In Brief
- The same problem coming back means you’re treating a symptom, not the cause.
- Deming distinguished between common-cause variation (in the system) and special-cause variation (signals).
- Single-loop fixes follow the existing rules; double-loop questions whether the rules are right.
- The 5 Whys and the fishbone diagram are simple tools for finding the real cause.
- Most “fixes” fail because they were applied before the actual problem was found.
There’s a quiet pattern in most small businesses that runs the founder ragged. The same problems keep coming back.
A different staff member missed a deadline. A different client complained about the same kind of thing. A different process broke in the same way. The names change. The pattern doesn’t.
If you’ve been fixing the same problem in different costumes for two years, you’re not solving the problem. You’re solving symptoms of a problem you haven’t found.
This post is about how to find the actual one.
The misdiagnosis loop
Most businesses fix problems the way a tired parent settles a sibling fight. Pull them apart. Issue a rule. Move on. The fight comes back the next afternoon, dressed slightly differently.
At work, the equivalent is fixing the visible thing. The deadline got missed — tell the person to plan better. The client complained — apologise and offer a discount. The process broke — add a check. Each of these treats the surface. None of them touch what produced it.
This isn’t laziness. It’s the path of least resistance. Surface fixes are fast, feel decisive, and don’t require asking uncomfortable questions about the system. The problem is they don’t work. They just delay the next version of the same thing.
The founders I interviewed in my PhD research who built lasting businesses didn’t do this. They asked harder questions, more often, about smaller things. That’s the practice. Not the tool — the practice.
Deming’s two types of causes
W. Edwards Deming spent decades watching managers attack the wrong problems. As part of what he called Profound Knowledge, he divided what was producing those problems into two categories.
Common cause variation. This is the everyday noise built into the system. Some days are busier than others. Some clients are harder than others. Some hours produce more output than others. None of these are special events — they’re how the system behaves on a normal day. Common cause variation is fixed by changing the system, not by reacting to any individual instance.
Special cause variation. This is something genuinely outside the system. A specific event that’s signalling something new — a major supplier failing, a new client behaviour, a key person leaving. Special cause variation is a real signal. It deserves a specific response.
The mistake most managers make is treating common cause variation as if it were special. Someone missed one deadline — they get pulled aside. Variation in the system gets misread as a personal failing. The lecture changes nothing because the system that produces missed deadlines hasn’t moved. The next deadline gets missed by someone else.
Deming called this “tampering.” Reacting to common variation as if it were a signal makes the system worse, not better. You’re now adding new rules and consequences to a problem that wasn’t actually a problem — it was just how the system behaves.
The other mistake is the opposite. Genuinely new signals get absorbed into “that’s just how it is.” A client complains about something new and unusual. You write it off as a one-off. Six months later, you realise it was the early warning of a pattern you should have caught.
The skill is telling these apart. The discipline is asking, every time, which one you’re looking at.
Single-loop vs double-loop
Chris Argyris described a related pattern from a different angle. He noticed that most leaders, faced with a problem, ran a fast mental loop. Spot the error. Apply an existing rule. Move on.
He called this single-loop learning. It works fine when the rules are right. It fails when the rules themselves are the problem.
Double-loop learning is the harder move. It asks not just “what should I do about this” but “should I be doing this at all? Are the rules I’m using producing this problem?” Most leaders avoid double-loop learning because it’s uncomfortable. It questions decisions you made. It might mean undoing something. It makes you look — to yourself, mostly — like you might have been wrong.
But every recurring problem in your business has a double-loop question behind it.
If staff keep missing deadlines, the single-loop fix is “improve planning.” The double-loop question is “is the way we’re scoping these projects realistic?”
If clients keep complaining about the same kind of issue, the single-loop fix is “train the team.” The double-loop question is “is the way we’ve structured the work setting people up to fail?”
Single-loop is faster. Double-loop is the one that actually fixes things.
The 5 Whys — when the surface answer isn’t enough
Toyota popularised the simplest tool in this space. When something goes wrong, ask “why” five times. Each answer becomes the next question.
The point isn’t to be annoying. It’s that the first answer is almost always the symptom, and the real cause is usually three to five levels deeper.
A worked example.
The deadline was missed. Why?
Because the report wasn’t finished in time. Why?
Because the data came in late. Why?
Because the previous step is being done by one person who’s overloaded. Why?
Because nobody else has been trained to do it. Why?
Because cross-training has been on the to-do list for nine months and never gets prioritised.
The actual fix isn’t “tell people to meet deadlines.” It’s “schedule cross-training this week, because the system has a single point of failure.”
Most managers stop at the second or third “why” because the answer there feels close enough. The discipline is going one or two more.
If you can’t get a satisfying answer after five whys, you probably have a system question, not a person question.
The fishbone — when the cause isn’t a single line
The 5 Whys works well when the cause is mostly linear. Sometimes it isn’t. Sometimes a problem has multiple contributing factors that interact.
The fishbone diagram (also called the Ishikawa diagram after the engineer who developed it) maps these out. You write the problem at the head of the fish, then draw “bones” for each major category of cause. Common categories are people, processes, equipment, environment, materials, and measurement.
Under each bone, you list specific factors that might be contributing. Then you look at the whole picture.
The value isn’t in the drawing. It’s in being forced to consider categories you’d otherwise skip. A founder looking at a recurring client complaint might only consider “the team” as the cause. The fishbone makes them also look at the process, the tools, the timing, the way performance is measured, and the wider environment. Often, the actual cause turns out to be something they wouldn’t have looked at without the prompt.
You don’t need software for this. A whiteboard and 20 minutes is enough.
Why this connects to continuous improvement
None of these tools are a one-off exercise. They’re meant to be used continuously, every time something goes wrong, on small things before they become big things. This is the same discipline Deming captured in the PDSA cycle — Plan, Do, Study, Act, repeated.
The businesses that don’t keep solving the same problem aren’t smarter. They’re more disciplined about studying what happens, instead of just acting and moving on. That study step is where the right problem gets identified.
What to do this week
Pick one problem in your business that’s come back at least three times in the last year.
Run the 5 Whys on it. Get to a fifth answer before you decide what to do. If you stop at the second or third, you’ll be fixing a symptom again.
Ask yourself the double-loop question. Is there a rule, assumption, or decision you’ve been holding that’s producing this problem? Be honest. It probably is.
If the cause isn’t linear, draw a fishbone. Twenty minutes on a whiteboard. List the factors under people, process, equipment, environment, materials, measurement. See what shows up that you hadn’t been thinking about.
Decide whether you’re dealing with common or special cause variation. If this is something the system naturally produces, you need to change the system. If it’s a real signal, respond to the signal specifically. Don’t confuse them.
If you’d like a structured way to look at the recurring patterns across your whole business, the workplace diagnostic helps you spot the system-level causes you might otherwise treat as one-offs.
The right problem isn’t always the visible one. Most of the time, it isn’t. The discipline of asking the harder question — every time, on small things, before they become big things — is what separates businesses that keep solving the same thing from businesses that keep growing.
That’s the work. It’s not glamorous. It also works.
For the broader picture across all seven patterns, see What 19 Australian founders taught me about workplaces people don’t leave.
