From Numbers to Decisions: 7 Lessons from My Guest Lecture at Jindal Global Business School

• A metric without context rarely tells you what to do.
• Good analytics moves from measurement to diagnosis, hypothesis, decision and learning.
• Acquisition only makes sense when you connect it to activation, retention and unit economics.
• The goal of experimentation and forecasting is not certainty—it is better decision-making under uncertainty.

Imagine someone tells you:

Our conversion rate is 5%.

The obvious next question is:

So what?

Is 5% good?

Was it 7% last month?

Is one acquisition channel converting at 9% while another converts at 2%?

Did something change in the product?

Are high-intent users dropping out at a particular step?

Are the customers who convert actually coming back?

A metric becomes useful only when it is put in context.

One simple framework I use is:

Measure → Diagnose → Hypothesise → Decide → Experiment → Learn

Most analytics work should eventually move through this chain.

If it stops at Measure, you have reporting.

If it reaches Decide, you have decision-making.

There is a big difference between the two.

One of the easiest mistakes to make with analytics is opening a dashboard first and looking for something interesting.

That often creates analysis without purpose.

A better starting point is a question.

For example:

Why has first-purchase conversion fallen?

Now the analysis has direction.

You might break the problem into:

• traffic source
• device
• geography
• new vs returning users
• funnel stage
• payment method
• customer cohort

The data is no longer the destination.

It is evidence being used to answer a question.

This sounds obvious, but it changes the way you work.

Instead of asking:

“What does the dashboard show?”

ask:

“What decision am I trying to make?”

Then work backwards to the information you need.

Growth problems often look large until you break them into stages.

Suppose 100,000 people visit a product.

20,000 create an account.

12,000 complete onboarding.

6,000 begin a transaction.

4,500 complete it.

Saying:

“4.5% of visitors convert”

is technically correct.

But it doesn’t tell us much.

The interesting questions are between the steps.

Why did 80% of visitors not sign up?

Why did 40% of registered users not finish onboarding?

Why did 25% of people who started the transaction not complete it?

Each drop-off may represent a completely different problem.

One could be messaging.

Another could be trust.

Another could be product friction.

Another could be pricing.

Another could be a technical failure.

The funnel does not give you the answer.

It tells you where to look for the answer.

That distinction matters.

Acquisition is easy to celebrate because it moves quickly.

Users go up.

Installs go up.

Sign-ups go up.

Traffic goes up.

But businesses are not built only by getting people through the door.

They are built by delivering enough value that the right people stay, transact again, subscribe, renew, refer someone or come back when they need the product.

Suppose two campaigns each acquire 10,000 customers.

Campaign A has a lower acquisition cost.

Campaign B costs more.

At first glance, Campaign A wins.

But imagine customers from Campaign B are twice as likely to make another purchase three months later.

Which campaign was actually more efficient?

Now the answer is less obvious.

This is why acquisition metrics should rarely be viewed in isolation.

CAC without customer quality can be misleading.

Sign-ups without activation can be misleading.

Activation without retention can be misleading.

The further you can connect an acquisition decision to downstream behaviour, the better the decision usually becomes.

Growth percentages can sound impressive.

“Revenue grew 50%.”

“Users doubled.”

“Transactions increased 80%.”

But eventually every business has to answer:

Did we create more value than it cost us to create it?

That is where unit economics becomes useful.

Consider a simplified example.

If it costs ₹1,000 to acquire a customer who eventually generates ₹400 in contribution, scaling acquisition simply scales the problem.

If the customer generates ₹5,000 over a sufficiently long relationship, the exact same acquisition cost might be extremely attractive.

This is why metrics like:

• customer acquisition cost
• lifetime value
• contribution margin
• repeat rate
• payback period

matter.

Not because finance terminology makes a strategy sophisticated.

Because growth eventually has to survive contact with economics.

A lot of business decisions are made in situations where nobody knows the answer.

Should we change onboarding?

Should we introduce an offer?

Should we move this step earlier?

Should we change pricing?

Should we send a reminder?

Should we remove a field?

Everyone can have an opinion.

The more useful question is:

How could we find out?

That is the purpose of experimentation.

You create a hypothesis.

For example:

Reducing the number of fields during onboarding will increase completion without reducing customer quality.

Then you decide what evidence would support or reject that hypothesis.

You define:

• the audience
• the change
• the control
• the metric
• the duration
• the success criterion

Then you test it.

The important mindset shift is from:

“I think this will work.”

to:

“Here is how we can determine whether it works.”

That is a much stronger way to run a business.

Forecasting sometimes creates a false sense of precision.

If a spreadsheet says revenue next quarter will be ₹12.4 crore, that does not mean the future has revealed itself.

It means we have created an estimate based on assumptions.

What happens if conversion falls 10%?

What happens if CAC rises?

What happens if retention improves?

What happens if transaction frequency changes?

A good forecast should help you understand those relationships.

In other words:

The value of forecasting is not certainty. It is preparedness.

The objective is not to predict one perfect future.

It is to understand what the business might look like under different plausible futures—and what decisions those scenarios would require.

If I had to compress the entire session into one idea, it would be this:

Data → Context → Insight → Decision → Experiment → Learning

Data tells us what happened.

Context tells us whether it matters.

Insight gives us a possible explanation.

Decision determines what we should change.

Experiment tests whether our thinking was right.

Learning improves the next decision.

And then the cycle starts again.

That, to me, is what analytical thinking in business actually looks like.

Not dashboards for the sake of dashboards.

Not metrics because everyone else tracks them.

Not complicated mathematics because complexity looks impressive.

Just progressively better decisions.

Preparing for the lecture created an unexpected test.

At work, it is easy to use words such as:

CAC.

LTV.

Activation.

Cohorts.

Incrementality.

Retention curves.

Conversion funnels.

Everyone around you may already understand the vocabulary.

But when you’re speaking to students, terminology cannot do the work for you.

You have to explain the underlying idea.

Why does CAC exist?

Why should you care about retention?

Why do we need a control group?

Why is a conversion rate useful?

What decision does the metric actually improve?

And that forced me to revisit some ideas from first principles.

I came away with a simple reminder:

If you understand something well, you should eventually be able to explain why it matters without hiding behind jargon.

Teaching is a very effective way to find out whether you can.

Being comfortable with numbers is valuable.

But most people in business will never be rewarded simply for knowing how to calculate a percentage.

What matters is what happens afterward.

Can you identify the right problem?

Can you ask a better question?

Can you separate signal from noise?

Can you make a reasonable hypothesis?

Can you decide what to test?

Can you change your mind when the evidence disagrees with you?

That is where numbers become useful.

The spreadsheet is not the decision.

The dashboard is not the decision.

The metric is not the decision.

They are tools that help us make one.

And that was really what I wanted this session to be about.

What is the difference between reporting and decision-making?

Reporting explains what happened. Decision-making uses that information to decide what should happen next. Good analysis connects a metric to context, a hypothesis and an action.

Which metrics matter most for product and growth?

There is no universal list. The useful metric is the one connected to the decision being made. Acquisition metrics might matter when evaluating a channel; activation and retention become more important when understanding whether customers actually receive lasting value.

How do you turn data into an actionable insight?

Start with a specific business question, examine the relevant data, identify where behaviour changes, form a plausible explanation and determine what action or experiment could test that explanation.

Why are experiments important in business?

Because most business decisions involve uncertainty. Experiments help distinguish an idea that sounds reasonable from one that actually changes customer behaviour.

Guest Lecture: From Numbers to Decisions: The Math Behind Product, Growth & Business

Audience: BBA students

Institution: Jindal Global Business School, O.P. Jindal Global University

Date: September 16, 2026

Thank you to Jindal Global Business School, Prof. Navneet Kumar Singh and the students for the opportunity and the thoughtful participation.

I enjoyed the experience far more than I expected.

And I would happily do more of it.

I’m Saurav Mishra. I work across growth, product, lifecycle and analytics, with a particular interest in how customer behaviour and data translate into business decisions.

Explore my case studies →