Data Driven Decision Making: 7 Decisions Leaders Get Wrong

Data driven decision making does not mean letting a dashboard make the decision for you. It means using evidence to improve your judgment before you commit money, people, or time to a choice.

There is an important difference.

I have sat in enough meetings where the decision was effectively made in the first five minutes and the next forty-five minutes were spent finding data to support it.

The chart changes. The conclusion somehow survives.

That is not data driven decision making.

That is decision-driven data selection.

The better test is simple: could the evidence have caused you to make a different decision?

If the answer is no, the data was decoration.

Below are seven common business decisions where leaders can use data well, use it badly, or accidentally use it to become more confident in a decision they already wanted to make.

What Data Driven Decision Making Actually Means

A useful definition is simpler than most consulting diagrams make it:

Data driven decision making is a repeatable process for using relevant evidence, business context, and human judgment to choose among alternatives and learn from the result.

Notice that judgment did not disappear.

It should not.

Data tells you what has happened, what is happening, and sometimes what is likely to happen next. It can show patterns your experience missed. It can also be incomplete, stale, badly defined, biased by the way it was collected, or perfectly accurate and still irrelevant to the decision in front of you.

The leader still has to decide.

The goal is to make that judgment harder to fool.

Data Driven Decision Making Example 1: Cutting a Product

Suppose one product has been losing money for three quarters.

The spreadsheet says kill it.

This looks like an easy data driven decision until someone points out that the product is frequently the first purchase made by customers who later buy your much more profitable services.

Now the question changes.

Instead of asking:

“Is Product A profitable?”

you need to ask:

“What happens to total customer value when Product A is the entry point?”

The original data was not wrong. The decision was framed too narrowly.

This happens constantly in business. A metric answers the question it was built to answer, while the executive assumes it answered a larger one.

Leadership lesson: Before accepting a metric, write down the decision it is supposed to inform. If the metric and the decision are measuring different things, stop.

Data Driven Decision Making Example 2: Choosing Which Project to Fund

Imagine three projects competing for the same $1 million.

Project A promises $2 million in annual benefit.

Project B promises $1.5 million.

Project C promises $900,000.

If we sort the spreadsheet, Project A wins.

Then you look closer.

Project A’s estimate depends on six assumptions that nobody has tested and requires eighteen months before the first benefit appears.

Project B already has a successful pilot with measurable results.

Project C has the smallest benefit but solves a regulatory problem the company cannot ignore.

Now the numbers still matter, but the ranking is no longer obvious.

Good data driven decision making does not ask which project has the biggest number.

It asks how much confidence you have in that number, how long it takes to realize the benefit, what must be true for it to happen, and what happens if you do nothing.

I would rather see an investment proposal with a believable $1.5 million than a heroic $4 million held together by twelve optimistic assumptions.

Data Driven Decision Making Example 3: Deciding Whether a Team Is Performing Well

This one causes trouble because leaders love scorecards.

Imagine a data engineering team reports:

• 99.9% platform uptime
• 18% faster incident resolution
• 27% more pipelines delivered
• 92% of sprint commitments completed

Everything is green.

Then the executive asks, “Is the business getting more value from this team?”

The scorecard may not contain the answer.

The team measured what its machinery produced. Leadership needs to understand what changed outside the team.

Did Finance close faster?

Can Supply Chain adjust orders weekly instead of monthly?

Did analysts stop spending Monday morning reconciling three versions of the same number?

Did business teams voluntarily stop using their private spreadsheets because they trust the shared data?

Activity is useful for operating a team.

Outcome measures are useful for deciding whether that activity deserves continued investment.

Leadership lesson: When reviewing a metric, keep asking “and what happened because of that?” until you reach something the business actually cares about.

Data Driven Decision Making Example 4: Responding to Conflicting Data

This is where being data driven gets uncomfortable.

Your ERP says inventory performance improved.

Your warehouse report says stockouts increased.

Your planning spreadsheet says you have too much inventory.

Three reports. Three conclusions.

The natural reaction is to ask which report is correct.

Sometimes all three are correct.

They may use different dates, definitions, grains, business rules, or points in the process.

For example:

• ERP may count an order when it is entered.
• The warehouse may count it when goods ship.
• Finance may count it when revenue is recognized.

The problem is no longer “find the correct dashboard.”

The problem is deciding which definition is appropriate for the decision being made.

If you are deciding how much product to order next month, the answer may be very different from the metric you would use to explain quarterly revenue.

Leadership lesson: Do not resolve conflicting numbers by choosing the report you trust most. Resolve them by understanding what each number actually measures.

Data Driven Decision Making Example 5: Following Data That Contradicts Experience

Experienced leaders develop intuition for good reasons.

They have seen hundreds of situations that never made it into a dataset.

So what should happen when the analysis says one thing and an experienced executive says another?

I would not automatically choose either one.

I would ask the executive:

“What do you know that may not be represented in this data?”

Maybe a major customer is about to leave.

Maybe a supplier relationship deteriorated last week.

Maybe the market changed after the reporting period ended.

Maybe the executive is recognizing a pattern the model cannot see.

Or maybe none of those things is true and experience is simply anchoring the person to an old belief.

There is a second question that helps:

“What evidence would change your mind?”

I use that as a test of whether we are examining a hypothesis or defending a preference.

If someone cannot describe any evidence that would change the recommendation, more dashboards probably will not help.

Data Driven Decision Making Example 6: Acting With Incomplete Data

Another mistake is assuming a data driven company waits until it has enough information to remove uncertainty.

You may wait a very long time.

Leadership decisions often arrive before the data is complete.

Should we enter this market?

Should we reorganize the team?

Should we invest another $2 million in this product?

Should we move an AI pilot into production?

There may never be a dataset that tells you exactly what to do.

The better approach is to make the uncertainty visible.

For a major decision, I like separating the evidence into three buckets:

• What we know: Evidence we consider reliable
• What we believe: Assumptions supported by some evidence but still uncertain
• What we do not know: Gaps that could materially change the decision

Then decide whether any item in the third bucket is important enough to delay the decision.

This prevents two bad extremes.

One is pretending you have certainty because executives are expected to sound confident.

The other is analysis paralysis, where the team keeps asking for more data because nobody wants to own the call.

You rarely need perfect information.

You need enough information to make the next decision responsibly.

Data Driven Decision Making Example 7: Judging Whether a Decision Was Good

This may be the most overlooked part of the process.

Imagine you approve a risky expansion and revenue jumps 30%.

Great decision?

Maybe.

What if the assumptions behind the decision were terrible and the company got lucky because a competitor unexpectedly exited the market?

Now imagine another leader makes a careful decision based on the best information available, but an unexpected regulatory change causes the project to fail.

Bad decision?

Again, maybe not.

Good decisions sometimes produce bad outcomes.

Bad decisions sometimes produce good outcomes.

If you judge every decision solely by the result, your organization will eventually reward luck and punish reasonable risk-taking.

A better review asks:

• What did we believe would happen?
• What assumptions did we make?
• What evidence did we use?
• What actually happened?
• Which assumptions were wrong?
• What should we do differently next time?

This is where data driven decision making becomes a learning system instead of a meeting ritual.

A Simple Data Driven Decision Making Framework

You do not need a twelve-box methodology for every business choice.

For decisions that actually matter, I would use six steps.

1. Write the Decision Down

Be precise.

“What should we do about declining sales?” is a discussion topic.

“Should we move 20% of next quarter’s marketing budget from enterprise acquisition to mid-market retention?” is a decision.

2. Define What Success Means

What outcome are you trying to improve?

Revenue? Cost? Retention? Risk? Time? Quality?

Agree before opening the dashboard.

3. Identify the Evidence That Matters

Ask what information would genuinely help distinguish between the alternatives.

This step prevents the meeting from becoming a tour of every metric the organization happens to collect.

4. Write Down the Assumptions

This part is important because assumptions have a strange habit of turning into facts after being repeated in three meetings.

Write them down while everyone still remembers they are assumptions.

5. Ask What Would Change Your Mind

Do this before the final decision.

If Strategy A is currently preferred, what evidence would make Strategy B the better choice?

If nobody can answer, you may already have a conclusion looking for support.

6. Schedule the Review Before You Leave the Room

Do not say, “We should revisit this later.”

Later is where decision reviews go to die.

Choose a date.

When that date arrives, compare what you expected with what actually happened.

The One-Page Decision Record

For expensive or hard-to-reverse choices, I would write down the decision before the meeting ends.

This can fit on one page.

That is intentional.

If documenting the reasoning takes twenty pages, nobody will do it consistently.

Three Questions I Would Bring Into the Next Leadership Meeting

You can improve the quality of a meeting without changing a single dashboard.

“What decision are we actually making?”

This stops a thirty-minute metric discussion that has no decision attached to it.

“What evidence would change our current recommendation?”

This exposes confirmation bias very quickly.

“When will we know whether we were right?”

This turns the decision into something the organization can learn from later.

I would take those three questions over another executive dashboard with seventeen tabs.

What Data Driven Decision Making Is Not

It is not letting the person with the best dashboard win.

It is not replacing experienced judgment with a score.

It is not waiting for perfect information.

It is not asking analysts to prove the executive’s preferred answer.

It is not declaring every good outcome evidence of a good decision.

And it certainly is not putting “data driven” in a strategy presentation and assuming the organization has become one.

Those are all easier.

The harder version requires leaders who are willing to state what they believe, show the evidence behind it, admit what they do not know, change their minds when the facts change, and return later to find out whether their reasoning held up.

The Takeaway

Data driven decision making is really a discipline for making judgment more inspectable.

You define the decision before looking for supporting evidence. You distinguish facts from assumptions. You ask what could change your mind. You make the call even when some uncertainty remains. Then you come back later and compare what you believed with what actually happened.

Do that consistently and the value of your data changes.

It stops being something people display in meetings.

It becomes something that changes decisions.

I go deeper into how senior data and analytics leaders create this kind of decision discipline, communicate with executives, build trust, and make the data function matter to the business in The Data-Driven Executive: How Data and Analytics Leaders Build Influence and Lead in the Age of AI. If your decisions involve AI investments specifically, AI to ROI for Business Leaders covers the execution side from defining outcomes through proving business impact.