September 17, 2026
Checkout-free stores use cameras, sensors and retail systems to track what customers pick up and charge them automatically when they leave.
Amrita Bhambhani
At Tesco GetGo and Aldi Shop & Go trial stores, a shopper could pick up a sandwich, a bottle of water and a packet of crisps and walk straight out. Cameras and sensors tracked what left the shelf and what went back, then charged the shopper's linked payment method once they'd gone.
For that to work, the store had to keep an accurate basket as the person moved around, knowing what they picked up, what they put back and what they eventually left with before charging the right amount and updating its stock. That becomes harder once people start shopping as they normally do.
Someone might change their mind and leave a product on the wrong shelf, reach across another person or briefly block a camera without realising it. Families might shop together but pay through one account. If the system loses track of an item along the way, it needs more sensor data to work out what happened or a member of staff to check the footage.
Staff were needed elsewhere too, from restocking shelves and helping customers find products to explaining the app and handling ID checks for age-restricted purchases. Taking away the till changed one part of the shopping trip, but much of the day-to-day work of running the store stayed the same.
That balance between a simpler checkout for the customer and a more demanding operation behind it helps explain why the model has struggled to spread across UK grocery. Tesco, Sainsbury's and Aldi have all closed their checkout-free trials, while Amazon said in September 2025 that its 19 UK Fresh stores would stop trading under that name, with 14 closing and five converting to Whole Foods Market. Smaller deployments have continued, including the Atrium Market at the University of Sussex.
The experience of those trials suggests that getting checkout-free technology to work is only part of the challenge. Running it across a large store network means installing and maintaining cameras and sensors, connecting them with existing retail systems and dealing with the many small uncertainties that come with people shopping every day, all of which has to make financial sense compared with simpler checkout options.
This guide follows that process from the moment a product leaves the shelf to the point where the customer is charged, looking at how the systems work together, where they can struggle, what they cost to run and where checkout-free retail can still be a practical choice.
A checkout-free store allows you to take what you need and leave without scanning each product or stopping at a till. The store keeps track of the basket as you shop, using cameras, shelf sensors, RFID or a combination of them to register what you pick up and what you put back. Payment can be linked through an app, a card tapped on entry or a badge at the gate, with the final basket charged when you leave.
That does not mean every checkout-free store gets rid of tills altogether. Some are built entirely around this model, while others keep staffed tills or self-checkouts and give customers the option to use them.
You will also see the same model described as cashierless, till-less or autonomous checkout. These terms refer to removing the conventional checkout step, not removing staff from the store. Nor does checkout-free necessarily mean cashless. A store can offer checkout-free shopping and still accept cash through a staffed till if it chooses to keep one.
There are several ways to let customers pay without going through a staffed till, but they do not all remove the checkout in the same way. With self-checkout or scan-and-go, for example, the customer still has to scan each product. Smart trolleys can take on some of that work, while RFID exit lanes and checkout-free stores can identify products without an individual scan.
Model | Who identifies each item? | Does the shopper scan? | Basket visibility | Infrastructure |
Staffed till | Store colleague | No | At the till | POS and scanner |
Self-checkout | Shopper at kiosk | Yes | At the kiosk | Kiosk, scales and payment terminal |
Mobile scan-and-go | Shopper using a phone or handset | Yes | Usually live | App, barcode scanning, payment and exit validation |
Smart trolley | Trolley or shopper | Sometimes | Usually live | Instrumented trolley and payment integration |
RFID exit lane | RFID readers | No individual scan | Usually at exit | Tagged items, readers and gate |
Checkout-free store | Store system | No | Live or after exit | Cameras, sensors or RFID, and commerce systems |
What separates checkout-free from most of these models is who keeps track of the products going into the basket. At a self-checkout, the customer scans them at the end. With scan-and-go, they scan them as they move through the store. In a checkout-free store, that identification happens in the background, so the customer can shop without scanning each item themselves.
A customer can spend only a few minutes in a checkout-free store, but in that time the store has to keep an accurate record of the entire trip. It needs to know when the trip starts, which products are picked up or returned and what belongs in the final basket before prices and promotions are applied, payment is taken and the sale is passed into the retailer's other systems.
That depends on the store having an accurate picture of what is on its shelves in the first place. Products are mapped to the shelves and sensor zones where they are expected to be, with the right SKU and catalogue information attached to them. Cameras and sensors are set up around the physical layout of the store, so when packaging changes, products move or shelves are rearranged, the digital version needs to change with them.
The shopping trip can then be opened with an app QR code, a contactless card tap or, in places such as workplace and university stores, a staff or student badge. Some setups work differently and allow customers to enter normally before dealing with payment at an automated exit gate. Where entry is used to start the trip, the system can assign a temporary token and use that to keep track of the customer as they move around the store.
This also means the cameras do not need to identify someone by their face. Doing so would involve biometric data and bring additional requirements under UK data protection law. Instead, what happens in the store can be connected to the trip opened at entry. Amazon, for example, says its Just Walk Out system tracks customers anonymously as they move through a store rather than using facial recognition to identify them.
Once inside, a customer might take something from a shelf, look at it and put it back, or reach for a product at the same time as someone standing next to them. Cameras can follow movement around the shelves, but they will not always have enough information on their own to know what ended up in the basket. Shelf sensors can register that weight has been removed or returned, while RFID can identify the exact item or variant when products are tagged.
The system can compare those different signals when the answer is unclear. If a camera briefly loses sight of an item, for example, a change in shelf weight may confirm that it was taken. If someone puts a product back in a different place, the system has to recognise that the expected shelf position no longer tells the whole story. Combining the available evidence is what allows it to keep correcting the basket as people move around and shop in ways that are not always predictable.
Technology | What it contributes | What it cannot solve alone |
Overhead cameras | Shopper location, hand movement, product and action recognition | Payment, pricing or moments when the view is blocked |
Shelf weight sensors | Confirms whether weight left or returned to a shelf zone | Distinguishing products of similar weight or identifying a misplaced product |
RFID | Identifies a specific tagged item or variant | Associating that item with a shopper or identifying untagged products |
Sensor fusion | Combines evidence from cameras, weight sensors and RFID | Poor catalogue or store-map data |
Edge computing | Processes store events locally to keep response times low | Tasks that depend on wider commerce systems |
Digital store map | Connects shelves and sensor zones with the products expected there | Remaining accurate when the physical store changes without an update |
Machine-learning models | Help recognise products and interpret more difficult sequences of events | Problems caused by missing or incorrect source data |
POS and payment integration | Turns the completed basket into a priced and paid transaction | Working out which physical products the customer took |
Once the system is confident about what is in the basket, the rest of the transaction can run through the retailer's existing commerce systems. The current prices, promotions, loyalty discounts and tax are applied there, so the cameras and sensors only need to work out what was taken, not what each product should cost.
The final amount can then be charged against a payment method authorised at entry or collected at an automated exit gate, depending on the setup. The customer gets a digital receipt, while the sale is recorded across the retailer's stock, inventory and loyalty systems in much the same way as any other purchase.
Someone picks up a bottle, changes their mind and puts it back two shelves away. Two people reach for the same product at almost the same time. A family moves around the store together while everything is being charged to one account. These are ordinary ways to shop, but they make it harder for a checkout-free system to keep the basket right. It has to follow what actually happened to each product, including where it ended up and which trip it belonged to.
There are other parts of a shopping trip that cameras and shelf sensors cannot handle on their own. Fruit and other products sold by weight may still need to be weighed, and age-restricted purchases can require a member of staff to check ID. Promotions and meal deals come from the retailer's pricing systems, so once the products in the basket have been confirmed, those systems still need to calculate what the customer should pay.
Things can go wrong after that too, particularly when a payment fails, a receipt does not arrive or a customer spots something they did not buy. Stores need a quick way to investigate and correct those problems, as well as another way to take payment if a camera, sensor or network failure means the checkout-free system cannot complete the transaction.
Where it gets more interesting is when the system has seen an event but cannot confidently decide what happened. Some providers send these low-confidence cases to trained reviewers, who can look at a short section of footage and help decide whether a product was taken, returned or added to the wrong basket. Human review like this raises its own questions: who does the work, where it is done, how long footage can be accessed and how quickly a disputed charge gets corrected.
Reporting on Amazon's Just Walk Out system, first published by The Information and covered widely since, found that human reviewers have been involved in checking a share of transactions when the system needs help. Amazon disputes the idea that reviewers watch shoppers live, saying the review happens after the fact and covers only a small share of visits. Either way, it's a useful reminder that retailers need to look beyond the experience at the exit when assessing how much work the model actually takes to run.
A busy shop will always have products left in the wrong place, people getting in one another's way, payments that do not go through and equipment that occasionally stops working. Removing the till changes how those situations are dealt with, but it does not make them disappear.
Faster throughput during busy periods
Benefit: When a lot of people want to buy a few items at once, such as during a lunchtime rush or a stadium interval, removing the need to scan each product can get more customers through the store and reduce queues at the checkout.
Trade-off: The queue can move rather than disappear. Entry gates, app issues or age verification can become the new bottleneck, so retailers need to look at the whole process and not just the time saved at checkout.
Longer opening hours
Benefit: Without a till that needs to be staffed throughout the day, stores can open earlier or stay open later in locations where keeping a conventional checkout running would otherwise be difficult to justify.
Trade-off: Those extra hours are not free. The store is still running its cameras, sensors and other systems, and may need remote support or human review, so the additional sales need to cover the cost of staying open longer.
More time for staff elsewhere in the store
Benefit: Time that colleagues previously spent scanning products can go towards replenishing shelves, checking freshness, helping customers or other work around the store.
Trade-off: Removing the till does not automatically reduce labour costs. Staff are still needed for many parts of the operation, and the value comes from what they can do with the time that has been freed up.
More usable store space
Benefit: Fewer tills and self-checkout machines can release space for more products or allow a retailer to operate from a smaller footprint, which can be particularly valuable in expensive or compact locations.
Trade-off: Checkout-free equipment takes up space too, including entrance and exit gates, cameras and sensors. The useful measure is how much selling space the new format actually creates and whether that translates into better sales per square metre.
Faster stock updates
Benefit: Picks and returns can be recorded as they happen, giving stock systems a more current view of what is moving through the store rather than waiting until products are scanned at checkout.
Trade-off: Faster data is only useful when it is accurate. If the system identifies the wrong product or the store map is out of date, incorrect information can reach the stock record just as quickly.
A quicker shopping trip
Benefit: For customers comfortable using an app, card or badge, checkout-free removes the need to scan every item or wait for a till, which can make a small, routine shop considerably quicker.
Trade-off: That convenience can disappear when entry is difficult, a payment hold is unexpected, a receipt is delayed or a wrong charge takes too long to correct. Privacy and trust matter as well. A 2025 study in the Journal of Retailing and Consumer Services on customer benefits and satisfaction in cashierless shopping systems found that customers weigh these alongside speed when deciding whether to use them.
The balance will look different from one store to another. A stadium shop may care most about getting more people through during a short interval, while a university store may get more value from staying open later with fewer staffing requirements. The question is whether the benefits that matter in that particular location are large enough to cover the equipment, integration, support and day-to-day oversight that come with the format.
Step 1: Be clear about what you want to improve
Start with what is happening in the store today. If queues build between 12 and 2, how long are people waiting and how many leave without buying? If longer opening hours are the goal, what sales are being missed while the store is closed? The same applies to staff time and floor space. Looking at sales, basket sizes, queues and staffing by time of day gives the pilot a clear starting point and makes it easier to judge later whether anything actually improved.
Step 2: Check whether checkout-free is the best way to solve it
Once the problem is clear, compare checkout-free with the other ways it could be addressed. Another self-checkout might be enough for a short lunchtime rush, while scan-and-go or a smart trolley may work better for a different type of store or basket. Checkout-free brings more infrastructure with it, so it needs to deliver enough additional value to justify choosing it over a simpler option.
Step 3: Get the store data ready
A checkout-free system needs to know exactly what is on the shelves and where it should be. SKUs, variants, product images and shelf locations all need to match the physical store, and someone needs to update them whenever packaging changes or products move. Similar-looking items are worth testing early too, particularly if the system has trouble telling them apart.
Step 4: Test the shopping trip from beginning to end
Try the trip the way a customer would actually experience it, including the less straightforward versions. How do they enter? Can they see what is being added to their basket? What happens if a parent enters with two children, a payment method is declined or someone wants to dispute an item on the receipt? Refunds, payment holds and the alternative for customers who cannot or do not want to use checkout-free should be worked through before the pilot opens.
Step 5: Track each part of the transaction
When a charge is questioned later, the retailer needs to be able to trace the transaction without piecing together different records from several systems. Product information, prices, promotions, payments and stock should each have a clear source, with the same transaction ID following the purchase through the basket, receipt, inventory update and any later return or customer query.
Step 6: Put the difficult cases through the pilot too
Do not test only the smooth version of the trip. Put a product back on the wrong shelf, have two people reach for the same area, try an age-restricted item, make a payment fail and see what happens when the network drops. Each case needs a clear owner and an expected turnaround time, so staff aren't improvising when a real customer hits one of these situations.
Step 7: Include privacy and accessibility in the testing
Customers need to know what data is being collected as they move through the store, why it is needed, who can access it and how long it is kept. The retailer also needs to complete any data protection work required for the setup being tested. At the same time, disabled customers and people using assistive technology should try the trip themselves, along with customers who do not have the app, smartphone or payment method the checkout-free route expects them to use.
Step 8: Decide what would make you keep it
Set the numbers before the pilot begins. If the aim is to reduce queues, agree how much they need to fall. If it is about longer hours, work out how much additional contribution those hours need to generate. Basket accuracy, disputed charges, human review, uptime and cost per transaction can sit alongside those measures, depending on the reason for the pilot.
A wider rollout is only one possible outcome. Checkout-free may earn its place in a handful of stores, during the busiest trading periods or in locations where queues are a particular problem. In other cases, the numbers may favour self-checkout or another simpler option. The point of the pilot is to find that out before committing to a larger investment.
The testing should include power cuts, network outages, failed payments and gate failures, with staff clear on how the store continues to operate in each case and customers able to leave safely even when a transaction has not completed. Running the pilot in stages or using a comparable store as a benchmark will give the results some context and make it easier to judge what checkout-free itself has changed.
The simplest way to look at checkout-free ROI is to compare what the store gains with what it spends. The gains might come from serving more customers at busy times, staying open for longer, using staff time differently or freeing up selling space. Better stock accuracy or lower shrink can also count, but only when there is evidence that checkout-free actually caused the improvement.
Add up the full cost
The initial cost covers the cameras, sensors, gates and installation, but the spending continues after the store opens. A complete calculation should include:
Hardware and installation
Software and transaction fees
Maintenance and calibration
Human review of uncertain transactions
Disputes and refunds
Privacy and security requirements
Accessibility and alternative checkout options
Some of these costs will be small individually, but together they determine what the format really costs to run.
Work out the cost per transaction
Take the total cost of running checkout-free over a given period and divide it by the number of transactions completed through it. This gives the retailer a cost per transaction, which only means something next to the same number for self-checkout, staffed tills or scan-and-go. If checkout-free costs more per transaction than self-checkout, for example, the extra cost needs to be justified by something else, such as more sales at peak times, longer opening hours or better use of the store.
Check what is changing in the store
Area | What to measure | What to check |
Sales | Transactions and sales during busy periods | Are more customers actually completing purchases? |
Queues | Waiting time at peak hours | Has the queue disappeared or moved to entry or age checks? |
Accuracy | Correct baskets and charges | How many disputes, refunds or chargebacks are there? |
Reliability | System uptime | How often does the store need another checkout method? |
Staff time | Hours spent on different store tasks | Has time actually been freed up for other work? |
Customer use | Trips completed without help | How many customers need assistance or cannot use the format? |
Test it under the same conditions
A busy week at one checkout-free store cannot simply be compared with a quiet week at another store. Promotions, holidays, local events and different opening hours can all affect sales and customer numbers.
A better test is to compare the pilot with a similar store over the same period, or introduce checkout-free gradually and measure what changes. Customers who need help, give up partway through a checkout-free trip or choose a staffed till instead are telling the retailer something useful, and that group should be counted rather than left out of the final numbers. A shorter checkout time on its own is not proof of a good return. It only counts as one once it has been weighed against everything the format costs to run.
What happens to customer data?
Checkout-free stores use cameras and sensors throughout the shop, which means privacy has to be considered before the system goes live. Retailers need a lawful basis for collecting and using the data and, where required, a data protection impact assessment. Customers should also be told clearly what is being collected and why before they enter the store.
How long that information is kept also matters. Video footage, events recorded during a shopping trip and the final transaction do not necessarily need to be stored for the same length of time, so each should have a clear retention period rather than everything being kept by default.
Tracking a trip is different from recognising a face
Most checkout-free systems do not need to know who a person is to follow their movement through the store. They can connect activity to the trip started with a card, app or badge instead.
Facial recognition changes the privacy requirements because using a face to identify someone involves biometric data, which can be special category data under UK GDPR and bring additional legal requirements. Retailers therefore need to establish exactly how a provider tracks customers rather than treating all camera-based systems in the same way.
Customers need a way to question a charge
If someone finds an item on their receipt that they did not buy, it should be easy to raise the problem and understand what happens next. An itemised receipt, a clear route for disputing the charge and information about whether footage will be reviewed as part of resolving it all help build that trust.
The rules around fully automated decisions changed in the UK in February 2026, when the Data (Use and Access) Act replaced the old blanket restriction with a set of required safeguards instead. Retailers should check whether an automatic charge counts as a "significant decision" under the new rules, since that determines what information, human intervention and challenge rights need to be built in.
Accessibility isn't a given
A checkout-free format needs testing with disabled customers and people using assistive technology, not an assumption that it works for everyone. Staff should be on hand to help rather than leaving someone to figure out an app alone.
UK retailers can generally choose which payment methods they accept, but reasonable adjustments may be required if a disabled customer would otherwise be seriously disadvantaged. A format that only works with a smartphone and a linked card excludes some customers by design, not by accident.
This section provides general information and should not be treated as legal advice.
Checkout-free is one way to change how customers pay, and the case for it will vary from store to store. Basket size, footfall, peak trading periods, available space and staffing all affect what makes sense. A supermarket handling a weekly grocery shop has very different requirements from a convenience store where most people are buying two or three items.
For stores with smaller baskets, mobile scan-and-go can remove much of the time spent at checkout without requiring cameras and sensors across the shop floor. Fynd & Go lets customers scan a QR code in store and use their phone to scan and pay for their products through the browser, so there is no app to download. Promotions are applied as they shop, and the completed purchase generates a QR code that can be checked at the exit. The format is currently used across more than 5,000 stores and handles over 1.5 million transactions a year.
Fynd Scan & Go runs through Fynd Store OS, alongside POS, self-checkout kiosks, endless aisle and order management. Inventory, pricing and customer data can therefore be shared across these different checkout formats instead of each one needing its own set of systems and integrations.
That shared foundation matters because changing the way someone checks out does not change the basic requirements of the transaction. The product still needs the right price, stock needs to be updated and promotions need to work as expected. Once those pieces are in place, retailers have more room to choose the checkout experience that suits each store rather than forcing one format across the entire estate.
Explore Fynd's Scan & Go
Cameras track your position and hand movement near the shelves, while shelf weight sensors and, in some stores, RFID confirm whether an item was actually removed or returned. These signals are combined to work out what belongs in your basket, including anything you pick up and then put back.
Not necessarily. Most checkout-free systems, including Amazon's Just Walk Out, track a shopper anonymously through the store rather than identifying their face. Your trip is linked to the card, app or badge used at entry instead. A retailer using facial recognition would need to meet extra requirements under UK data protection law, since that counts as processing biometric data.
You should get an itemised receipt and a simple way to flag the problem. Some checkout-free systems use human review of footage as part of resolving disputes, and a good retailer will be clear about whether that happens and how quickly a mistake gets corrected.
With scan-and-go, you scan each item yourself as you shop. In a checkout-free store, cameras and sensors do that identification for you, so there's nothing to scan at all.
AI stylists help shoppers find products and build complete outfits around their needs and preferences, using live catalogue data and clear styling logic.
How the Fynd AI solution and a custom shoot planning module are helping the home decor brand manages multiple SKUs across its website and marketplaces.
Virtual makeup try-on helps shoppers compare shades and finishes before buying, making online choices easier.
Fill out the form
Share your contact information to get started
Speak to an expert
A member of our sales team will get in touch with you