Two similar risks get very different answers, depending on who looked at them.
Organizational Noise Audit
You may not have a people problem. You may have a noise problem.
A company makes thousands of judgment calls every day. The question is not whether every call is perfect. The question is whether the same kind of situation keeps getting different answers for no good reason.
Noise vs bias
Noise is not bias
Bias is everyone leaning the same wrong way. Noise is everyone leaning a different way, at random. The difference matters because they have different causes and different fixes.
I use this as one tool among several, not as the answer to everything. Not every problem in a company is noise. Saying it is would be its own kind of mistake.
Based on the book Noise: A Flaw in Human Judgment by Daniel Kahneman, Olivier Sibony, and Cass Sunstein.
What it looks like at work
Two managers read the same rule in two ways that cannot both be right.
A project gets an easy pass or a hard time, depending on who presents it.
The answer changes based on who happens to be in the room.
The first answer said out loud in a meeting becomes the answer.
People who all follow the “standard” way are each doing it differently.
Three kinds of noise
Not all disagreement is the same. Each kind has its own fix.
When a group disagrees with itself, the useful question is what kind of disagreement it is. The audit splits it into three, because each one points to a different fix.
Between teams
Whole teams grade harder or easier than other teams. (The book calls this “level noise.”)
Sales says a risk is low. Legal says the same risk is high. The fix: give everyone the same scale, with real examples, so a “7” means the same thing in every team.
Person to person
One person’s own habits. Tough on some things, easy on others. (The book calls this “pattern noise.”)
A manager who is strict about money but easy on people. The fix: break the decision into parts and score each part on its own. Then average several people’s answers, made separately.
Day to day
The same person answers differently on a different day. Mood, timing, and what came just before all play a part. (The book calls this “occasion noise.”)
A decision made tired, rushed, or late at night comes out different from the same decision made fresh. The fix: pick a good time, control what people see first, and get everyone’s answer before the group talks.
Splitting noise into these three tells you which fix to reach for: line up the teams’ standards, rein in personal habits, or clean up how and when decisions get made. The audit reports all three, for the whole company and for each department and level, so the fix matches the real problem.
What we look at
Eight areas of how the company works
We have about 120 to 150 questions in all. Each person sees only 25 to 40 of them, picked to fit their job.
How decisions get made
Did people have enough facts? Did the group agree too fast? Were other options looked at? Do decisions get reversed later?
Who gets to speak
How much a job title changes the answer. What it costs to disagree. Who talks first. Whether the most confident person wins.
How information moves
What gets lost along the way. Questions that get asked over and over. Sources that contradict each other. Things only one person knows.
Who owns what
Whether each task has a clear owner and a clear next step. Work getting done twice. Workarounds. Who has to approve what.
Meetings
Whether people know why they are there. Whether the meeting was worth it. Whether it ended with a real decision or another meeting.
Tools and systems
The same data typed in twice. Numbers that have to be fixed by hand. Too many alerts. Side systems people built because the real one does not work.
Whether work matches the goal
Tasks that do not connect to a purpose. Numbers that get gamed. Teams doing what is best for the team, not the company.
The conditions people work in
Interruptions, time pressure, and workload. We record these as the conditions people were working under, never as a judgment about them.
Measured at every level
The power is in the comparison
One person’s answers tell you almost nothing. The gap between what the top leaders believe, what managers deal with, and what the front line reports tells you a lot. Showing people that gap usually ends an argument that has been going on for a year.
Top leaders
The big decisions. How good their information is. Whether they trust the numbers. Whether departments are pulling the same way. What they assume about how work really gets done.
Managers
Turning the plan into work. Priorities that fight each other. Who has to approve what. How much say they have. Workload. Hand-offs.
Employees
How the work really gets done. Workarounds. Interruptions. Tools that get in the way. Whether meetings are worth it. Work done twice. Whether it is safe to disagree.
Customers
How easy it is to get help. Whether they get the same answer twice. Having to repeat themselves. Delays. Mixed messages. Whether anyone seems to own their problem.
Across departments
A few people from each department, so we can compare how the same process looks from different sides.
Sample results
One big number, always broken into parts
See a full sample report — the three kinds of noise, by department →
Noise score58Sample numbersHow much people disagree · 0 to 100 · lower is better
How to read these numbers
The big number is the noise score. It runs from 0 to 100 and shows how much people in the same company disagree when they describe the same thing. Lower is better.
- 0 to 34: people mostly agree. They see things the same way.
- 35 to 54: some disagreement.
- 55 to 100: a lot of disagreement. Different people describe the same work very differently.
In this example, 58 is high. It is just past the line where disagreement becomes a problem. That is a warning sign, not a disaster.
The bars work the other way.Each bar scores one area from 0 to 100, and higher is healthier: 75 and up is strong, 60 to 74 solid, 45 to 59 mixed, 30 to 44 strained, and under 30 weak. Here the two lowest areas are “Meetings are worth it” (36) and “Teams pull together” (43). Both are strained. The strongest area is “Same question, same answer” (72).
These are sample numbers. A real audit also shows how sure we are about each result, and it never reports on a group of fewer than 5 people.
Area scores · 0 to 100 · higher is better
Same question, same answer
72
Details survive hand-offs
61
Teams pull together
43
Someone owns each decision
49
Titles do not sway the answer
68
Meetings are worth it
36
People hear how it went
55
Everyone knows the next step
63
- Solid (60 to 74): 4 of 8Same question, same answer, Details survive hand-offs, Titles do not sway the answer, Everyone knows the next step
- Mixed (45 to 59): 2 of 8Someone owns each decision, People hear how it went
- Strained (30 to 44): 2 of 8Teams pull together, Meetings are worth it
What this is, and what it is not
The scores describe how the company works and how much people agree. They do not measure how smart, skilled, healthy, or productive any employee is. This is a tool for understanding a company. It is not a personality test, a mental health test, a way to watch employees, or a way to rank people. Results are only ever shown for groups, and we never show a group so small that someone could be picked out.
Still being built
Ask about an audit
I am building the audit and the reports right now. The first companies to do this work directly with me, and what we learn together shapes the product.