Find OpEx drivers and where spend is accelerating
Breaks operating expenses into categories over three months and separates one-time step-ups from spend that is genuinely accelerating.
Copy and customize
You are a controller analyzing operating expense trends.
Context: expenses are in the Amount field, the category is {category_field}, and
a secondary driver is {driver_field} (e.g. department or cost center). The
reporting date is Reporting Month. Use the last three months.
{opex_data}
Task: show OpEx by category over the three months, identify the biggest drivers,
and find where spend is accelerating rather than simply rising once.
Output format:
- A table: category, the three monthly figures, and month-over-month change for
the latest two months.
- The categories whose month-over-month growth is itself increasing, labeled as
accelerating, kept separate from categories that stepped up once.
- For the top accelerating category, a breakdown by {driver_field}, with the
share of rows covered if some are unattributed.
Run it in four steps
- Export at least three consecutive months of OpEx from the Amount field, with a controlled category column and a department or cost-center column.
- Paste it into
{opex_data}, and point{category_field}and{driver_field}at the right columns. - Run it to find the drivers and flag where growth is accelerating.
- Read the coverage figure on the driver cut; if much of a category is untagged, the concentration claim covers only the tagged portion.
When to reach for this prompt
Use this when OpEx is creeping and you want the cause early, before it becomes a budget conversation. Three months is enough to tell a trend from a blip without dragging in stale history. It needs a controlled category list rather than free-text memos, and it needs at least three months, since acceleration cannot be seen in two points.
What you can expect back
OpEx by category — last 3 months (Amount)
| Category | M1 | M2 | M3 | MoM M2 | MoM M3 |
|---|---|---|---|---|---|
| Headcount | 720K | 742K | 781K | +3.1% | +5.3% |
| Software / SaaS | 138K | 151K | 172K | +9.4% | +13.9% |
| Marketing programs | 190K | 186K | 181K | -2.1% | -2.7% |
| Facilities | 62K | 62K | 63K | 0.0% | +1.6% |
| Total | 1,110K | 1,141K | 1,197K | +2.8% | +4.9% |
Accelerating (growth rate rising): Software/SaaS, where month-over-month growth moved from 9.4% to 13.9%, and Headcount, from 3.1% to 5.3%. Marketing is declining rather than accelerating.
Software/SaaS by department (M3 increase): Engineering accounts for roughly 64% of the increase, suggesting it is concentrated rather than broad-based.
This prompt has real limitations you should understand.
"Acceleration" sounds like a measured quantity, but over three data points it is a fragile one. The label rests on two month-over-month deltas, and one more month can flip it entirely, so accelerating categories are candidates to watch rather than conclusions. The prompt will still name them with conviction.
The category cut underneath has its own trap. If a cost was recategorized mid-window, that category appears to accelerate simply because rows arrived in it, not because spend grew. The secondary driver breakdown is meant to localize the cause, but it only works where the driver field is populated; when 40% of a category carries no department tag, a claim like "Engineering is 64% of the increase" is 64% of the attributed portion, which is a materially different statement than it appears.
Trustworthy driver analysis needs every expense row tagged to a stable category and a consistently populated cost-center dimension, held constant across the window so that movement reflects spend rather than reclassification. That tagging discipline is a property of how the data is structured upstream, and it is the difference between an acceleration signal you can act on and one that is an artifact of coding.
What your data needs to look like
- An Amount field at monthly grain across at least three consecutive months
- A controlled OpEx category field, not free text, stable across the window
- A secondary driver dimension (department or cost center) populated on most rows
- No mid-window recategorization of existing costs
See how FinanceOS handles this prompt on real financial data.
Book a 20-minute walkthrough. We’ll run this exact prompt against a sample dataset reconciled through FinanceOS, and show you what changes when the data underneath is right.
Book a walkthrough