For years, unattended micro markets — self-serve shops that replace traditional vending machines — were something you could only get if you were big enough. Dental practices, boutique offices, fitness studios, co-working spaces and small hotels all heard a version of the same answer: come back when you have more people.
That answer was never really about technology. It was about route economics. An operator has to send a truck, and a truck has to be worth sending.
What has changed is that you no longer need an operator to run a self-serve shelf. If you're willing to own the stock and the restocking yourself, the placement threshold that kept you out stops applying — because nobody has to justify a route stop.
Why operators say no to small locations
This is worth understanding properly, because it explains the whole situation and it isn't arbitrary.
VMFS USA's 2026 micro market operator guide sets out three bands by the number of people at the location. Under 150 people, it is blunt: "the service economics of a micro market do not justify daily visits for this revenue volume." Between 150 and 300 it calls the case borderline and worth evaluating carefully. Over 300 people in a secure setting, it describes a clear financial case. Vending reaches lower — VendSoft's 2026 profitable-locations guide characterises high-traffic vending sites as ones seeing 50 or more people daily.
Those are rational positions for a business with a truck and a route. They are also the reason a twelve-person office, a yoga studio with a few dozen members, or a dental practice cannot get a market placed at all, no matter how good a location it would be.
The threshold is about the operator's costs, not yours. Remove the operator and you remove the threshold.
What actually changes when you run it yourself
Not magic, and not a cost saving we're going to invent a number for. Four structural differences:
- Hardware is generic. An ordinary Android tablet on a stand, rather than a purpose-built terminal. If you have a spare tablet, that line is zero. There is no iOS build — TallyTill runs on Android and in Chrome, so an old iPad won't run the app, though a browser on a laptop will. One caveat worth knowing before you plan around it: tap-to-pay on the device itself needs the Android app and an NFC-capable tablet. It is not available on the browser path, where you would use a separate reader instead.
- No route to fund. Restocking is your walk to the storeroom rather than someone's drive across the county. This is the whole ballgame for a small site.
- You own the stock, the pricing and the mix. No planogram to negotiate, no minimum assortment, no waiting on a service visit to change what's on the shelf.
- You own the data. Sales history sits in your own account and exports as CSV, rather than arriving as whatever reporting a contract specifies.
The trade you're making is equally clear: you become the operator. Restocking, spoilage, shrink and pricing are now your problem. If what you actually wanted was for somebody else to handle all of that, an operator contract is the better product and you should keep asking around.
For what the software costs, see the pricing page — there's a free plan with no seat cost and no card required. Machine and kiosk pricing varies enormously by configuration and region, so get quotes rather than trusting a range in an article.
Attendance: who's in the building decides the payment method
This is the detail almost nobody screens for up front, and it's worth two minutes now rather than a surprise later. Card processors classify a self-serve location by whether staff are available — not by whether anyone is watching the tablet. Stripe's own wording:
- Attended — "Attended means staff support is available, either per device or per location."
- Semi-attended, which Stripe folds inside attended — "This includes semi-attended scenarios, where a customer independently interacts with the card reader, but staff may be available in the store or nearby if needed."
- Unattended — "no staff is on location to help the customer if needed."
Stripe supports attended and semi-attended retail with its pre-certified readers and Tap to Pay; it does not certify them for unattended.
Here's why that's rarely a problem for these locations. A micro market lite almost never sits in an empty building. It sits in a dental practice with a front desk, an office with a receptionist, a gym with staff on shift, a co-working space with a community manager, a hotel with a 24-hour desk. What counts is staff being available if a customer needs help — not customers being present. That's semi-attended, and readers and Tap to Pay are fine, even though nobody stands at the market and nobody watches the checkout.
The case to plan for is the stretch when the building genuinely empties of staff — a lobby overnight, an office on a weekend, a 24-hour keyfob gym after the last instructor leaves. Those hours are unattended even if members are wandering in, and what's unavailable then is card-present capture: Tap to Pay and the reader on your tablet, plus offline store-and-forward, which rides on the same reader.
Cards themselves keep working. A QR checkout opens a hosted payment page on the customer's own phone. Card, Apple Pay and Google Pay go through at 2 a.m. in an empty lobby exactly as they do at noon — provided the tablet has a working connection, which is a separate requirement covered below. Cash and check are unaffected too.
So the practical rule is simple: the payment method follows the staffing model. Staffed hours get the full menu. Genuinely unstaffed hours run on QR-to-phone. Most small locations have staff on site nearly all the time their market is used.
Our guide to attended, semi-attended, and unattended Stripe rules walks through the classifications in full and the configuration steps for each.
Connectivity, precisely
Unattended retail has moved heavily cashless — Cantaloupe's 2025 Micropayment Trends Report found 96% of micro-market transactions were cashless in 2024, and that tap-to-pay accounted for 77% of all cashless vending sales that year. So it's worth knowing exactly what needs a signal.
- Barcode and QR scanning run locally on the device and work with no connection at all.
- AI photo recognition needs a connection, because identification happens server-side.
- Offline card sales are possible but genuinely narrow. Stripe store-and-forward requires an external mobile reader over Bluetooth or USB — Tap to Pay has no offline mode whatsoever. It must be switched on for your account in advance with a risk acknowledgement, requires that reader to have connected online at that location beforehand, is capped by default at around $50 per sale and $500 held unforwarded, and leaves you carrying the loss if a card later declines. Treat it as a short outage bridge, not an offline strategy.
- QR checkout is not an offline route. The customer pays on their own phone, but your tablet has to reach the server to create the checkout session and confirm the payment. QR solves the attendance problem; it does not solve the signal problem.
- Cash and check always work offline. Emailed receipts queue until the connection returns.
For an indoor location on building Wi-Fi this is mostly academic. It matters if you're putting a shelf somewhere with genuinely patchy service.
Where this actually works
Consider it when:
- You have a steady, repeating group of people — the same faces most weekdays, rather than passing traffic. Repeat customers learn the shelf; strangers need signage and patience.
- Baskets are big enough to be worth processing. Very small single-item purchases get eaten by fixed per-transaction costs. Check current rates on the pricing page and Stripe's, and do the arithmetic against your own average basket.
- Customers are somewhat captive — office workers, gym members, patients, hotel guests who can't easily leave to buy elsewhere.
- You can genuinely restock on a rhythm. This is the commitment people underestimate. An empty shelf teaches customers to stop looking.
- Extended hours add value — before hours, at lunch, after hours. During the stretches when no staff are around, QR-to-phone carries the card sales.
Skip it when:
- Traffic is sporadic — occasional visitors rather than a regular population.
- Purchases would be mostly cash, mostly tiny. The friction and the margins don't justify a system.
- Restocking isn't realistic — the location is remote, or you don't have reliable supply access.
- Connectivity is genuinely bad. Cash, checks and barcode scanning survive an outage; card payments largely do not, apart from the narrow store-and-forward case above.
What you actually need to start
- A tablet or phone to run the checkout and display your catalog. Bring your own, or buy an Android tablet — with NFC if you want tap-to-pay on the device itself. Prices move constantly; check current models rather than a range in an article.
- Somewhere to put the goods. A cooler, shelving, or a rack. Frequently you already have something that works.
- A card reader, optionally, for the hours your location is staffed. QR checkout needs no hardware at all — just a connection at the tablet.
- Software. See the pricing page.
- Restocking labor. Your time, or somebody's. This is the line people leave off the sheet and it is the one that decides whether the thing survives past month two.
Where it fits, location by location
Small offices. A shelf in a small law firm or accounting office replaces snack runs and vending downtime. With staff in the office, the location is semi-attended, so tap and reader payments are available all day.
Medical and dental practices. Waiting rooms are well suited — patients with time to fill and no easy alternative nearby. Self-checkout means no receptionist disruption, and the front desk is what makes it a semi-attended location.
Fitness studios. A boutique gym or yoga studio can offer post-class refreshments without the cost of a café. Staffed class hours are semi-attended; if you offer 24-hour keyfob access, the unstaffed overnight window runs on QR even though members are present.
Co-working spaces. Low foot traffic per desk is exactly what makes traditional markets uneconomical here, and exactly what stops mattering when there's no route to fund. A community manager on site keeps it semi-attended.
Hotel back-of-house. Staff break rooms in small hotels can stock employee food and drinks. Because a hotel almost always has someone at the desk, it stays semi-attended around the clock.
Workshops and maker spaces. Craft studios with regular members need refreshments but can't support a café.
The numbers to watch
There is no industry benchmark for a shelf like this, and we're not going to invent one. What's useful is tracking your own from day one so you have something to compare against at the end of the first month:
- Transactions per day, and whether the trend is up, flat or down after the novelty wears off.
- Average basket, against the per-transaction cost of processing it.
- Weekly sell-through by item — which things move and which sit. This is the number that tells you what to stock.
- Gross margin after everything, including your restocking time valued honestly.
- Share of sales that are card versus cash, which tells you whether the payment methods you enabled match how people actually want to pay.
The decision rule is simple: if the shelf is covering its costs and your time after four weeks and the trend isn't falling, it's working. If it isn't, you've learned that for the cost of a tablet and some stock rather than a service contract.
Getting started: the test-drive approach
- Pick one location with a regular population and workable restocking access.
- Check the staffing picture — write down the hours staff are on the property and the hours there are none. Five-minute exercise; it tells you which payment methods to switch on. The attendance guide has the configuration steps.
- Set up the hardware — tablet, optional reader, cooler or shelf.
- Stock conservatively — a couple of dozen items, weighted toward things you're confident will move.
- Run it for four weeks and collect the numbers above.
- Do the arithmetic against your real costs, including your own time.
Most locations know within a month. If it works, you've added a revenue stream for very little risk. If it doesn't, the loss is a tablet you can use for something else and some stock you can eat.
The bottom line
The barrier to a self-serve shelf in a small location was never that the technology was too expensive. It was that somebody else's truck had to be worth sending. Once you're the one restocking, that constraint disappears — and with it the reason you kept being told to come back when you were bigger.
Test one location, watch your own numbers, and let them tell you whether to expand.
Sources, retrieved August 2026: VMFS USA, "Micro Markets vs. Vending Machines: 2026 Operator ROI Guide"; VendSoft, "Profitable Vending Machine Locations: 2026 Guide"; Cantaloupe Micropayment Trends Report 2025 (April 2025). Attendance definitions are from Stripe's support documentation. Operator thresholds are trade guidance, not rules — ask any operator you're considering for their own placement criteria.