Most Growth Problems Are Diagnosis Problems

A disappointing result tells you something is wrong. It doesn’t tell you what to fix. Here’s how I separate symptoms from causes and pick a useful next test.

growth diagnosis - conceptual image

The most expensive sentence in a strategy meeting is a confident one.

“We know what the problem is.”

Sales is down, so sales needs fixing. Traffic is flat, so marketing needs more reach. People are leaving, so culture needs work. The product isn’t growing, so it needs more features.

Sometimes that’s right. Often it just names the place where the pain showed up. It’s the symptom.

This matters because most companies are good at execution once they think they know the problem. They start a project, form a team, and move money over. They set targets and push hard.

A weak diagnosis leads to fast, disciplined work in the wrong direction.

That’s worse than doing nothing.

Problems show up in one place and start in another

Most reporting is built to show symptoms. A dashboard says conversion fell from 8.2% to 6.9%. A forecast says the quarter will miss. A survey says manager confidence has dropped. A pipeline report shows one stage getting worse.

All useful. None of it explains itself.

It’s tempting to treat the place where the problem appeared as the place where it started. Conversion fell on the landing page, so fix the landing page. The miss showed up in sales, so rebuild sales. People say they’re burned out, so add wellness benefits.

We assume the problem started where we first saw it. It showed up here, so it must start here.

Companies don’t work that neatly.

A team can miss its number because the positioning stopped being relevant six months ago. Sales can slow down because a pricing change added someone new to the buying committee. Marketing can look wasteful because the way buyers find you changed and the measurement didn’t.

Burnout is often less about hours than about decisions that never get made, incentives that pull against each other, and hard work that never seems to turn into progress.

The problem is real. You just don’t know the cause yet.

Good execution can hide a bad diagnosis

This is why strong teams are most at risk.

When execution is weak, the failure is obvious. Missed deadlines, unclear ownership, and sloppy work.

When execution is strong, a wrong idea can survive for a year.

The team works hard, the plan looks solid, and weekly meetings get tighter. Everyone can point to activity. There are some results, but not enough.

So the company does more of it. More campaigns, more training, more pressure. They get a new leader, a new agency, a new tool.

Eventually someone decides the strategy failed. But the strategy never got a fair test, because it was answering the wrong question from the start.

The biggest cost is trust. People stop believing the next plan is any different from the last one. They learn that when results disappoint, the company just pushes harder.

That’s when execution really starts to fall apart.

Three things worth keeping separate

Most companies lump these together. Each one needs a different kind of answer.

1. The problem is what’s happening.

  • Revenue is under plan. Retention is falling. Deals stall. Customers are confused. Decisions take eleven weeks.
  • Describe it. Don’t explain it yet. If your sentence has the word because in it, you’ve already left this step.
  • It may feel like a waste of time, but it’s the cheapest step to get right.

2. The mechanism is what’s causing the problem.

  • Maybe customers don’t see the value fast enough. Maybe details get lost in a handoff. Maybe the buying process changed. Maybe managers can’t make the call at the level where it needs to be made.
  • This is a claim about cause, which means it can be wrong. That’s the point. A claim you can test beats a confident label everyone nods at.

3. The condition is what keeps it going.

  • A decision stays slow because four executives have to approve it. A bad customer experience stays bad because the two teams who own either side of it are paid on opposite numbers. An experiment program stays fake because being wrong hurts your career, so nobody runs a test that could fail.

Here’s the simplest way to keep the last two apart. Mechanisms are verbs. Conditions are nouns.

Details get lost in the handoff. That’s a verb.

Two teams own the same customer and are measured against each other. That’s a noun.

A project can fix a mechanism. If the condition stays, the mechanism comes back.

That explains a pattern you’ve probably seen. Fix the problem, and you get relief. Fix the mechanism, and you get relief. Change the condition, and you have a shot at something that lasts.

Most projects that worked and then stopped working fixed the mechanism and left the condition in place.

How to tell whether you actually have a diagnosis

It should do more than sound reasonable. It should predict something.

If this is really what’s causing it, we should also see this other thing.

That one sentence will improve a strategy conversation more than any framework.

If you think low conversion comes from weak messaging, then customer interviews should show confusion about value before anyone brings up price. If you think slow growth comes from bad leads, then win rates should be clearly better on the leads your team calls good. If you think burnout is about workload, then cutting workload should help even when the slow decisions and unclear roles stay exactly the same.

You don’t need to turn your leadership team into a lab. You just need to stop rewarding explanations that can’t be wrong.

An explanation that fits every possible outcome tells you nothing.

The other way this goes wrong

This can go wrong in the other direction too, and it’s harder to spot.

Once a team learns to look at conditions, “it’s structural” becomes the answer to everything. It sounds smart. It can’t be disproven. And it lets everyone off the hook.

So don’t call a condition finished until you’ve answered two more questions:

  • Who could change this?
  • What would changing it cost them?

If nobody can change it and saying it costs nothing, it isn’t a diagnosis. It’s a complaint that sounds smart.

What AI changed

It made this more urgent.

You can now produce analysis, plans, options, research, forecasts, and a full rollout deck in an afternoon. That’s a real advantage.

It also means a weak diagnosis can turn into a polished plan before anyone notices it’s wrong.

When doing things gets cheap, choosing what to do matters more. Answers are easy to get now. Asking the right question is harder. So is judging the evidence and how sure you should be.

The risk is that AI makes early, unearned certainty look professional.

Diagnosis and action go together

I want to be careful here, because this argument gets misused.

Diagnosis isn’t a reason to slow down, and it doesn’t replace doing the work. A good diagnosis is worth nothing if nobody acts on it. Its only job is to make sure you work on the real problem. Done well, it’s the fastest way to your goal, because you stop spending time on the wrong things.

You need both.

That’s why the most useful next step is usually a small test. Run the smallest thing that could actually change your mind.

And it’s rarely one thing. Companies usually have several conditions holding them back at once. But an organization that tries to change five things changes none, because attention and political capital run out long before ideas do.

The honest version is that you can only change one thing at a time with enough force to matter.

That’s about order. There’s rarely a single cause. You work on one change while you set up the next one. Some conditions take a quarter to set up and a week to execute. If you only work on them one at a time, back to back, you spend most of the year waiting.

Two better questions

Instead of asking what should we do about this?, start one step earlier:

What must be true for our explanation to be right?

That question moves the conversation from opinions to claims you can check. It brings assumptions out in the open. It gives the skeptics something useful to argue with.

Then ask:

What’s the cheapest test we could run that would make us more sure?

Cheap in time, money, and energy. Now you’re moving again, and this time the work tells you something.

Try it on something real

Take the problem your company is spending the most money on right now.

Write down the symptom without explaining it. Write your current explanation as a claim that could be wrong, with a confidence number next to it. Then write one thing you’d expect to see if that claim were true, and one thing you’d expect to see if it were false.

If you can’t fill in that last part, you don’t have a diagnosis yet.

Better execution still matters.

But first, ask the right questions just long enough to find the right problem.

Written by Aaron Edgell

Growth executive. Builder. Writer.

Get new essays and notes by email.

One email when I publish something new.

Previous:
Next: