Fifteen dollars, 38 items, member 8.6 miles from the store. Estimated tip of $9 sitting right under the pay.
Took it. Got nothing.
That tip figure on your offer screen is doing something more complicated than most shoppers realize, and once you know what it is, you stop reading it as a promise.
Two different numbers wearing the same outfit
Shipt's own help pages spell this out and almost nobody reads them.
If a member adds a tip at checkout, that pre-tip gets shown to shoppers before the order is claimed. Real money, already attached, entered by a person who decided to do it.
If they don't pre-tip, you may still see a tip figure. That one is an estimate Shipt builds for you, and their help page lists the ingredients: the member's previous tipping history, their order history, the total items, and shopping trends.
Same spot on the screen. Completely different meaning.
One of them is a commitment. The other is a prediction about a stranger, generated by a model, from records you aren't allowed to look at.
The estimate is a guess about history you can't see
Read that ingredient list again. Previous tipping history. Order history.
Shipt is telling you in writing that they hold a tipping record on this member. They compress it into one number and hand you the number.
You get the compression. They keep the record.
And an estimate carries no floor. Offer pay does. Shipt's explainer on shopper pay says they moved from showing a range to showing a single figure, and that the amount you see is the minimum you'll be paid for finishing the order. Nothing like that exists for the tip. The estimate can read $12 and the tip can come in at zero, and nobody broke a rule.
So the guaranteed part of the offer is usually the smaller part.
A pre-tip is better and still not banked
Better because a human did it on purpose.
Not banked because tips move after the fact. Shipt lets members add a tip at any point after delivery, and once a tip goes in there's a limited stretch where they can still change it. A $10 pre-tip can turn into $25 after you text a photo of the empty yogurt shelf and haul six bags up to a landing. It can also shrink.
Mostly it climbs. I've had pre-tips double. Twice in four years I've watched one get cut.
Since the shopper groups get this backwards constantly: a pre-tip does not reduce what Shipt pays you. Their help page on early tips says a tip is in addition to the pay you already receive for the order, and that shoppers get 100% of it. Nobody is quietly docking your offer because somebody was generous at checkout.
Signals that don't work
Everybody's got a theory. Most of them are junk.
Big house, big tip. Nope. Two of my most reliable no-tippers live in places I could not afford on my best year.
Apartments tip worse. Not in my zone. My single best tipper for two years straight was a third floor walkup who left $20 every time and apologized for the stairs.
Bigger order, bigger tip. Sometimes this one runs exactly backwards. A 62 item haul from someone who tips 5% pays worse per hour than an 11 item run from someone who hands you $15 at the door.
Store, neighborhood, delivery window, whatever else people swear by. The store doesn't tip you. A person does.
One soft signal I'll defend: whether they pre-tipped at all. A member who reached for the tip field before you'd done a single thing has told you something honest about how they think about the job.
What actually predicts a tip
The address does.
Tipping is a habit, not a mood. Somebody who tipped $12, $15, and $11 across three orders will land in that band the fourth time. Somebody who's stiffed you four times has a policy, and you're not talking them out of it on their porch.
Three visits is enough. One is noise. Two is coincidence. Three is a customer.
Which becomes usable a few hundred orders in, when you're delivering to the same doors over and over. Same complexes, same cul-de-sacs, same guy who orders La Croix every Thursday. Shipt's estimate compresses a member's whole tipping record down to a digit or two. Yours isn't compressed at all: every visit, every amount, sitting in one place and belonging to you.
The same order, three ways
Back to the one I opened with. 38 items, $15 offer pay, 8.6 miles out, late July.
Shopping ran 48 minutes with two substitution texts. Fifteen minutes to the house, five at the door, fifteen back toward the store. Total 1 hour 23 minutes, 17.2 miles round trip, because the drive home is the same distance as the drive out.
At the IRS business rate of 76 cents a mile, that driving cost $13.07. Gas, tires, brakes, the resale value of the vehicle.
No tip: $15 less $13.07 leaves $1.93. That's $1.39 an hour.
Estimate turns out right, $9 tip: $10.93 after the drive. $7.90 an hour.
Address history said $18, $22, and $16, and they came in at $19: $20.93 after the drive. $15.13 an hour.
Same 38 items every time. Same 17.2 miles, same 83 minutes, same conversation about whether they wanted the organic bananas. The gap between $1.39 an hour and $15.13 an hour was information, and I didn't have it.
What to do in the twenty seconds you get
Practical version, since that's all you have before the offer is gone.
A pre-tip counts. Treat it as a floor rather than a ceiling, because it's more likely to climb than fall.
An estimate with no pre-tip behind it should be treated as zero on anything past about eight miles. Then ask whether the offer pay by itself covers the drive. Run the mileage rate against the round trip once and you'll find it often doesn't.
Any address you've been to before outranks everything else on the screen. Not close.
The catch on that last one is that it only exists if somebody recorded it, and the tip usually lands hours after you've left, when you're four orders down the road and no longer thinking about the yogurt shelf. Which is why most shoppers have a strong opinion about their good tippers and no actual record of them.
Auto Tip Map is an Android app that saves every order, tip, and address automatically, including tips that arrive hours after delivery, and puts an address's full history on the offer screen while the timer is still running.