Was This a Problem of Knowing, Deciding, or Doing?

In the previous essay I showed how product and engineering teams consistently miss their objectives because they conclude before examining all three dimensions of a failure. I called this The Causal Stack. And I showed how four invisible forces make sure the conclusion forms before the investigation starts.

Why Does Your Team Keep Shipping and Missing?

Shipping and missing? The failure is rarely where you’re looking. The Causal Stack reveals why teams conclude too early and what to ask before the next one does.

Thought MunchiesPrashant Bhargava

If you haven't read it, start there. This one is where the framework becomes something you can actually use.

When the Conclusion Forms Too Early

You shipped a feature customers had been asking for, some for over a year. You had the right signals. Desperate requests from account managers. Sales data. A business case that pointed to 20% adoption. The team was energized. This was the one.

Or so you thought. Three months later you're still sitting at 3%. The retro is in an hour.

Before that room jumps to a conclusion, one question needs to be asked.

Was this a problem of knowing, deciding, or doing?

Knowing points to the bet.
Deciding points to the solution.
Doing points to the execution.

Most teams never ask it. The explanation hardens into a conclusion. The retro produces action items. The team moves on. And the real failure, the one that will repeat in the next initiative and the one after that, is never found.

This pattern shows up far beyond software. And sometimes the clearest way to see a failure you're living inside is to watch it happen somewhere else first.

A Real World Example

You're part of a team working on a pilot for a public library.

The library footcount is down. The municipality is watching. Funding is quietly being questioned.

After some research the team lands on a bet: Young adults aren't coming because the library isn't relevant to their lives.

The solution: Introduce Adulting Courses - filing taxes, building resumes, navigating careers.

It feels right. It feels meaningful.

Your library runs a six month pilot. The team puts everything into it. However, attendance stays low and engagement, poor.

The staff reconvenes. They are drained, not just by the effort, but by the outcome. Nobody wants to say it out loud. Nobody wants to be the one who questions the work everyone just gave six months of their life to.

"The courses seemed valuable, but nobody attended."

"AI can teach this stuff better than we can. To compete we'd need certified professionals. That's not in the budget."

The pilot is reluctantly scrapped and the team moves on.

But you're still sitting there.

Something doesn't feel right. The team worked hard. The problem was real. Young adults genuinely struggle with adulting. The library was a good place to help. And yet the only answer anyone found was "AI is cheaper and certified professionals cost too much."

You start turning it over in your head.

Maybe the bet was wrong.

Maybe young adults in this area don't actually want structured courses.

Maybe they're fine figuring things out on their own.

But wait. Were they even asked?

Did anyone actually go talk to them before building the program?

Or maybe the bet was right but the solution was wrong?

Maybe adulting courses were never the answer. Maybe a different format or different topics would have brought them in.

Or maybe both were right and it was just the execution.

Wrong timing.

Wrong instructors.

Wrong awareness. The right people never even knew it existed.

And then it hits you.

Nobody in that room asked any of these questions. The conversation went straight from "the pilot failed" to "AI is too hard to compete with."

Three months of work. Six months of pilot. And the only question anyone asked was how to exit this cleanly.

Was this a problem of knowing, deciding, or doing?

That's the question that should have been asked before the pilot was scrapped. And there are three places to look for the answer.

Knowing. Deciding. Doing.

1. Knowing - Was the Bet Ever Real?

The bet was simple: adulting courses will bring young adults back to the library.

But every bet has assumptions buried inside it. Assumptions that feel obvious. Assumptions that nobody writes down. Assumptions the whole project gets built on top of.

Did anyone actually talk to young adults before designing the program?

Not survey them. Not assume what they need. Actually talk to them. Watch how they currently solve these problems. The team assumed young adults need help with taxes and resumes. But did anyone ask whether they want that help from a library in the first place?

Reasonable isn't the same as tested.

How are young adults already solving these problems?

This is where the AI conversation gets interesting, not as a budget problem, but as a signal. If young adults are already using YouTube, Reddit, or AI tools to figure out their taxes and resumes, that's not a competitor. That's data. It tells you the assumption, that people want a structured course in a physical space, was already being beaten by behavior that existed before the pilot launched.

The team called AI a cost problem. It was actually a knowing problem. The environment had already changed. Nobody checked.

Was low footcount caused by what the team thought it was?

Low footcount has many causes. Maybe young adults don't know what the library offers. Maybe the hours don't work. Maybe the space doesn't feel like theirs. Adulting courses solve one version of the problem. The bet assumed it was the right version.

Was it?

You haven't solved anything yet. You've just made the bet visible. And visible is where the work starts.

2. Deciding - Did Anyone Actually Decide?

Let's say the bet held. Courses are the right way to bring young adults back. The knowing layer checks out.

Now the harder question. Not whether to run courses, but what those courses actually looked like. Who decided that. And whether anyone ever really decided at all.

Was this solution chosen, or did it drift into place?

Think back to the planning meetings. Did anyone sit down and say "we are choosing tax filing courses because young adults in this area specifically told us this is what they need and this is how they want to learn it?"

Or did it just... happen? Someone mentioned taxes. Someone else nodded. A template from a previous program got dusted off. The agenda filled up. And before anyone realized it, six months of work was committed to something that was never really chosen. It just became the path of least resistance.

That's not a decision at all.

And the painful thing about this approach is that everyone involved did their job. Nobody was lazy. Nobody was careless. The drift happened in the space between good intentions and the questions nobody thought to ask.

Who actually shaped what the courses looked like?

Was it the person who had run programs before and knew what was easy to deliver? The loudest voice in the planning meeting? The one with the most conviction that tax filing was the right topic?

Whoever shaped the courses had assumptions baked in. And those assumptions were probably never checked.

Did anyone go back to young adults and ask what topics they actually wanted help with? And how they wanted to learn? In a group or alone. In person or online. Structured or on demand. Topics and format are both deciding layer questions. Both are almost never validated.

Were these courses built for young adults, or for the team's idea of young adults?

There is a difference. And it's the kind of difference that doesn't show up until six months later, when attendance is low and the room is drained and someone says AI is cheaper than certified professionals and everyone nods because they're too tired to push back.

The courses weren't wrong because the team was wrong. They were wrong because the right questions were never asked when it still mattered.

You haven't solved anything yet. You've just made the roadmap visible. And visible is where the work starts.

3. Doing - Did the Pilot Get a Fair Chance?

Let's say both held. The bet was right. The decisions about topics and format were right. Young adults want this. The library is the right place. The courses were the right shape.

Now ask the hardest question of all.

Did the pilot actually give any of that a fair chance?

When did the courses run?

Wednesday evenings. Maybe a Saturday morning. Someone picked those slots because the room was available. Because the instructor could make it. Because it fit the schedule of the people planning it, not the people it was designed for.

Young adults are not free on Wednesday evenings. They're finishing work. Commuting. Cooking. Exhausted from a day that started too early. Nobody asked them when they were free. The slot got picked and the invitations went out and then everyone wondered why nobody showed up.

The pilot didn't test whether young adults want adulting courses. It tested whether young adults will rearrange their lives for a Wednesday evening program at a library.

Those are not the same experiment.

Who taught the courses?

It matters more than anyone admitted in the planning meeting. A librarian who researched tax filing is a different experience from someone who files taxes for a living and genuinely loves helping people understand it. People feel the difference in the first five minutes. They decide in those five minutes whether they're coming back. Whether they're telling a friend.

If the instructor wasn't credible because they were the most available option, the course never had a chance to succeed on its own merits.

How did people find out it existed?

A flyer on the notice board. A post on the library's Facebook page. Maybe an email to existing members.

That's not marketing to young adults. That's marketing to the people who already come to the library. Your target audience wasn't in that room because nobody went to find them where they actually are. Instagram. TikTok. A WhatsApp group. A Reddit thread.

The program ran. For the wrong audience. In the wrong place. At the wrong time. With the wrong messenger.

The pilot didn't fail. It never really ran.

At least not in a way that actually tested the bet or the decisions made around it. The execution had so many gaps that the results can't tell you anything useful about whether the courses were right or wrong.

Scrapping the pilot was premature and based on evidence that was never evidence at all.

You haven't solved anything yet. You've just made the execution visible. And visible is where the work starts.

What One Question Unlocked

Two hours ago the library team had one exit.

Scrap the pilot. Blame the budget. Move on.

And in an hour your team walks into a retro with the same exit waiting.

A feature that 97% of your customers never touched.

A room full of people who already have an explanation.

And four invisible forces making sure the conclusion hardens before anyone asks the right question.

Look at what happened when one question was asked instead.

1. The brain stopped finding the nearest exit

The nearest cause was "AI is cheaper and certified professionals cost too much." That felt like a conclusion because it was recent, concrete, and nobody had the energy to fight it.

But the knowing questions don't accept budget answers.

You can't respond to "did anyone actually talk to young adults?" with "AI is cheaper."

The question has no exit. It forces the brain upstream whether the room wants to go there or not.

2. The assumption finally left a trail

Nobody wrote down "we believe young adults in this area want structured courses and will attend them on Wednesday evenings."

It emerged from planning meetings and became invisible the moment it stopped being questioned. There was nothing to audit.

"Did anyone actually talk to young adults before designing the program?" forces the assumption into the open.

Not by finding a document.

By making the team reconstruct it and say it out loud for the first time.

Now it's visible. Now it can be right or wrong instead of just assumed.

The question didn't find the trail. It created one.

3. The right people are now named

The pilot debrief had everyone in the room. But not the right everyone.

Whoever shaped the bet needs to answer the knowing questions.

Whoever drifted into the course format needs to answer the deciding questions.

Whoever picked Wednesday evenings and put a flyer on the notice board needs to answer the doing questions.

Three layers. Three conversations. Three different rooms with three different people.

The question separated the diagnosis but also the accountability. And accountability is the only thing that makes the next attempt different from this one.

4. The layers are speaking the same language

"AI is cheaper than certified professionals" was execution vocabulary being used to make a strategic decision.

It shut down a conversation that needed to happen at the knowing layer, in the language of hypotheses and assumptions, not budgets and costs.

Each layer now speaks its own language.

Knowing questions only accept answers about what was validated.
Deciding questions only accept answers about what was chosen and why.
Doing questions only accept answers about how it was run.

The conclusion that formed in execution language can no longer stand in for a strategic insight.

None of this required new data. No research report. No external consultant. No budget.

Just one question, asked before the room defaulted to the nearest exit.

Was this a problem of knowing, deciding, or doing?

And now instead of one premature conclusion, the library team has three honest conversations to have. Each aimed at a layer that was invisible two hours ago. Each with the right people. Each in the right language.

The pilot doesn't need to be scrapped. It needs to be understood.

Now Make It Yours

That library team moved on. The same way your team will move on after the retro today.

Unless someone asks the question first.

Think about the feature sitting at 3%. The signals were real. The business case made sense. The team worked hard. And yet here you are, an hour away from a room where everyone already has an explanation and nobody has an answer.

Was this a problem of knowing, deciding, or doing?

Was the signal real or filtered through too many hands?

Was the solution chosen or did it drift into place?

Did the launch actually reach the right people at the right time with the right message?

You don't know yet.

And that's the point. Nobody in that room stopped to find out. The conclusion formed. The exit looked reasonable. And the real failure, the one that will repeat itself in the next initiative and the one after that, was never found.

You still have time before the retro starts.

Ask the question.

Ask It Earlier Next Time

You asked the question. You looked at all three dimensions. You found where the conclusion formed too early.

That feels like enough. And for today, it is.

However, the library team didn't just fail to ask the question, but they also failed to ask it when it mattered. While the courses were being designed. While the timeslots were being booked. While the flyer was being printed.

The earlier you ask, the less there is to recover from.

Before a bet gets locked.
Before a solution drifts into place.
Before an execution plan gets committed to.

One question. Three dimensions. Ask it before the room needs it, not after.