Payroll data usually tells you what happened.
It shows what instructors were paid, which classes ran, how many people attended, and how much labor cost changed from one period to the next.
What it does not always tell you is why.
That is where AI-assisted analysis becomes useful.
By exporting your Jebra Payroll Detail as a CSV and uploading it to Claude, you can ask targeted questions about attendance, instructor pay, class revenue, scheduling, capacity, and profitability.
The goal is not to replace your judgment. It is to help you see patterns buried inside the data faster.
Before asking Claude to analyze anything, use this instruction:
First check whether any pay periods in this export overlap, and deduplicate repeated shifts before analyzing.
This step matters more than it may appear.
During testing, two out of three payroll exports contained overlapping periods that silently inflated the results by approximately 23% to 25%. Without removing those duplicates, the analysis could make payroll, attendance, and class performance look very different from reality.
Once the data is clean, these ten prompts can help turn payroll records into practical operating decisions.
Prompt:
For each class type, work out how many check-ins it needs to cover the instructor's pay, using pay per session and revenue per check-in. Compare that to actual average attendance. Which classes are chronically below break-even, and by how many people? First deduplicate any shifts that appear in overlapping pay periods.
This analysis shows the minimum attendance required for each class to cover instructor pay.
Instead of simply labeling a class as "good" or "bad," you can see whether it needs one more attendee, several more attendees, a schedule change, or removal from the calendar.
A class that consistently averages three people but needs four to break even has a very different problem from a class that needs ten and averages four.
Prompt:
For each class type, calculate revenue per check-in and instructor pay per check-in, and the gap between them. Rank classes by that gap. Is my profitability problem pricing, or attendance? Tell me which classes to fill and which to reprice or cut.
This separates pricing problems from attendance problems.
A studio may assume that prices are too low when the real issue is that instructor cost is being spread across too few attendees.
On the other hand, a consistently full class with very little contribution per attendee may point toward pricing, package structure, or instructor compensation.
This prompt helps prevent the common reflex of changing prices before understanding the actual cause.
Prompt:
Calculate instructor pay as a percentage of estimated revenue for each class type. Flag anything over 100% - classes where I pay out more than they bring in - and show what the best performers look like by comparison.
This creates a straightforward view of how much class revenue is being consumed by instructor pay.
Anything above 100% means the class is paying out more in instructor compensation than it is estimated to generate.
The comparison with stronger formats is especially useful. Instead of looking at weak classes in isolation, you can see what healthy payroll-to-revenue ratios look like inside your own business.
Prompt:
Where two or more instructors teach the same class type, compare their fill rates and revenue per session. Important: check whether they teach different time slots before crediting the difference to the instructor - only compare like-for-like slots, and tell me when the data can't separate the two.
Raw instructor rankings can be misleading.
An instructor teaching weekday evenings may appear to outperform someone teaching early Friday mornings, even when the schedule is doing most of the work.
This prompt asks Claude to control for time-slot differences before assigning performance differences to the instructor.
That makes the analysis more useful and much fairer.
Prompt:
Build a fill-rate table by day of week and time of day, weighted by sessions per slot - and recompute fill as check-ins divided by capacity rather than trusting any occupancy field. What share of my schedule sits in below-average slots, and where is the biggest supply-demand mismatch?
This prompt helps answer a deceptively simple question:
Are you scheduling classes when people actually want to attend?
It recalculates fill rate using actual check-ins divided by capacity rather than relying on an occupancy field that may be incomplete or misleading.
The result can reveal situations where a studio is running many sessions in weak time slots while offering too few classes during stronger periods.
Prompt:
Audit every session: pay minus estimated revenue. What share of my sessions lose money, and how much? Then find the pattern - is it a class, an instructor, a time slot, or a combination? Test the combinations before blaming any one factor, and tell me the single change that eliminates the most losses.
This is one of the most operationally useful prompts in the set.
It looks beyond broad class averages and evaluates each session individually.
More importantly, it asks Claude to test multiple explanations before blaming a class type or instructor.
A losing pattern may come from one specific class at one specific time, while the same instructor performs well elsewhere.
That distinction matters because the correct response may be to move one class, not overhaul the entire schedule.
Prompt:
How often do classes sell out or reach 80% of capacity? Compare each class's stated capacity to its 90th-percentile attendance. Which capacity numbers are fiction - and do any sell-outs suggest demand for an extra session or a price rise?
A class may technically have space for 20 people, but that does not mean 20 is a realistic operating capacity.
This prompt compares stated capacity against actual attendance patterns.
It can reveal:
The result is a more realistic picture of what "full" means inside each format.
Prompt:
Compare my total payroll period over period. Split each change into two parts: did I run more or fewer classes, and did the pay per class change? If pay per class rose because classes got fuller and instructors hit higher attendance tiers, say so - that's growth, not overspend.
An increase in payroll is not automatically bad.
Payroll may rise because:
This prompt separates those causes.
That distinction helps studio owners avoid treating healthy payroll growth as a cost-control failure.
Prompt:
Show average fill rate month by month, and revenue per class-hour against pay per class-hour. Is occupancy trending up, and is the trend big enough to change which classes look viable?
A class that appeared weak three months ago may no longer be weak.
This prompt tracks fill rate, revenue, and instructor pay over time to show whether the studio is gaining momentum.
It also helps identify classes that are moving toward viability, rather than evaluating every format using one frozen snapshot.
Prompt:
If I paid a per-head bonus for every check-in above break-even, what would it have cost last period, and what share of class margin is that? If my pay rates or bonus column already vary with attendance, tell me instead how much of pay is attendance-linked and whether it correlates with instructor fill rates.
Attendance bonuses can sound attractive, but they need to be tested against margin.
This prompt estimates:
That gives studio owners a clearer basis for designing incentives instead of relying on guesswork.
That is the point.
In Jebra's testing, the same prompt set produced very different findings across three studios:
The export format was the same. The questions were the same. The answers reflected the reality of each studio.
AI analysis becomes most useful when it is grounded in your own operational data.
It does not give every studio the same playbook. It helps each studio understand its own.
At the end of any prompt, add:
Show your working and give me the top three actions.
That small addition asks Claude to provide both the analysis and a practical next-step list.
Instead of receiving only a conclusion, you get the reasoning behind it and a clearer path forward.
Jebra Payroll Detail already contains more than payroll history.
It contains signals about attendance, scheduling, instructor compensation, class profitability, demand, capacity, and growth.
These prompts help bring those signals to the surface.
Export the data. Check for duplicates. Ask better questions.
Then use the answers to make better studio decisions.
Export your Jebra Payroll Detail as a CSV and use the full set of Claude prompts to explore class profitability, attendance, scheduling, payroll changes, and growth.