by Konstantin Tskhay​

June 15th

What If the Most Important Data Is in the Tail?

↓

Hi Friends,

And as all of us in people analytics, I’ve been thinking about the normal distribution while walking my dog in the morning after the workout.

The nice, familiar bell curve. I don't know if it was the hill by the dog park that triggered the vision. Or it's just how my brain is functioning.

It sort of reminded me of my grad school professor who said something along the lines: if you love stats, you are probably thinking about variance while in the shower.

Back to normal distribution, however.

Most observations sit somewhere in the middle.

Most employees are somewhere around average performance.

Most turnover looks something like last year.

Most compensation changes aren't wildly different from the historical range and the local market.

Most workforce outcomes are normal.

And that's useful information for prediction science.

Until it isn't.

In fact, most things that truly matter don't happen in the middle.

They happen in the tails.

Let me explain:

The Problem With Normal

People Analytics is very good at telling us what's normal.

We calculate:

• Average turnover
• Average engagement
• Average compensation
• Average performance
• Average time to hire

Then we build models around those averages.

Again, useful.

But imagine you're looking at a perfectly normal distribution.

Everything looks fine.

Then something happens way out on the right side of the curve.

A critical group of employees leaves because of the CEO comment.

A regulation changes.

An economic shock hits your industry.

COVID-19.

Suddenly, the average doesn't look quite so useful.

The average didn't warn you.

The tail did (somewhat).

Enter the Black Swan

Nassim Taleb popularized the idea of the Black Swan.

An event that's difficult to predict and has a major impact on your organization.

And, once it happens, everyone suddenly has a story about why it was obvious.

That's an uncomfortable idea for People Analytics.

Because we're often asked:

"What's most likely to happen?"

That's a reasonable question.

But I'd add another one:

"What could happen that would completely change our decision?"

Those are very different questions.

The first asks us to predict.

The second asks us to prepare and assess alternative scenarios.

I think COVID-19 pandemic was a great example of this.

Think about it. All our preparation and thinking did not consider COVID at all. Most of our models focused on the average or the status quo: people will continue to perform, be engaged, and turnover at the set rate. Absenteeism is going to stay where it is and we will have the same revenue and profitability as we have outlined.

But COVID happened.

Companies lost their revenues. People went into a lock down. Company could not move engagement of people. People got sick and did not show up to work even in protective equipment. We had to spend time on this protective equipment.

This was a tail end scenario.

But even though it broke out models.

It also unlocked a new world where remote work stuck around, where we invested more into employee wellness, where we refined our definition of productivity.

The tail end produced an actual change.

Spend Some Time in the Tail

I'm not suggesting we abandon classical statistical thinking.

Quite the opposite.

You need to understand the center of the distribution before you can understand the tail.

But once you know what's normal, start asking some uncomfortable What if questions.

• What if turnover isn't 12% next year—but 25%?
• What if it drops to 5%?
• What would we actually do differently?
• What if a role we assumed would be difficult to automate becomes one of the easiest? Akhem, AI.
• What if a skill we thought was abundant suddenly becomes scarce?
• What if our most important business assumption is simply wrong?

These aren't forecasts.

They're counterfactuals.

And that's the point.

We're not trying to predict the impossible.

We're trying to understand what happens if our assumptions fail.

The 10% Test

One of the simplest ways to do this is to stress-test a number.

Take a metric.

Now imagine it moves 10%.

What changes tomorrow morning?

If the answer is:

"Nothing."

Maybe it's not a very useful metric.

But if the answer is:

"We'd need to change the hiring plan."

"We'd have to revisit our compensation strategy."

"We'd need to rethink our workforce plan."

Now you've found something worth paying attention to.

And sometimes 10% isn't enough.

Because the world doesn't always move in 10% increments.

Sometimes it jumps.

This Is Where People Analytics Gets Interesting

I think we sometimes define the People Analytics job too narrowly.

Predict the outcome. Report the number. Explain the variance.

But leadership doesn't only need to know what's likely. They need to know what happens when the likely thing doesn't happen.

That's where the tails become interesting.

Because the goal isn't to predict every Black Swan.

We can't.

The goal is to become less surprised when the world stops behaving normally.

Until next time,

K

---

Whenever you’re ready, there are 2 ways I can help you:

#1

If you’re still looking to get started in People Analytics, I recommend starting with my affordable course:

​Practical People Analytics: Build data-driven HR programs to 10x your professional effectiveness, business impact, and career. This comprehensive course will teach you everything from building an HR dashboard for business results to driving growth through more advanced analytics (i.e., regression). Join your peers today! ​
​

#2

If you are looking for support in your human capital programs, such as engagement, retention, and compensation & benefits, and want to take a more data-driven approach, contact me at Tskhay & Associates for consulting services. Or simply reply to this email!