Why We Keep Solving the Wrong Problems

Article thesis: “Most businesses don't fail because they lack data. They fail because they solve the wrong problem.”

Imagine you're driving a car, and suddenly the engine warning light comes on.

You have two options.

The first is to cover the warning light with black tape.

The second is to open the hood and understand why the warning light appeared in the first place.

The first option removes the symptom. The second solves the problem.

Surprisingly, many businesses operate exactly like the first option. They react to symptoms rather than understanding the system that created them.

Sales dropped? Increase marketing.

Employee productivity is low? Hire more people.

Customer complaints increased? Expand the support team.

Sometimes these decisions work. Often, they only postpone the real problem. A fast response can be sensible; it simply should not be confused with a verified diagnosis. Looking beyond an apparent problem can expose root causes and reduce the risk of merely treating symptoms (Interaction Design Foundation).

This is exactly why Systems Thinking exists.

What Is Systems Thinking?

Systems Thinking is a way of understanding how different parts of a system influence one another over time.

Instead of asking, “What happened?” it asks, “Why did it happen?” and, more importantly, “What interactions inside the system could have caused it?”

Every organization, whether it's a startup, a hospital, a manufacturing plant, or even a family, is a collection of interconnected systems. No decision exists in isolation. Incentives change behaviour. Capacity limits create queues. Information arrives late. One team's sensible local response can become another team's constraint.

That does not mean every outcome has one hidden “root cause.” It means the boundary of the question matters. Systems thinking asks us to look beyond the part that is easiest to see and examine the wider context before assigning cause (Interaction Design Foundation). The result is not automatic certainty. It is a more inspectable explanation.

“A visible symptom tells you where pressure appeared. It does not necessarily tell you where intervention belongs.”

A Simple Example

Hypothetical illustration — the 15% figure below is invented and is not reported Netflix data.

Imagine Netflix notices that monthly revenue has dropped by 15%.

A traditional response might be:

• Increase advertising
• Launch discounts
• Send promotional emails

These aren't necessarily bad ideas. But they all assume one thing: marketing is the problem.

A systems thinker would pause before making any decision. Instead, they'd ask questions like:

• Has customer retention changed?
• Has the quality of new content declined?
• Are support tickets increasing?
• Did the recommendation algorithm change?
• Did pricing affect customer behavior?

Suddenly, the problem looks very different. After connecting these pieces, the team might test a chain like this:

Support response time increased → Customer satisfaction declined → Subscription renewals reduced → Revenue fell

In this hypothetical, the revenue problem wasn't caused by marketing. It was caused by customer experience. If Netflix had simply increased advertising, it could have attracted more customers into a broken system.

The chain is a hypothesis, not proof. Each arrow would need evidence: trend timing, customer cohorts, support records, renewal behaviour, and plausible alternatives. That factual label matters because a convincing diagram can create false confidence. The purpose of the example is to show how the diagnosis changes when the team connects functions—not to claim knowledge of Netflix's actual operations.

“The chain is a hypothesis, not proof.”

Everything Is Connected: Why Systems Thinking Matters More Than Ever

One of the biggest mindset shifts in Systems Thinking is realizing that businesses don't operate as independent departments.

Marketing influences sales. Sales influences customer expectations. Customer expectations influence support. Support influences retention. Retention influences revenue. Revenue influences hiring. Hiring influences product development.

Everything is connected.

This is why changing one small part of a system can create unexpected consequences somewhere else. The word “can” is important. A connection is not automatically a material cause, and a map is not a substitute for measurement. But ignoring the connection can be equally costly when a decision changes demand, workload, incentives, or customer behaviour outside the team that owns the metric.

A practical systems question is therefore: If this action works locally, what will it change elsewhere—and when will that effect become visible?

Why Dashboards Aren't Enough

Modern companies have more dashboards than ever before:

• Revenue dashboards
• Sales dashboards
• Marketing dashboards
• Product dashboards

The problem isn't the lack of data. The problem is that dashboards mostly tell us what already happened.

They rarely explain, on their own:

• Why it happened
• Which decisions caused it
• What will happen next
• Which action will have the greatest impact

That limitation is not a criticism of dashboards. A well-designed dashboard is valuable for monitoring defined measures. But a display of observations does not, by itself, establish causality or choose an intervention. If teams also distrust the underlying data, the dashboard becomes less useful as a shared reference and cross-functional decisions can slow (Trailique).

This is where Systems Thinking becomes powerful. Instead of looking only at isolated metrics, it focuses on the relationships between those metrics. A useful review pairs the dashboard with a relationship map that labels what is observed, assumed, and not yet measured. That prevents a correlation or a neat arrow from quietly becoming a fact.

“A dashboard can monitor a signal. It cannot, by itself, establish its cause.”

The Birth of Systems Thinking

The formal foundations of this approach were developed in the 1950s by Professor Jay W. Forrester at MIT.

While working with managers at General Electric in the mid-1950s, Forrester examined how internal hiring and layoff decisions interacted with the wider organisation. The System Dynamics Society's history records that his stock-flow-feedback calculations showed employment instability could arise from the firm's internal structure rather than an external business cycle (System Dynamics Society).

This led to the development of System Dynamics—a formal approach for modelling how complex systems behave over time. Its language represents stocks, flows, relationships, feedback, and time delays (Thwink). Systems thinking is the broader lens; system dynamics is one modelling method that can turn a dynamic explanation into something that can be examined or simulated.

Illustrative application areas from the original article include:

• Supply chain management
• Public policy
• Climate modelling
• Healthcare
• Manufacturing
• Urban planning
• Business strategy

The list is illustrative, not a claim that every field uses the same method or evidence standard. Even after more than 60 years, the core idea remains surprisingly simple: everything is connected. The difficult work is deciding which connections matter for the decision in front of you.

Why This Matters More Than Ever

Editorial interpretation: Today's organizations often feel significantly more complex than they were in the 1950s. A single business decision may affect marketing, finance, product, customer success, engineering, operations, and sales—all at the same time.

In everyday decision-making, linear stories are easier to hold in mind than systems containing feedback, delayed effects, and competing explanations. That is one reason organizations can make decisions that seem logical in the short term but create larger problems later.

There is also a boundary. Not every fault needs a system model. If a problem is bounded, directly observable, and safely reversible, fix it and measure the result. Use the wider lens when the issue keeps returning, crosses functions, contains meaningful delays, or allows one team to improve its own metric while the enterprise outcome worsens. Systems thinking should reduce avoidable confusion, not manufacture complexity.

“Systems thinking should reduce avoidable confusion, not manufacture complexity.”

Enter Artificial Intelligence

For decades, Systems Thinking required experts. Building a formal model meant understanding feedback loops, mathematical equations, stocks, flows, and simulation software.

Today, AI changes part of that equation—but not the need for evidence or judgement.

Instead of manually reviewing every record, AI-assisted tools may help teams search for patterns and candidate relationships across large amounts of organizational data. Imagine bringing together:

• CRM data
• Customer feedback
• Financial reports
• Project plans
• Product analytics

Instead of generating another dashboard, a team could ask: “What relationships might exist between these systems?”

AI could support pattern discovery, scenario construction, or explanation of model assumptions. Columbia Business School describes organizations using generative AI for competitive intelligence and predictive analysis based on historical data and assumptions (Columbia Business School). Those words—historical data and assumptions—set the boundary. A predicted relationship is not a demonstrated cause, and a simulation is only as credible as its structure, inputs, and validation.

Current AI systems also have material technical limitations; the UK's AI Security Institute identifies multiple capability limits and evidence that would be needed to show progress (AI Security Institute). Work reviewing AI in market research similarly stresses the continuing importance of the human element (Greenbook).

Rather than replacing human decision-making, AI has the potential to become a reasoning partner: useful for proposing relationships, surfacing contradictions, and exploring scenarios, while people remain responsible for boundaries, causal evidence, values, and the final decision.

“A predicted relationship is not a demonstrated cause.”

Looking Ahead

The future of decision-making isn't simply better dashboards or more reports. It's building systems that help organizations understand how they behave—systems that don't just answer “what happened?” but also ask:

• Why did it happen?
• What happens if we change this?
• Which decision creates the best long-term outcome?

Systems Thinking is no longer just an academic concept. As AI becomes more useful for exploring relationships and simulations, it may become one of the foundational building blocks of how organizations make decisions in the future. That is a possibility, not a guarantee: causality still needs evidence, simulations still need validation, and consequential choices still need accountable human judgement.

A simple relationship check

• Observed — Meaning: A pattern appears in relevant data.; Next question: Is it reliable, repeated, and decision-relevant?
• Assumed — Meaning: A proposed relationship may explain the pattern.; Next question: What evidence would support or disprove it?
• Not yet measured — Meaning: A potentially important variable is missing.; Next question: Can it be measured before the decision scales?

Conclusion

Because the best decisions rarely come from looking at isolated events.

They come from understanding the system that created them.

Map one symptom before you scale the fix

Choose one live business problem. Write down the visible symptom, the connected teams and variables, any likely delays, and the strongest competing explanation. Label each relationship observed, assumed, or not yet measured before committing the larger response.