Gross, then the carve-outs
The shift starts with average earnings per trip times trips - the number the app congratulates you on. Twenty trips at $12 is $240 of gross, and everything after that is a subtraction rather than a bonus.
The platform commission comes off first because it is a percentage of that gross; the field accepts 0%-50% and is labeled as the app’s share, so a fee built into your per-trip figure should be entered as zero rather than counted twice. At the default 15%, $36 leaves the $240 before the car is even considered.
Then the vehicle cost: miles driven multiplied by your cost-per-mile rate, so 80 miles at 65 cents is $52. Other shift costs - parking, tolls, delivery bags, an apportioned slice of insurance or a phone plan - come off last, because they do not move with the trip count but still come out of the same shift. What remains is the net for the shift, and the page prints each line so the difference between "earned" and "kept" cannot hide.
The cost per mile is the honest guess
Fuel is real but only part of it: tires, oil, maintenance and depreciation all accumulate per mile, and nobody tracks them to the cent. A defensible all-in range for a typical car is roughly 50 to 80 cents a mile, which is why the field holds a rate you set rather than a rate the page asserts.
Raising that rate by a dime moves the answer more than most drivers expect. At the defaults, 80 miles at 65 cents costs $52 and leaves $19 an hour; at 75 cents the same shift leaves $18, and at 50 cents it leaves $20.50. That spread is the point of the page: it is calibrated to show how sensitive the number is, not to hand you one warm figure.
The mileage field is meant to hold everything the shift put on the odometer for the platform - including the miles driven between trips and while waiting. Dead miles are among the largest hidden costs in delivery and rideshare work, and leaving them out is the easiest way to make an unprofitable shift look acceptable.
Waiting is work
The hours field runs from the first fare to the last and is deliberately the whole shift, with a floor of half an hour and a ceiling of 24. If two of those hours are spent waiting between trips, the two hours belong in the denominator, because that is the time the work actually took.
A per-active-trip figure silently drops the waiting and flatters every gig; a per-hour figure that includes it is the only one that survives contact with a real day. The hero number is net across the whole shift divided by those hours, and the line underneath restates the hours the answer implies, so a shift that paid $152 over eight hours reads as $19 an hour rather than as $7.60 a delivery.
The guard rails match the arithmetic. Trips are held between 1 and 200, miles between 0 and 1,000, and a shift with no trips, no hours or no positive per-trip earnings returns a warning instead of a per-hour number, because dividing by a zero-hour shift is not an answer this page can print.
What it does not model
The net is pre-tax and rough. Self-employment tax, insurance, and precise vehicle depreciation are real and often larger than new drivers expect, and they vary too much by market and vehicle to fold in honestly. The page prints the pre-tax net and says so rather than pretending the figure is take-home pay.
The cost-per-mile rate is the largest honest guess on the page, and the fixed-cost field is where a driver puts the costs that do not scale with miles. Neither turns this into a tax calculation, and neither captures depreciation precisely enough to price a car’s replacement.
Read the result as a physical: what the shift actually paid per hour after the platform, the car and the waiting. If the number looks disappointing, that is the arithmetic working, and it is information worth acting on rather than a flaw in the model.
Danfio