Skip to main content

Drivers Explained: When a Driver Is Statistically Correlated and When It’s Not

Learn how RunSmart identifies which drivers are statistically correlated using R-squared and how you can view or change them.

Updated over 3 weeks ago

In RunSmart, a driver is something that helps explain why a line item in your Profit & Loss (P&L) statement or Balance Sheet changes over time. For example, if your Sales Revenue goes up and your Shipping Costs also increase, Sales Revenue might be driving Shipping Costs.

But not everything that moves together is truly connected. To make sure we’re only using meaningful relationships, RunSmart uses a statistical test called R-squared to determine whether a driver is actually correlated to a line item—and reliable enough to use in your forecasts.


What Is a Driver in RunSmart?

A driver is a financial input that helps explain or predict the behavior of another line item on your P&L or Balance Sheet. Some common examples:

  • More Sales Revenue might increase your Cost of Goods Sold

  • A higher headcount could raise your Payroll Taxes

RunSmart automatically links certain line items to likely drivers based on industry best practices, so you don’t have to start from scratch. But if you prefer, you can modify those drivers and choose your own.


How Does RunSmart Decide If a Driver Is Statistically Correlated?

To determine if there’s a real connection between two line items, RunSmart analyzes your historical financial data and calculates something called R-squared.


What Is R-squared (in Simple Terms)?

R-squared is a number between 0 and 1 that shows how closely changes in one line item explain changes in another.

  • 1.0 = Perfect correlation (one line item fully explains the other)

  • 0.0 = No correlation (no relationship at all)

RunSmart’s Rule:

A driver is only considered statistically correlated if the R-squared value is 0.75 or higher. That means at least 75% of the changes in one line item can be explained by its driver.


Example: Sales Driving Shipping Costs

Let’s say you want to see if Sales Revenue drives Shipping Costs. Your historical data might look like this:

Month

Sales Revenue

Shipping Costs

Jan

$10,000

$2,000

Feb

$12,000

$2,400

Mar

$14,000

$2,800

Apr

$16,000

$3,200

As Sales Revenue goes up, Shipping Costs go up at the same rate. RunSmart calculates:


R-squared = 0.95

✅ This means Sales Revenue is strongly correlated with Shipping Costs. RunSmart will treat it as a valid driver in your forecasts.

If the R-squared had been just 0.60:
❌ The relationship isn’t strong enough. RunSmart will not treat Sales Revenue as a driver of Shipping Costs.


What Happens If R-squared Is Below 0.75?

If the correlation between two line items is weak (R-squared < 0.75), RunSmart does not treat the relationship as a valid driver. This helps ensure your forecasts are based only on meaningful, statistically reliable patterns—not chance or noise.

Instead, when a driver isn’t statistically correlated, RunSmart automatically applies advanced forecasting algorithms to project the line item reliably. This ensures you still get a solid, data-driven forecast—even when no strong driver exists.


View Drivers and Correlations in RunSmart

You can see which drivers are linked to which line items—and whether or not they’re statistically correlated—by toggling “Show Correlation” on the P&L or Balance Sheet pages.

For each line item, RunSmart shows:

  • The driver we’re using

  • A simple “Yes” or “No” to indicate if the correlation is statistically valid (R-squared ≥ 0.75)

This gives you transparency into how your forecasts are built, so you can adjust them if needed.


Customize Your Drivers

While RunSmart automatically links line items to commonly used drivers based on industry best practices, you can change them if something else on the list of selectable drivers makes more sense for your business.

You’re in control—and we’ll still let you know whether your custom driver is statistically correlated or not.


Summary

R-squared

What It Means

What RunSmart Does

0.75–1.0

Strong correlation between line items

Uses it as a driver in your forecast

Below 0.75

Weak or no reliable correlation

Does not use it as a driver

Did this answer your question?