A laundromat app can run the whole business except the part everyone asks for first.

What a phone can read about a machine, and what it cannot

Start with the hard limit, because everything else follows from it. A phone has a camera, a microphone, radios for Wi-Fi and Bluetooth, and a location sensor. None of those can tell it whether washer seven is running, how many minutes are left, or whether somebody has already emptied it. A laundromat machine status app never reads machines. It reads a feed that networked machines produce, and where there is no feed there is nothing to read.

It helps to see what that feed costs in physical terms. Dexter's own DexterLive networking manual lists a location kit built around a gateway and a paired router. Then a washer kit: a sixteen port Ethernet switch and Cat5e cables, enough for ten washers. Then a matching dryer kit for ten dryers. Each cable is routed through the back of a machine to a connector on its control board, one machine at a time. The manual says an average location would ideally have at least 15 Mbps download and 5 Mbps upload. That is an installation across the whole floor, not a feature you add to an app.

So there are two honest positions, and a plan should pick one out loud. Either the machines are networked, in which case ask the supplier in writing whether you may read that status into your own app, and on what terms. A portal you can log into is not the same thing as a feed you can query. Or the machines are not networked, in which case the app says nothing about machine state and does the other useful things instead.

There is a cheap middle worth knowing about. A numbered QR sticker on every machine costs almost nothing. It gives the app one fact it can trust: which machine the customer is standing at. From there a customer can start their own countdown, report a fault against a real machine number, or pay for that specific unit. The countdown is still a guess, and the screen should say so rather than dress a guess up as a reading.

Dexter Laundry, DexterLive networking installation manual (part 8507-475-002)

Telling a customer the cycle finished

Assume you do have a status feed. The next question is how the customer hears about it, and this is where laundry apps quietly overpromise. A push notification is a request to Apple or Google to deliver a message to a device. It is not a guaranteed delivery and it is not a phone call.

Apple's own documentation is plain about the model. APNs makes every effort to deliver notifications. It can store a notification for a device that is currently offline, then forward it when the device comes back online. Where it cannot deliver immediately, it may coalesce notifications for the same app. For a laundromat that means a dryer finished alert can arrive late, and that two alerts can turn into one. The customer may also have declined notification permission, in which case nothing arrives at all.

Design around that rather than against it. Show the true state of the customer's machine on screen every time the app opens. The alert is then a convenience, not the only path to the fact. Do not build a rule that depends on the alert landing, such as holding a finished machine for ten minutes. Not unless somebody on site can override it. Permission, device tokens and delivery are a subject of their own. If the headline promise of the app is telling someone the cycle finished, read that part properly first.

Apple, setting up a remote notification server

What a wash and fold order has to hold

The wash and fold side is where an app earns its keep, because the paper ticket is where the arguments come from. A wash and fold app is really one record with a life: dropped off, weighed, priced, washed, folded, on the rack, collected. Every stage has a time and a person attached to it, or the record is just a mood.

The fields that matter are the ones people leave out. Weight at intake, because the price usually depends on it and the customer should see the number that produced their total. Instructions in the customer's own words rather than a dropdown: no heat, separate the whites, the stain is on the left cuff. A ticket number that also lives on the bag. Where the finished order is sitting, because shelf 4B is the difference between a ten second handover and five minutes of hunting. Who released it, to whom, and when.

Two events need a plan before launch rather than after. An order nobody collects, which needs a stated rule and a notice instead of a growing pile in the corner. And an item that goes missing or comes back damaged. That needs a record starting at intake, because the argument is always about what was in the bag. Photograph anything already marked before it goes in the machine. If you also collect and return orders, the routing and handover part is a different shape of app. It is closer to a local delivery app than to anything in the store.

Prices, balances and the regulars you never record

A coin laundromat knows nothing about its customers. Money comes in and no name is attached to any of it. The first commercial reason to build anything is that an app turns anonymous traffic into a list. A list you can tell about a broken dryer, a price change, or a quiet Tuesday morning.

Balances are the part to get right from day one. If customers top up a balance in the app, that money is a liability rather than revenue on the day it lands, because it is service you still owe. Keep the balance on the server and never on the device. Record every top up and every spend as its own row, so the total can always be rebuilt. Ask your accountant how unused balances should be treated in your books. The rewards layer on top, a free dry after ten washes or a discount before noon, is the same machinery as any loyalty card app.

Pricing rules are usually where an off the shelf product stops fitting. A student rate after four on weekdays. A different wash and fold price for bedding. A standing account for the barber shop next door that gets invoiced monthly instead of paying per load. None of those are exotic. They are simply not on the list of options somebody else built for a different store.