Read time: 3 min, 26 secs

Hey there - it's Brian 👋

I have a hot take on data:

“If you torture data long enough it will confess to anything.”

This week, a consulting firm hired me to help them figure out which market to go into.

And in certain situations we rely too much on data!

It’s semi-controversial but…

I found corporate executives would ask for market sizing models they knew were a lie, just to handle their emotions.

Not to make a better decision.

So this week I’ll share which part of the data they knew was made up and how you can avoid it to make better market entry decisions.

This issue is for you if you’re struggling to know:
➟ Should I focus on a specific customer type?
➟ Launch in a specific country?
➟ Launch a new product / service?

Let’s make your business an outlier: 👇

For my friends: What’s going on with Brian?

We had to pause promoting the staffing business!

We had more demand than we expected.

And there’s nothing more important than quality client delivery.

But somehow people are still finding our calendars to book sales calls!

We’re hiring internally to keep up with demand. All I can think about for this week is our own values to make sure we built the best culture on the planet.

5 years ago I would have laughed.

In consulting we were more focused on process than people. I had no idea how critical values + culture are to really scaling a business.

Anyway…

Anyway this weekend I rode my first dune buggy outside of Lima. INCREDIBLE experience.

Check out the picture.

And now on to market entry (101): 👇

What’s wrong with data?!

So let’s clarify real quick. What does I mean by “torturing data to confess?”

I did a market entry project where we had to decide if a $1B consulting business should start giving Wealth Management services.

We’ll call it WM.

Partner says:
“Hey I just want to make sure the market is growing. Can you tell me if it’s big enough for us?”

So you get exploring.

You buy a data set. You find free datasets. Research reports. And marry it all together.

Sounds simple. But if you’re a good data analyst you start asking questions about the data.

And that’s where you find a big problem.

Every data set defines things in a way that YOUR business likely disagrees with.

Example quotes from that project:
➟ Well we don’t count that as WM! Cut out that section.
➟ You think Tech strategy for WM will grow THAT fast? Adjust the growth rate.
➟ We don’t target Tech Ops! We need to cut them out
➟ Why aren’t you including the Private Bank?! We need to add them in

So you snip data out. Add from other sources. Adjust growth rates.

And BOOM. You have a custom dataset that actually applies to your business.

But…

It’s YOUR assumptions.

So you’ll find you can tweak the slices until…

The data confesses to you what you need to hear.

If you don’t make adjustments… it won’t apply to you.

If you DO make adjustments… there’s room for error.

So what do you do?

How to not torture your data

So the quality of your decisions is only as good as your inputs.

You can have the most advanced model in the world but if you feed it fuzzy data you’ll get fuzzy answers.

Market entry data is fuzzy. You can’t treat it with such precision.

Because that’s fake.

So for fuzzy data, logic needs to win BEFORE you check with data.

Here’s the 6 steps we did to check the logic on market entry:

1) List your business strengths / advantages

2) List markets we can use those strengths (this is your initial hypothesis)

3) Data check (are these industries are growing)

4) Look at trends (see headwinds / tailwinds to plan your focus)

5) What do we need to build or buy to enter? (How long will it take? Cost?)

6) Go-To-Market strategy (What channels? What’s our advantage?)

And if that made sense - we entered the market.

🧔🏻‍♂️ Brian’s nerdy side rant:

See how data wasn’t the main decision-maker there?

It raises flags if we see something we don’t expect. But it’s not the whole thing.

An example on the opposite end of the spectrum I did a theft-prevention strategy for a mega pharmacy. We had receipt data which was wildly accurate.

So we built a big model and let data make a lot of the decisions.

It was crazy.

I hit “go” on the model and it suggested an employee was stealing. We interviewed the employee and BOOM he was!

Market entry data tends to be fuzzy. Not thief catching clarity.

Quality inputs = quality outputs.

Please don’t torture data

In other words don’t create a false level of precision.

If you have fuzzy data… set expectations that you’ll give fuzzy answers.

If you have accurate data… you can catch thieves stealing pharmacy gift cards.

See you next Thurs.

If you liked this post, and don’t want to miss future issues, sign up for free:

If you found this helpful, please forward this email to 1 friend or colleague. They'll appreciate you and you'll help grow the community.

See you next Thursday 👋

Brian
Twitter
LinkedIn

🙋 Vote: How did we do?