The Precise Prediction application uses two advanced statistical modeling techniques to:

• Predict a salary range for full-time employees and
• Predict the total salary of a specific employee based on multiple factors (location, education, experience, etc.)

Pro+ members can use this app to:

• Quickly generate a second-opinion on hard to price employees.
• Distill multiple sources of data into a quick, expert-level prediction.
• Create a bespoke salary range based on the precise budget of a church—one that discards preset or custom ranges entirely.
• Create salary packages for employees and churches where data is thin or non-existent.

Overview

The Precise Prediction app on ChurchSalary is designed to quickly empower Pro+ members to distill the entire landscape of numbers inside of a salary report down into a precise salary range and predicted total salary.

To replicate that human-driven process, Precise Prediction follows the same step-by-step methodology that structures all of our salary reports:

1. Find a range → based on nationwide patterns
2. Tune the range → based on location
3. Place employees within the range → based on compensation factors

Precise Prediction replicates the first, second, and last steps of this process in a few seconds using advanced statistical modeling techniques. In a few clicks this powerful application can distill an entire report down into a predicted range and precise salary figure.

But how exactly does ChurchSalary use quantile regression and a proprietary AI model to generate a set of precise predictions?

In-Depth Explanation

Finding a Salary Range Using Quantile Regression

Normally, in order to find a salary range you need to group churches together using a budget and/or attendance range. This gives you a cohort of “similar employees serving at similar churches” that you can use to accurately predict compensation.

Similar Employee

• Position
• Status (FT/PT)

Similar Churches

• Budget Range
• Attendance Range (optional)

The Basic and Hybrid Reports enables Basic and Expanded members to quickly define “similar churches” using two sets of 10 predefined budget and attendance ranges. That allows the report to find the 25th, 50th, and 75th highest paid employees within those predefined budget or attendance ranges. 

The Precise Report empowers Pro and Pro+ members to define a custom range that places their church in the middle of the distribution. Once that custom range is defined, the same math is used to find the 25th, 50th, and 75th highest paid employees in the cohort of “similar employees serving at similar churches.”

By contrast, the Precise Prediction application uses Quantile Regression to continuously predict a set of percentiles.

Rather than grouping employees together based on a budget range, this type of regression analysis finds a continuous line that predicts any quantile (i.e., percentile) within the distribution of similar employees. Using this method, we can predict a salary range at any point on the graph by simply solving a set of math formulas. Rather than finding a specific set of employees and doing math, this method is creating a salary model for all similar employees.

When you select any full-time position and feed a budget number into the Precise Prediction app, it automatically predicts the entire set of percentiles based on that precise budget.

By discarding ranges, our new Quantile Regression model can generate salary ranges even if the underlying data is thin or non-existent. This makes it especially useful for pricing megachurch employees.

For example, a megachurch with a $35M budget will struggle to find salary data to price their employees because there are very few churches with $35M budgets in the real world and not all of them have shared their information with ChurchSalary. Our new advanced quantile regression model enables ChurchSalary to extrapolate a salary distribution for executive pastors across  all churches, regardless of size.

Predicting Total Salary Using ChurchSalary AI

Pricing an employee requires leaders to make a series of decisions, after distilling tons of information, with the goal of arriving at a single number. What makes this process even harder is that every leader will make a slightly different set of decisions even if they are handed the exact same set of figures.

The Precise Prediction application simplifies this decision-making process by using the machine learning model built into our ChurchSalary AI model.

ChurchSalary AI is a proprietary, math-based machine learning tool that was designed using math as both the input and output—the training data and the prediction. Because we built this model in-house, using our knowledge of which variables impact salary the most, we were able to train the model to replicate the decisions made by all of the leaders who set pay at the over 20,000 churches in our database.

The end result is a prediction of how a human might price this specific employee based on the inputs you provide. It won’t necessarily give you the exact same prediction that you may make. And it is definitely not infallible. It is making inferences based on the data we’ve accumulated from leaders—not all of whom are making the most rational, consistent decisions.

But we think it’s pretty amazing. And it’s really good at making predictions.

AI can be a scary word these days. Keep reading if you have more questions about how our ChurchSalary AI model works.

Questions about Our ChurchSalary AI Model?

Q: Does the ChurchSalary AI model rely on a large-language model (LLM) built by another company?

A: Absolutely not! To protect the privacy of churches and employees, ChurchSalary only creates and maintains internal, in-house AI models of our salary database.

Q: What do you mean by proprietary?

A: To protect the privacy of churches and employees, we paid an expert AI researcher to build a math-based model from scratch. None of ChurchSalary’s database was fed into a tool or technology controlled by some other company. We deliberately used in-house tools and storage (and will continue to) to protect the privacy of churches and employees.

Q: What do you mean by math-based?

A: Most of the “AI” that you use is powered by large-language models or LLMs. These LLMs are extremely complex ways to encode and predict human language patterns. While the data that powers LLMs is technically encoded using math, the model that ChurchSalary uses starts and ends with math. In essence, we trained a machine learning algorithm using pure numbers, not words encoded as numbers. That algorithm generates a predicted number by weighting a set of input variables much like a human working through a set of decisions.

Q: Does ChurchSalary sell or share data?

A: ChurchSalary does not sell data to third parties. Neither ChurchSalary, nor our parent company, are data brokers. 

ChurchSalary maintains its own website and database for safety and security. We are assisted in this work by a world-class team of engineers and privacy experts from our parent company, Gloo, who help us adhere to the highest standards of safety and security.

Your data continues to be used in the same way it always has: to analyze salary and help churches and church networks make informed decisions about compensation.