Is This Uber Ride Worth It? The 20-Second Math I Run on Every Offer

Last Tuesday I was parked outside a coffee shop on Ninth Street in Durham when a request came in. $19.80. On the card that looks like a good ride. I let it time out.

Two blocks later a $11.40 request came in and I took it.

That sounds backwards, and for about six months of driving it would have been backwards for me too. I took the big numbers and skipped the small ones, and at the end of the week I could never work out why the bank balance did not match the effort. The problem was never the dollar figure on the card. It was everything the card does not show you.

Here is the arithmetic I actually run now, why those two offers came out the way they did, and how to make the same call in the two or three seconds you get before the request disappears.

The offer card is a headline, not a number

An Uber offer shows you a payout, a trip distance, a trip time, and a pickup distance. Four numbers. Three of them are incomplete and one of them is actively misleading if you read it the way it is presented.

The payout is gross. Nothing has come out of it yet. Not fuel, not tyres, not the fraction of your car's value that this ride consumes.

The trip distance is only the paid leg. The miles you drive to reach the rider are real miles on a real car and they are not in that figure.

The trip time is the same story. Your shift does not start when the passenger gets in.

So before anything else, the offer needs two corrections: add the pickup, and subtract the cost.

Step one: know your cost per mile

This is the number most drivers have never calculated, and it is the one that decides everything downstream.

DriverSignal's default is $0.26 per mile. That is not a figure I invented to look conservative, it is built from two parts:

| Component | How it is derived | Per mile | |---|---|---| | Fuel | 28 MPG at $3.20 a gallon | ~$0.11 | | Wear, tyres, brakes, oil, depreciation | remainder of the total | ~$0.15 | | Total | | $0.26 |

The fuel half is easy to check against your own car and your own pump price. Swap 28 MPG and $3.20 for whatever is true for you and the number moves immediately. The wear half is the one people leave out, and it is more than half the cost. Tyres do not feel like an expense until the day they are. Depreciation never feels like an expense at all, right up until you sell the car.

If you take one thing from this article: every mile you drive with the app open costs you money whether or not a passenger is in the seat. Once you accept that, the rest of the math writes itself.

Step two: count the pickup

The pickup, deadhead, in driver shorthand, is where good-looking offers go to die.

Pickup miles cost you at your full cost per mile and pay you nothing. Pickup minutes stretch your hourly rate without adding a cent to the fare. It hits the numerator and the denominator at the same time, which is why a long pickup can gut an offer without the payout figure changing at all.

I run a veto on this: more than 6.0 miles or more than 12 minutes of pickup and the offer is off the table. Not because the math never works out, sometimes it does, but because a long pickup carries risk the math cannot see. The rider cancels while you are eight minutes out. The trip drops you somewhere you then have to drive back from. Traffic on the way to a pickup is unpaid in a way traffic on a trip is not.

There is one exception I allow, and it is deliberately narrow: if the offer still comes out at 1.5× my floor or better after all the costs and all the pickup time, the ride is strong enough to be worth the risk. A genuinely excellent offer should not be thrown away by a rule of thumb. But it has to be genuinely excellent. A few percent over the floor does not buy an exemption.

Set your own thresholds if 6.0 and 12 are wrong for your city. Just have thresholds.

Step three: the actual formula

Three lines:

Total miles = pickup miles + trip miles Net pay = offer − (total miles × your cost per mile) Effective hourly = net pay ÷ (pickup minutes + trip minutes, in hours)

That last number is the one that matters. Not dollars per mile, not the payout, not how the card feels. Dollars per hour of your life, after costs.

Now back to Tuesday.

The $11.40 ride I took

• Offer: $11.40
• Trip: 5.2 miles, 18 minutes
• Pickup: 3.1 miles, 9 minutes

Total miles: 5.2 + 3.1 = 8.3 miles Cost: 8.3 × $0.26 = $2.16 Net pay: $11.40 − $2.16 = $9.24 Total time: 18 + 9 = 27 minutes = 0.45 hours Effective hourly: $9.24 ÷ 0.45 = $20.53 an hour

Not exciting. But it clears my floor, and the pickup was short enough that I was earning again quickly.

The $19.80 ride I let go

• Offer: $19.80
• Trip: 14.6 miles, 26 minutes
• Pickup: 6.4 miles, 14 minutes

That pickup, 6.4 miles, 14 minutes, trips both halves of my veto before I even get to the arithmetic. But run it anyway, because the result is instructive:

Total miles: 14.6 + 6.4 = 21.0 miles Cost: 21.0 × $0.26 = $5.46 Net pay: $19.80 − $5.46 = $14.34 Total time: 26 + 14 = 40 minutes = 0.667 hours Effective hourly: $14.34 ÷ 0.667 = $21.51 an hour

Here is the honest part: on hourly rate alone, the big ride edges out the small one. $21.51 versus $20.53.

But $21.51 against a $20 floor is 1.08×, nowhere near the 1.5× that would earn a pickup-veto exemption. So the veto stands, and I let it go. Forty minutes of commitment, 21 miles on the odometer, and a drop-off well outside the pocket I was working. Versus 27 minutes and 8.3 miles that left me where the requests are.

This is the whole point. The $19.80 offer was not a bad offer. It was a differently shaped offer, and the card gave me no way to see that in three seconds.

Step four: pick a floor and hold it

A floor is the hourly number below which you do not accept. Mine started at $20 an hour, net, after costs.

I picked $20 because it was a round number I could hold in my head at a red light, not because it was optimal. That is fine. A floor you actually apply beats a perfect floor you re-litigate on every ping.

The problem with a fixed floor is that it knows nothing about your market. $20 might be leaving money on the table on a Friday night and might be unreachable on a wet Tuesday afternoon. So after enough offers have gone past, a better floor is one learned from your own stream: take the 60th percentile of the offers you have actually been scored on, meaning you accept the better-than-typical 40% and pass on the rest, and clamp it into a sane $12 to $35 an hour band so one weird night cannot drag it somewhere silly.

That needs data to be meaningful. It takes about 50 scored offers before a learned floor beats a number you picked yourself. Until then, use the fixed one. There is more on how this fits together on the DriverSignal home page.

About declining: it is not free, and it is not fatal

I want to be straight about this, because there is a lot of confident nonsense on both sides.

Passing on offers protects your take-home. That is the entire point of having a floor. But it does have a price, and pretending otherwise would be doing you the same disservice the offer card does.

A low acceptance rate can cost you Uber Pro tier perks. Turn down several requests in a row and you may see a brief cooldown before new ones come through. Those are real costs and you should weigh them, especially if you are chasing a quest or a tier threshold, a $2-per-ride bonus genuinely changes which offers clear your floor, and it should be in the math rather than in the back of your mind.

What passing on offers does not do is put your account at risk. Skipping protects your take-home; it may cost you Pro-tier perks or a short cooldown, not your account.

Doing this in the three seconds you actually get

Everything above is fine on paper. In the car, with a request on screen and a countdown running, nobody is dividing 9.24 by 0.45.

What worked for me was practising the shape of it rather than the arithmetic. Roughly: a quarter a mile, all the miles, all the minutes. After a few hundred offers you stop calculating and start recognising, this one is fine, this one is a trap, this one is close.

That recognition is exactly what I built DriverSignal to do faster than I can. It reads the offer already on your screen, subtracts your cost per mile, adds the pickup you were about to forget, and shows the net and the effective hourly before the card times out. It runs on the device, it does not touch your driver account, and it does not accept or decline anything, the call stays yours. You can see what it costs or read the rest of the offer-math articles.

But the tool is downstream of the habit. Even if you never install anything, do this: work out your cost per mile this week, pick an hourly floor, and start counting the pickup. Those three moves change more than any app will.

The dollar figure on the card is the last thing you should be reading.