Overtime Analysis in Power BI: Staffing or Spike?

Overtime analysis in five cuts: department, month, reason, role and employee. See how to tell a staffing problem from a one-off spike with a Power BI…

The overtime line is over budget again, and the first question in the review is always the same: do we hire, or do we wait it out? A single total cannot answer that. Two departments can burn the same 100 overtime hours, one because it is a person short every week, the other because of a deadline that is already over. Hiring for the second one wastes a salary. Waiting out the first one wears people down.

This guide shows how to tell the two apart with five cuts of the same overtime data. Every example uses the Overtime Spending & Workforce Cost Intelligence dashboard, a one-page Power BI template, and its built-in sample data, so the numbers illustrate the method and are not a benchmark. If you want the wider payroll picture first, read our guide to payroll and overtime cost control in Power BI. This article is about one decision.

The short version: five cuts, two patterns

Overtime that is a staffing problem is constant, concentrated and recurring. Overtime that is a spike is clustered, explained and temporary. Each cut below checks one of those words, and the table shows what to look for in each.

CutWhere to look in the dashboardLooks like a staffing problemLooks like a spike
1. Department and roleHours and records bars by department and by job titleA few departments and roles carry most of the hours, period after periodHours spread thin, or one department high in one period only
2. MonthOT Distribution Heatmap (department by month)The same department is shaded in most monthsA few dark cells, then blank ones
3. ReasonTotal OT Hours by Overtime Reason donutEveryday workload reasons recur (call volume, bug fixing)A named one-off or calendar reason (a launch, a year-end report)
4. PeopleTops Employees by OT table and the Emp. ID filterThe same names lead month after monthThe names change from one period to the next
5. Cost trendMonthly cost line and the YoY OT Cost Change % cardCost keeps rising from a similar base of employeesOne bump that returns to the earlier level

Start with the five KPI cards as a baseline

The dashboard opens with five cards: Total OT Hours, Total OT Cost, Total Unique Employees, Avg OT Cost per Employee and YoY OT Cost Change %. Use them as a baseline before you cut anything. One division is worth doing by hand: hours divided by unique employees. In the sample data that is 366.5 hours across 15 people, about 24 hours each.

How two numbers move together tells you something. Hours up with the same number of people means the same people are working more. Hours up with more people means the load is spreading, which usually points to a wider process issue. Because every chart on the page cross-filters the cards, they also show how big whatever you select is: click one department and the cards re-count for that department alone.

The four analysis panels of the Overtime Spending & Workforce Cost Intelligence dashboard, shown with its sample data.

This part of the page is where the analysis happens: a heatmap, a donut by reason and two bar charts. The cuts below take them one at a time.

Cut 1: Where does overtime concentrate? Department and job title

The two bar charts show Total OT Hours (purple) and Total OT Records (teal) by department and by job title. Hours tell you how big the load is. Records tell you how often it happens, and the ratio between the two is a clue.

Purple is total overtime hours, teal is the number of overtime records. Sample data.

Then compare the two charts. In the sample data IT leads by hours, followed by Engineering and Finance, and the top roles are Operations Manager, IT Support Specialist and Network Engineer. When the top roles sit inside the top department, as the two IT roles do here, the load is concentrated in one skill set, which you can hire or cross-train for. When the top roles are scattered across unrelated departments, the cause is more likely process than headcount.

One caution: with a handful of employees a single role can be one person, so check Total Unique Employees before you read too much into a bar.

Cut 2: Is it constant or does it spike? Read the heatmap

The OT Distribution Heatmap puts departments in rows and months in columns, with darker cells for more overtime. The Cost and Hrs buttons switch what the colour measures: hours show the workload, cost shows the money.

Department by month. Darker means more overtime; the Cost / Hrs buttons switch the measure.

You are looking for the shape of each row, not the darkest cell. Three shapes show up in the sample data:

Now click the IT bar in the department chart and the whole page re-filters to IT.

Filtered to IT: one heatmap row shaded in 11 of 12 months, all five reasons present, and two IT job titles highlighted.

IT's overtime is 69 hours from just 2 employees, about 19% of all the hours in the sample. Its heatmap row is shaded in 11 of 12 months. The donut shows all five reasons present, so no single event explains it, and the job-title chart highlights IT Support Specialist and Network Engineer. Constant, concentrated and multi-cause is the staffing-problem signature. It is also a retention risk: two people carrying a department all year.

Cut 3: Why is it happening? Let the reason decide the fix

The donut shows Total OT Hours by Overtime Reason. The sample data uses five reasons: High Call Volume, Emergency Maintenance, Bug Fixing, Annual Report and Project Launch.

Overtime hours by reason in the sample data. High Call Volume is the largest slice at 51.5 hours (27%).

Sort whatever reasons you have into three groups, because each one calls for a different response:

Click a slice and the heatmap redraws for that reason alone. Here is High Call Volume, the largest slice:

Filtered to High Call Volume: 51.5 hours, 8 employees, seven departments.

It touches 8 employees across seven departments, and it returns to the same places: IT and Operations each show it in three different months. That recurrence is what makes a reason a capacity problem. Now compare Emergency Maintenance:

Filtered to Emergency Maintenance: 39.0 hours, seven departments, nine different months.

It also reaches seven departments, but across nine different months, with one or two cells in most departments (Finance is the exception). Each hit reads like an event, not a pattern.

One data warning: a reason field is only as good as the people filling it in. If most of your overtime lands in "Other" or stays blank, fix the capture before you draw conclusions.

Cut 4: Who is carrying it? Top employees and the Emp. ID filter

The Tops Employees by OT table ranks the biggest contributors, with their hours, hourly rate, total and monthly average. The Emp. ID filter in the top corner isolates one person, and the page redraws for that employee. This is the cut that confirms or kills the staffing story. If the same names lead month after month, the workload is built into their roles. If the names change from one period to the next, the pressure moves around with the work, which fits a spike or a scheduling problem better than a headcount gap.

Treat this view as people data. Share it with HR and the managers who act on it, and read it as a workload signal, not a performance ranking.

Cut 5: Is the cost rising from a stable base?

The monthly line compares overtime cost across two years, and the YoY OT Cost Change % card sums that up in one number. Read it next to the headcount in the cards. Cost that climbs year over year while the number of overtime employees stays flat means the same people are costing more: a structural pattern. A single bump that falls back to the earlier line is a spike. If cost rises faster than hours, look at the rate column in the top-employees table before you decide anything.

A simple decision rule

Count how many of the five cuts point the same way: constant in the heatmap, concentrated in the bars, recurring reasons in the donut, the same people in the table, and a cost trend that keeps rising from a stable base. This is our rule of thumb, not a legal threshold.

  1. Four or five signals: you have a staffing case. Take the dashboard to finance and ask for the role the data points to.

  2. Two or three signals: fix the workload first. Rebalance work between roles, cross-train, or change shift cover, then watch the heatmap for a month or two.

  3. Zero or one signal: it is a spike. Ride it out, and put the next calendar peak into planning.

What the dashboard cannot decide for you

The template shows the overtime side of the argument: where, when, why and who. It does not hold the other side. Three inputs have to come from elsewhere before you hire anyone:

A fair comparison is the yearly cost of the recurring part of your overtime against the yearly cost of one more person. The dashboard gives you the first number and tells you which role to budget for in the second.

Try it with your own data

The Overtime Spending & Workforce Cost Intelligence template is a single Power BI page: five KPI cards, a monthly cost comparison across two years, the heatmap, the reason donut, the top-employees table, the two bar charts and the Emp. ID filter. It ships as a .pbix file with sample data, which you replace with an export from your payroll or HR system. To feed these views you need, at minimum, one row per overtime record with an employee ID, department, job title, date, hours, reason and hourly rate.

Browse the rest of our Power BI templates, or see the full set of HR dashboards in The 6 HR Dashboards Every Team Needs in Power BI.

How do I know if overtime is a staffing problem?

Check whether it is constant, concentrated and recurring. Constant means the same department has overtime in most months, not just one or two. Concentrated means a few roles or people carry most of the hours. Recurring means the reasons are everyday workload, such as call volume or bug fixing, not a one-off deadline. If four or five of the five cuts (department, month, reason, people, cost trend) point the same way, you have a staffing case.

How do I analyze overtime data in Power BI?

Cut the same overtime records five ways: by department and job title (where it concentrates), by month (constant or spiky), by reason (why it happens), by employee (who carries it) and by cost over time. The Overtime Spending & Workforce Cost Intelligence template puts all five on one page, and because the charts cross-filter, clicking a department or a reason re-draws everything else.

Is overtime cheaper than hiring more staff?

It depends on how much of the overtime is recurring. Compare the yearly cost of the recurring part of your overtime with the full yearly cost of one more person, including salary, benefits, recruiting and training. The dashboard gives you the first number and shows which role to budget for in the second; finance supplies the cost of the hire.

What causes excessive overtime?

Common causes are recurring workload that the team cannot absorb (high call volume, bug fixing), predictable calendar peaks (a year-end report, a project launch), emergencies (breakdowns), and uneven work distribution that leaves a few people carrying most of the hours. Which one you have decides the fix: hire, plan, repair the cause upstream, or rebalance.

What data do I need for an overtime dashboard?

One row per overtime record is enough: an employee ID, department, job title, date, overtime hours, a reason and an hourly rate. Those fields feed the views in the template: heatmap by department and month, hours by reason, hours and records by department and job title, the top-employees table and the cost trend.

Can I filter the overtime dashboard by department or manager?

Yes for departments: click a department bar, a job title or a reason and the whole page filters to it. The template also has an employee ID filter. If you want a slicer for manager names or another field, you can add one from the Power BI field list.

All Power BI HR Dashboard Templates