Four Kinds of HR Analytics a Small Employer Can Actually Use

HR analytics sorts into four questions: what happened, why it happened, what is likely to happen next and what you should do about it. A small employer can answer the first two with payroll records and a spreadsheet, the third with a simple headcount forecast, and the fourth with a written decision that a named person owns. Start with eight counted numbers before you attempt anything predictive.

I am Mac Malbrough. I hold a J.D. and an MS-HRM, I am not an attorney, and I help small employers turn records into decisions. The numbers in the examples are hypotheticals I made up to show the arithmetic. They are not benchmarks or client results.

Why does a small team need a way to sort its questions?

When people hear “HR analytics” they picture a platform with charts and a price tag. For a company of 15 or 80 people, the problem is different. You have payroll in one system, applicants in another, a leave log in a third, and a question from the owner that begins with “why do we keep losing people in the second year?” Nobody can answer it, because nobody has decided what kind of question it is.

Sorting the question tells you what data you need and how much effort the answer deserves. It also protects you from the most common mistake, which is trying to predict something you have never counted. If you do not know last year’s turnover, a forecast of next year’s turnover is a guess with a chart around it.

What is descriptive analytics, and what should you count first?

Descriptive analytics answers “what happened?” It is counting, done the same way every time. For a small employer the starting list is short: headcount at the start and end of the period, separations, hires, time to fill, offer acceptance, early exits in the first 90 days, absence and where each person’s pay sits in the range for the job.

I covered the definitions and the one-page layout in HR metrics a small employer should track, and the formulas in how to calculate HR metrics correctly. If you want the eight numbers in a workbook with one definition each, the HR Metrics Starter is built for a 15-minute monthly review.

For national context, the Bureau of Labor Statistics publishes the Job Openings and Labor Turnover Survey each month. In its September 2026 release, employers across the economy reported 5.4 million hires, 5.3 million total separations and 3.4 million quits, which BLS expresses as monthly rates of 3.4 percent, 3.3 percent and 2.1 percent. Those rates describe all industries and sizes together. They are useful for knowing whether you are in a hot or cold market, not for judging your own team.

What does diagnostic analytics add?

Diagnostic analytics answers “why?” by cutting the same numbers a different way. You are looking for a pattern that points to a cause you can change. The cuts that most often matter at small scale are:

• By manager: do exits cluster under one supervisor?
• By tenure: do people leave at month six, at the two-year mark, or right after a review cycle?
• By role: is the problem in one job family or everywhere?
• By pay position: are leavers mostly below the midpoint of the range?
• By source: do people hired from one channel leave sooner?

Take a hypothetical. A company of 40 people loses six employees in a year. Four of the six reported to the same manager and left between month 14 and month 20. That pattern does not prove the manager is the cause, but it tells you where to ask questions. Pair the cut with conversations: Gallup has reported that managers account for at least 70 percent of the variance in employee engagement scores across business units, which is why a manager-level cut is usually the first one worth making. My article on why a great manager is a retention strategy goes into the reasoning.

One caution on small numbers. With six exits, one more or one fewer changes the rate by a large amount. Read a small-sample pattern as a prompt for a conversation, not as a finding. Never present a percentage without the count behind it.

How do you forecast headcount without data science?

Predictive analytics answers “what is likely next?” At small scale that means a workforce plan built from a few inputs you can defend. The formula is plain arithmetic:

1. Start with current headcount by role.
2. Subtract the exits you expect, using your own last two years of separations by role as the base, not a national rate.
3. Add the roles the business plan requires: growth, new services, coverage gaps.
4. For each open slot, work backward from the start date using your own time to fill to find the date you must post.
5. Multiply by loaded cost to see what the plan costs for the year.

The Headcount and Hiring Forecast workbook does this with post-by dates, plan-year cost including federal payroll tax and a budget check. It is the tool I would use before asking for a hiring budget.

Also forecast the risk, not only the count. The Succession Planner for Critical Roles scores the roles where one person holds knowledge nobody wrote down. A role that scores high is a forecast in itself: if that person left in March, what would stop? To size the skills side of the plan, the skills gap analysis workbook compares what the business needs with what the team can do today.

Where does prescriptive analytics fit?

Prescriptive analytics answers “what should we do?” In large companies this often means software that recommends an action. In a small company, it is a decision with a name on it. You take the diagnostic finding, list two or three options, estimate cost and risk for each, and choose one. Then you write down what you chose, why, and when you will check whether it worked.

If you use AI to draft options, treat the output as a draft. A tool can summarize a spreadsheet, but it does not know your obligations, your culture or the person in the next office. I wrote about where the line sits in where AI helps HR and where a person must decide. For the written record, the AI Decision Log for HR captures what the tool did, what a person checked and who decided, which works equally well for decisions that did not involve AI.

What are the limits and risks?

Three limits matter for a small employer.

• Small samples mislead. Do not run percentages on groups of fewer than ten people, and report counts alongside rates.
• Privacy and fairness apply. If you cut results by age, sex, race or another protected category to look for disparities, keep the data access narrow and use the results to find and fix problems, not to make individual decisions. If a pattern suggests a legal problem, talk with an employment attorney.
• Records need to be accurate. A dashboard built on inconsistent dates and job titles will give you confident wrong answers. Agree on definitions first and write them down.

How do you present the findings to an owner?

Owners rarely want a chart. They want to know what is happening, whether it is a problem and what you recommend. Use a three-line format for each finding: the number and its trend, the likely reason in one sentence and the decision you want. For example: “Early exits were three this quarter against one last quarter, all in the same team. The likely cause is a mismatch between the job as posted and the job as done. I recommend a rewrite of the posting and a day-30 check-in.” Each line can be defended with the data behind it, and the owner can say yes or no.

What should you do this week?

Pick the one question that keeps surfacing in your leadership meetings, and say which of the four kinds it is. If it is a “what happened” question, count it. If it is a “why” question, make two cuts of the same data and write one sentence about each. If it is a “what next” question, build a one-page headcount plan. Do the smallest version that answers it.

Start with the HR Metrics Starter for the counting, then the Headcount and Hiring Forecast for the plan. If you want help setting the definitions and presenting the numbers to leadership, book a free consult through my HR services for small businesses.

Related reading: workforce management for small employers.

Sources

• U.S. Bureau of Labor Statistics, Job Openings and Labor Turnover Survey, September 2026, released November 3, 2026: hires, separations and quits
• Gallup Business Journal, “Managers Account for 70% of Variance in Employee Engagement,” April 21, 2015
• How to calculate HR metrics correctly, for the formulas used here

McKinley holds a J.D. but is not a licensed attorney. Articles here are general information, not legal advice. For your specific situation, talk with an employment attorney.

Data current as of November 2026. Sources are linked where each figure appears.