by Konstantin Tskhay​

April 30, 2025

The #1 Predictor You’re Probably Overlooking in People Analytics

↓

Hi again, friends!

A few announcements I wanted to make!

1. TMU: As some of you know, I created of the technical courses at the Toronto Metropolitan University's People Analytics Program last year. Well, guess what! I will be teaching the second course this coming semester, focused on data visualization and storytelling.
   ​
2. My Practical People Analytics Course is doing very well. So, I have decided to increase the price on the course, given other competitors are charging $1,000+ for content that does not meet the same standard. The new price will be $350 as I want to make sure it is still affordable and will kick in on May 10th. So, if you are interested in checking it out, it is only $150 for now!

Now, onto the newsletter!

This week, let’s talk about one of the most powerful—and underused—ideas in People Analytics:

I know.

It sounds almost too simple.

But if you learn to apply it properly, you’ll build better models and make better talent decisions.

Here’s what we’ll cover today:

• Why past behavior works
• The 3-point test to apply the concept effectively
• How to action past behavior in your analytics

Let’s dive in.

📈 The Science: Why Past Behavior Predicts the Future

Psychology and behavioral economics have long known:

People repeat patterns.

A 2002 meta-analysis by Frank Schmidt (University of Iowa) found that past performance was the single strongest predictor of future performance—stronger than cognitive ability, personality tests, or even structured interviews.

It makes sense:

• Behavior creates habits
• Habits create consistent outcomes
• Outcomes reinforce behavior once again

This creates a loop that is very difficult to break.

In fact, unless something disrupts the loop, most people will continue to behave exactly the same way they did previously.

And they will not even know they are doing it!

Now, what's cool about this concept is that we use it in HR a lot:

• For assessing potential based on past performance
• For hiring by looking over the resume and interviewing
• For succession planning, by looking at management skills

And, since we are so good at using it, we can also quantify it and model it.

🛠️ The 3-Point Test: How to Apply Past Behavior Smartly

When trying to predict a future outcome—turnover, promotion success, engagement—run this quick test.

Here are 3 questions you might want to ask:

1. Is this behavior consistent or a one-off?
2. Have conditions around the behavior change?
3. Are there forces strong enough to disrupt the behaviors?

Past behavior ≠ destiny.

However, it is more likely to recur if the conditions and the environment have not changed substantially.

At least, it will increase your odds of being correct.

Let's review a few ways to consider including past behavior in your models.

📚 Where Past Behavior Predicts Outcomes

First, if you have a measure of past behavior, I recommend you include it in your model.

You will observe that a past instance in behavior will be one of the strongest, if not the strongest, predictors of current behavior.

In simple terms:

• High performers will still be high performers
• Disengaged employees will be more likely to be disengaged
• And, departments with high attention will be losing more talent

Including past behavior in your model doesn’t just improve accuracy—it sharpens your view of what else matters.

Let me explain.

Before you control for past behavior, it can confound your results.
You might mistakenly think other factors (like team alignment, leadership style, training) drive performance, when really, it’s just that top performers tend to stay top performers.

When you account for past behavior first, you effectively account for it before making conclusions.

Only then do you see additional effects that were previously obscured.

Everything is better with an example:

Imagine you’re analyzing what predicts employee performance over the next year.

• If you don't control for past performance, your model might show huge effects for things like "team fit" or "engagement," but they could be inflated—because past high performers tend to be more engaged anyway.
  ​
• If you do control for past performance first, you can now see whether team fit adds predictive power beyond just being a historically strong performer.

You might discover:
​
✅ Past performance explains, say, 30% of the outcome
✅ But team alignment explains another 20%, independently

🔑 Use This Starting Tomorrow

If you’re a People Analytics practitioner (or leader working with data), here’s what you can start doing:

✅ Understand behavior first​
​
Before asking what an employee might do—ask what they’ve already done repeatedly and consistently and whether you have data on your desired outcome from the past.

✅ Context check
​​
You need to recalibrate if the environment has changed significantly (e.g., new leadership, massive hybrid shift). You have to make an assumption of a similar context for the effects to be pronounced.

✅ Prioritize interventions where past predicts risk
​​
For example, if someone shows repeat disengagement signals, you can target coaching or role realignment before they mentally check out and you have to terminate them or they quit.

✅ Build your models behavior-first
​​
In predictive modelling, behavioral variables (e.g., participation rates, promotion cadence, project stretch involvement) should be your first candidates before "sentiment" or "intent" measures. Make sure to include past data if you have it.

Everyone says they’ll do better next time.
Everyone promises they’ll stay engaged, committed, and aligned.

But behavior—real, repeated behavior—tells you the truth.

And truth is where real People Analytics starts.

Next time you’re trying to forecast retention, promotion success, or culture fit, don’t start with wish lists or promises.

Start with the question:

That’s where the signal lives.

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!