Excel Data Analysis: A Realistic (Messy) Walkthrough

A Confession About Excel Data Analysis Before We Start

I've done three separate dry runs of what you're about to read.

My intention was to document my analytical process, figure out the high points through repetition. But I realized one of my biggest personal gripes with analytics instruction: it's very sterile. Very fenced in. Very prescriptive.

There's a saying: Prescription without diagnosis is malpractice.

I am not in the business of conducting malpractice.

Analytics Is Not Linear

Analytics is, by virtue of what it is, not really a structured thing. It's a little bit amorphous. Non-linear.

A lot of common schools of thought give you clean environments, clean sandboxes, clean examples. Step 1, Step 2, Step 3—here's your picture-perfect analysis.

I don't want to do that. That's not the real world.

Maybe that approach works if you're trying to learn a specific technical skill. But the reality of doing ad hoc analysis—which is what most analytical requests are—is that you don't always have direction. You don't always know the exact order of operations.

It's curiosity. This is where it gets fun.

What You're About to Watch

You're about to watch me—someone who's been an analyst since 2004—struggling a little bit.

I'm not going to put the Instagram filter over this. This is going to be real. I'm probably going to forget how to do things. It's not going to be perfect.

But it's realistic.

I want you to experience that.

The Request

The board is asking to see how volume looked in Q2. I got some data but didn't have a chance to pull anything together. I think they just want to see Q2 2021 volume by region and wanted to know if everything was looking good.

That's it. Q2 2021 volume by region. Are things looking good?

The Lazy Analyst Trap

If I were a brand new analyst without a good role model, I might:

1. Build a pivot table
2. Make a quick chart
3. Send it back and say "Here you go!"

Don't do that.

That's lazy. You're effectively telling them: "Hey, how about you go ahead and do the job you just asked me to do."

What did the analyst actually contribute? You pushed a couple buttons, made a little chart, and sent it back open to interpretation.

You didn't add any value. Analytics exists up here (points to head), not in the toolset.

What I Actually Found

After digging into the data:

The Headline

Q2 year-over-year growth slowed from Q1 growth of 4% down to 2.7%.

The Story

• Latin America went from 9% growth in Q1 to 0% in Q2
• Two customers left the region in Q2, taking about 7,000 units of volume
• That's almost 50% of our total variance

The Rest

Same-store sales (existing customers) were slightly down across the board, but nothing dramatic. The big story is those two lost customers.

The Key Insight Everyone Misses

While reviewing North America, I noticed a new client that started in Q2 2020. They ramped up through the year and were contributing significant volume by Q1 2021.

But now we've anniversaried them.

Their volume was completely incremental in Q1 (no prior year comp). In Q2, we finally have a prior year to compare against. That's why growth looks slower—not because anything is wrong, but because the easy comp is gone.

This is the kind of insight that comes from walking away and coming back with fresh eyes.

The Real Deliverable

I'm not going to recite numbers and say "this was up, this was down." That's called elevator analysis—and it's a horribly common mistake.

Instead, I'm going to tell them:

1. Q2 grew 2.7% year-over-year (down from 4% in Q1)
2. Latin America drove most of the slowdown—two lost customers accounted for 50% of the variance
3. Same-store sales were slightly softer across regions
4. North America's growth is normalizing as we anniversary a large new customer

That's a story. That drives conversation. That leads to better questions.

Good Data Drives Good Questions

What's going to happen next:

• "Who were those customers? What happened?"
• "Is the same-store slowdown a trend?"
• "Should we be worried about seasonality?"

This is how analytics works. You present findings, people ask questions, you dig deeper. The flywheel starts spinning.

If you just send a pivot table? The conversation dies.

Common Questions About Excel Data Analysis

Q: How do I know when I've done enough analysis?

When you have a clear story that leads to action. If someone reads your analysis and doesn't know what to do next, you're not done.

Q: What if I don't find anything interesting in the data?

That's actually interesting. "Everything is trending as expected" is a valid finding. But make sure you've actually looked—don't just assume everything is fine.

Q: How much time should I spend on ad hoc requests?

Depends on the stakes. Board-level requests? Take the time to get it right. Quick operational question? 30 minutes max. Learn to calibrate effort to impact.