genuinely done. Sort them from quickest to slowest.
Most sit in a bunch near the left. A few trail off to the right. The average lands somewhere in the middle of nothing: slower than most of your work, quicker than the stragglers, and a perfectly accurate description of no item you actually did.
The stragglers are where your plans die. They're also where your problems live. A job that took five times longer than the rest didn't do that by accident. It was blocked, or far too big, or started and abandoned, or it needed something you didn't have.
The average buries every one of those stories in a single tidy figure.
| What you look at | What it tells you | What it hides |
|---|---|---|
| The average | A single comforting figure | Every slow item, which is exactly what broke your plans |
| The bunch near the left | What "normal" looks like for you | Nothing much. It's the boring, healthy part |
| The tail on the right | How bad the bad case gets | Nothing. It's the useful part, if you open each item |
| The gap between them | How predictable you really are | Not much, as long as you look at it as a shape |
When you read delivery data, a tight bunch with a mediocre middle is worth far more than a good average with a long tail. You can plan around the first. You can only hope around the second.
Free guide · 21 pagesMeasure the system, not yourself.The Delivery Metrics Playbook: is the machine healthy, without the stick.Get the free guide →Do this once a month, in twenty minutes
You don't need software. A list will do.
- List what finished. For each item, write the time from real start to real done. Use the endpoints you defined and wrote down. If you haven't, do that first.
- Sort it. Quickest to slowest. Look at the shape before you think about any figure.
- Circle the slowest few. Not all of them. The few that made you wince.
- Ask what happened to each. One sentence per item. Not "it was hard". What was it waiting for?
That fourth step is the one that matters, and the one that people skip because it feels less like analysis and more like detective work. It is detective work. That's why it works.
Each outlier is individually investigable. You can usually remember the story. Ten minutes on the slowest few beats an afternoon on aggregate charts, every time.
Read delivery data for the three usual suspects
Before you reach for anything clever, check these three. When flow is poor, the cause is nearly always one of them.
Too much in progress
Everything started, nothing finished. Every session begins with reloading what you were doing. If your working time is ninety minutes after dinner, that reload eats a visible slice of it. The item doesn't get slower because it's hard. It gets slower because it keeps going cold.
Stalled on a decision
The item is waiting on something you haven't decided. This one is sneaky when you work alone, because nobody is chasing you. A pending decision can sit for weeks and look exactly like work in progress.
Ask of each slow item: was I waiting on myself?
Too large
A piece of work that can't be finished in the sessions you actually have will sit half-done and rot. Size it against your real evenings, not your ideal ones.
All three are fixable by how you work. None of them is fixed by working more. If that sounds familiar, you may be mislabelling a flow problem as a capacity one, and how to tell a capacity problem from a flow problem shows how to separate them.
The idle time nobody sees
Here is a question worth asking of any slow item: how much of its life was spent not being worked on?
People who measure this for the first time are frequently startled. Most of the elapsed time wasn't building. It was waiting. On a decision you kept putting off. On something external. On an evening that didn't come.
In ordinary accounts of your work, that time is invisible. You were busy with something else. Nothing looked stuck. The calendar just says "it took three weeks".
The tail makes it visible. And once it is visible, it's usually the largest and cheapest thing to fix. You don't need more hours. You need fewer things waiting.
If the biggest lever turns out to be how many items you have open at once, how to limit work in progress is where to go next.
What it looks like in practice
Say you sell a productised service and you've finished a handful of jobs this month. Most went out in a few evenings. One took the best part of a month.
The average says you're slightly slow. Useless. So you open the slow one. It sat for a fortnight waiting on a pricing decision you kept meaning to make. Nothing was wrong with the work. You were the blockage.
Now you know the fix, and it isn't "work harder". It's "decide faster, or stop starting things that hinge on a decision you haven't made". That is a change to how you work. It costs nothing. And no average on earth would have shown it to you.
Do this three months running and a pattern appears. The same cause turns up again and again. That repeat is the real finding, and it's the thing worth fixing.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Sanity-check before you act
Data that doesn't match your month is a bug report about your measuring.
If the tail shows nothing slow and you remember a wretched fortnight, something isn't getting recorded. Perhaps work is happening outside the list. Perhaps a status means one thing in March and another in April. Fix that first. Don't conclude the month was fine, and don't conclude your memory is faulty.
The other direction matters too. If a number shocks you, check it before you react. One badly recorded item can fake a whole tail.
Then change one thing
You've found the cause. Resist the urge to fix everything.
Pick one change to how you work. Decide which number should move and roughly when. Then leave it alone long enough for the signal to respond, and look again next month.
Four changes at once teach you nothing. You'll improve, you won't know why, and you'll be stuck with habits you can't justify. One change and a check, repeated through a year, builds a real understanding of how you actually work. It also lets you drop things that didn't help, which is rarer and more valuable than adding new ones.
Free guide · 21 pagesMeasure the system, not yourself.The Delivery Metrics Playbook: is the machine healthy, without the stick.Get the free guide →What not to conclude from a bad tail
A long tail isn't a verdict on you. It's a description of a process. It points at work that stalled, not at a person who failed.
If a month's tail makes you feel worse about yourself, the number has stopped being useful. Step away until you can look at it as a fact about the work again. This is the single biggest way people lose the instrument, and it's laid out plainly in how to measure solo productivity without guilt.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Frequently asked questions
Why are averages bad for planning?
Because they describe a typical item you rarely have. What breaks plans is the slow tail, and the average hides it. Planning needs the worst case you should reasonably expect, which lives in the spread.
How many items do I need to read delivery data this way?
Fewer than you think. A month of finished work is enough to see a bunch and a tail. If you finish very little, look at what is stuck instead, because that is the tail in disguise.
What do I do with the slowest items?
Open each one and ask what happened. It was blocked on a decision, too big, started and abandoned, or missing something you needed. Each answer points at a change in how you work.
Does this replace tracking cycle time?
No. You still measure cycle time with clear start and end points. This is how you read delivery data: as a shape with a story in the tail, not as one number.
The full method, with the checklists, is in The Delivery Metrics Playbook, a free 21-page playbook.
How trapped are you, really?
Take the Corporate Suffocation Index, the free solopreneur readiness assessment.