Experience a no pressure, live demo of our software.   Talk with our sales team

Petrosoft / Field Notes
Case file 026 Fuel margin / AI prompts / 5 min read

Case Solved: Gallons were up, the bank account was down.

The prompt (and the data) that found the missing 8 cents a gallon.

SOLVED
Case file: solved
Owner
Jay, one independent gas station with a small c-store in western Pennsylvania
The problem
Gallons up about 3% this year. Money in the bank down.
What didn’t work
Generic AI questions with no data, and, before 2026, months of Excel that never reconciled
What worked
One structured prompt plus clean daily data exported from CStoreOffice®
The finding
Card fees grew from 5.8 to 8.1 cents a gallon as prices climbed: about $73,600 a year at today’s prices.
The number that mattered: net fuel margin per gallon, after card fees.
+2.3 cents/gal

This case is a composite of situations independent station owners commonly face. Jay’s store numbers are illustrative; fuel prices and industry figures come from the sources listed below.

01 / The problem

A good year on paper

Jay runs one unbranded station with eight fueling positions and a small store. He pumps about 2,500 gallons a day and processes cards through his own merchant processor. Every morning he sets his price from his supplier’s rack price and the street price at the chain store across the intersection, aiming for a 30-cent posted margin.

By October, 2026 looked fine on paper. Gallons were up about 3%, and his posted margin had held near 30 cents all year. But deposits kept coming in lighter than he expected, and his operating account was noticeably thinner than a year earlier.

02 / Dead ends

Two things that didn’t work

In 2025, on his old system, Jay spent two months of evenings in Excel. The processor batched by settlement date, the POS by business day, invoices by delivery and tank readings lived on a gauge printout. Nothing lined up, so he never trusted the totals. That’s a big reason he moved his back office to CStoreOffice® at the start of 2026.

This fall, he tried asking an AI assistant instead:

Why is my gas station making less money when sales are up?
A list of possibilities: theft, labor, shrink, competition. Nothing tied to his store.
Result: generic
What’s a good fuel margin per gallon?
National ranges. His posted margin was already in range. That was the confusing part.
Result: confusing

The AI wasn’t wrong. It just had nothing to work with. Without his numbers, it could only give him everyone’s answer.

03 / The motive

Why the numbers didn’t add up

Jay’s posted margin was measured in cents per gallon. Card fees are charged as a percentage of the sale. When prices rise, the cents stay the same and the fees go up.

And prices rose a lot. Gas started 2026 below $3 a gallon in most of the country. By November, the Pennsylvania average was about $4.22, and Pittsburgh-area prices had been running about 20 cents above the national average.

$4.22
Pennsylvania average gas price in November 2026, up from under $3 at the start of the year
$21.3B
Record card fees paid by convenience retailers in 2025 (NACS)
3x
Growth in swipe fees since 2009, now most retailers’ biggest operating cost after labor (NRF)
04 / The evidence

The data the prompt needed

His second attempt started with the data, not the question. He needed two files:

File 1
Daily since January 1
gallons sold avg retail price delivered cost/gal fuel sales paid by card book vs. tank variance

Exported from CStoreOffice®, where sales, EDI fuel invoices and ATG-integrated wet-stock reconciliation were already lined up by business day.

File 2
Monthly
total card sales total processing fees

Copied from his processor statements. That’s ten rows, ten minutes.

For Petrosoft customers, File 1 is a standard CStoreOffice® report. He removed account numbers and other sensitive details before uploading.

05 / The breakthrough

The prompt that cracked it

The prompt
I own [number] gas station(s). I’m attaching two files. File 1 has daily data from [start date] to today: date, gallons sold, average retail price per gallon, delivered cost per gallon, fuel sales paid by card, and book vs. tank variance in gallons. File 2 has my monthly card processor totals: month, total card sales, total processing fees. Use code to do the math. (1) Calculate my effective card rate by month. (2) Calculate my net fuel margin per gallon after card fees and tank variance, by week. (3) Show how much of the change in net margin comes from price changes versus margin changes. (4) Give me a formula for the minimum daily retail price that keeps a [X]-cent net margin. Show every formula and flag any weeks below [Y] cents.

Jay filled the blanks with one station, January 1, a 20-cent target and a 15-cent warning line.

Why it works: it names one metric, uses real numbers, makes the AI run the math in code, separates the price effect from the margin effect and asks for formulas he can check in Excel. Use an AI tool that accepts uploads and runs code, and spot-check a few rows by hand.

06 / The verdict

What the data said

Where a posted margin goes, in cents per gallon
Net margin kept Card fees
January, about $3.00 gas, 30c posted24c kept
24.2
5.8
November, about $4.20 gas, 30c posted22c kept
21.9
8.1
Two flagged weeks in October, matched down to 18c postedabout 10c kept
~10
8.1
12c given up to match
Effective card rate
2.4%
Steady all year, on about 80% of fuel sales
Card fees per day
$144 to $202
January vs. November, at 2,500 gallons a day
Extra fees from higher prices
$21,000
A year. His posted margin barely moved.

Tank variance was normal, so no leak or theft. The gap was card fees growing with the price while his margin target stayed in cents. The two flagged weeks were when the chain across the intersection cut its price and Jay matched down to 18 cents posted. At $4.20 gas, that netted him about 10 cents a gallon, half of what he thought he was protecting.

07 / Case closed

What Jay changed

1

He prices from net margin, not posted margin

The AI’s floor formula
floor price = (delivered cost + target net margin) / (1 – effective card rate x card share)
($3.90 + $0.20) / (1 – 0.024 x 0.80) = about $4.18

With a $3.90 delivered cost and a 20-cent net target, that’s about $4.18. Now he knows exactly how far he can match before he’s working for free.

How Petrosoft helps
Retail360 puts fuel performance and forecasting on Jay’s phone, so he can check his floor before reacting to the station across the street. CStoreOffice® updates delivered cost from EDI fuel invoices, so the floor uses real costs.
2

He reruns the check every month

The CStoreOffice® export comes out the same way every time, so he adds one month of processor totals and reruns the prompt. Minutes, not weekends.

3

He looked at his card mix

With a clear effective rate, he could compare processor pricing and see how much volume came from higher-cost cards. Thinking about separate cash and credit prices? Check your card agreements and state rules first.

4

He leaned on the store

Per NACS, fuel was 65% of c-store sales in 2025 but only 38.8% of gross profit, and 62% of customers went inside on their last fill-up. With fuel tighter than he thought, inside sales became part of the plan.

08 / Questions

Frequently asked questions

How do you calculate net fuel margin per gallon?+

Take retail price minus delivered cost per gallon, then subtract card fees per gallon and any inventory variance. Card fees per gallon are roughly retail price x effective card rate x share of fuel sales paid by card.

Why do credit card fees go up when gas prices rise?+

Card fees are charged as a percentage of the sale. If your margin stays the same in cents per gallon but the price rises, the fee per gallon rises with it. In this example, moving from $3.00 to $4.20 added about 2.3 cents a gallon in fees.

Can I trust AI to analyze my gas station numbers?+

Yes, if you give it clean, reconciled data, ask it to run the math in code, require it to show its formulas, and spot-check a few rows yourself. Remove account numbers and other sensitive details before uploading.

See the data behind Jay’s prompt in CStoreOffice® and Retail360.
Book a live, no-pressure demo
Sources
  1. NACS, “U.S. Convenience In-Store Sales Top $340 Billion” (State of the Industry, April 2026)
  2. NACS Magazine, “Who’s Buying Your Gas?” (2026 NACS Consumer Survey, April 2026)
  3. NRF, Swipe Fees policy page (2025 data)
  4. OilPriceAPI, Pennsylvania/Pittsburgh regional average gasoline price (updated November 16, 2026)
  5. “Gas in Pittsburgh about 20 cents above national average” (August 4, 2026)