Kiosks' Evolution and Impact on Autonomous Experiences

Kiosks have been a common machine supporting businesses for more than two decades. They automate many customer-centric processes and their use is increasing as businesses try to better serve customers and respond to changing market conditions.

The kiosk market measured by revenue is dominated by the self-checkout kiosks. ABI Research estimates that in 2024, self-checkout kiosks accounted for 47% of revenue worldwide. These kiosks are larger and can handle many tasks needed to complete the checkout of items, including weight scales, barcode readers, credit card scanners, and cameras, and they have a large enough footprint to enable conveyance of such products through the checkout process.

When measured in shipments, there is a more even worldwide distribution with self-checkout kiosks leading at 30% followed by self-ordering kiosks at 25%. Different from self-checkout kiosks, which have more prominence in Western markets, self-ordering kiosks are being adopted worldwide, particularly at fast-food and fast-casual restaurants.

The 2020 pandemic accelerated the use of kiosks to encourage business activity by reducing human interaction at retail and other places of business. However, retailers are pulling back from their aggressive purchases with some major suppliers such as Diebold Nixdorf and NXP showing negative revenue growth in 2023 and 2024.

Interestingly, the pullback is mainly with self-checkout kiosks by retailers in home improvement, grocery, and mass merchandizing. The rest of the kiosk market continues to show steady growth.

The kiosk market is retrenching in some ways, but also experiencing steady growth from new markets such as Buy Online, Pick-up In-Store (BOPIS)/Buy Online, Return In-Store (BORIS). In addition, kiosks give many retailers options for automating and managing employee costs either from increased wages or worker accessibility. Below are the key changes for the kiosk market looking ahead.

Artificial Intelligence (AI) Incorporation

AI is finding use in both small and large ways. At one end is health and beauty kiosks. Customers can select makeup from a menu and through a camera on the kiosk can virtually see how that makeup will look on them. Another is in the self-checkout process applying AI to address friction points. Friction happens with unbarcoded items, such as produce, or trying to find the item in a menu. AI can recognize the item for the customer.

Another friction point is AI use for age verification. Age verification can take time from both the employee and the customer, but can also lead to customer abuse and/or employee embarrassment. In China, self-checkout kiosks already have facial recognition capabilities. Elsewhere, kiosks can have ID scanners or use camera functions to take a picture and use a database for age verification.

Modularization

As penetration of kiosks reaches higher levels and new technology becomes available, businesses do not want to have to replace the whole kiosk when all that is needed are point upgrades. Many kiosk suppliers are changing their product lines to entirely stackable units or enabling more plug-and-play capabilities. The benefit goes beyond satisfying businesses. Modularization can also bring down maintenance costs. Rather than sending a technician out to assess and correct a kiosk issue, the customer can self-assess the issue assisted by Original Equipment Manufacturer (OEM) remote care and, if necessary, correct the issue in a plug-and-play manner with a new module.

Software and the Customer Experience

As kiosk penetration starts to peak in the traditional verticals of supermarkets and mass merchandizing, software and services are becoming more important. Kiosk OEMs have to find new revenue streams as hardware sales decrease; and kiosk customers want to use software to get the most out of their hardware. The latter trend has been accelerated as tariffs are driving some customers to want to extend the life of their hardware, rather than buying new hardware. To this end, vendors such as NCR are launching cloud versions of their software such as Aloha Cloud for Point of Sale (POS) software. Diebold Nixdorf with its focus on the end-to-end in-store customer experience offers its Vynamic Retail Platform, its cloud-native software for customer experience management across grocery, convenience and specialty/fashion stores.

An interesting last point related to software is the relationship between modularization and software. There is a lot of custom software in the kiosk world as customers want to replicate the functionality of other devices in the kiosk or for better back-end system functionality. As standardized hardware volumes increase, bringing kiosk hardware costs down, more money can be spent on software for customization.

Chinese Supplier Influence

Chinese suppliers typically target Western markets not only for unit shipment growth, but also for profitability. Interestingly, Chinese suppliers have not been as successful in the U.S. market specifically with self-checkout kiosks. Part of the reason for the slower influence of Chinese suppliers in Western markets has been China’s preference for employing its large population, rather than using kiosks. They simply do not have the experience in self-checkout kiosks in their home market. But this is changing. Europe already has a heavy Chinese kiosk vendor presence. In North America, more Chinese suppliers are building a sales presence (although recent geopolitical tensions seem to have slowed official office openings). One issue for Chinese vendors such as Sunmi is that their kiosks are based on Android. Most Western markets use Windows for the kiosk Operating System (OS). Supporting Android complicates support and services. Additionally, Android is not container friendly, which becomes an issue as kiosks begin to incorporate more AI features.

Kiosks provide a window into how the market is adopting technology that is transitioning workflows and customer experiences toward more autonomy. Below are three key trends highlighting how kiosks are leading technology adoption for a more autonomous world.

Camera-Based IoT Is Just Beginning

Camera technology has developed to the point that anything can have an embedded camera. But cameras are not nearly as powerful if the data from the camera cannot be easily processed and applied to various customer and business activities. This has changed with the availability of Vision Language Models (VLMs) allowing quick identification of an Internet of cataloged things.

More and more kiosks are incorporating cameras and applying AI for workflows that are common in many industries. They include identification and verification. Kiosk processes use it to verify if a customer is the right age to purchase an item. Outside of kiosks, this same process may apply to machinery or truck operation. A camera-enabled tablet can verify if a driver is allowed to operate the machine.

Another task is object identification. Kiosks use camera images and AI to assist with items not paid for in a cart or to ease the checkout process. Video surveillance camera OEMs are now just starting to apply this technology for security purposes. Parking systems are adding cameras to check-in and check-out vehicles.

Compute Power at the Edge

It is becoming clear that for AI experiences to become possible, AI is needed at the edge to address privacy, security, and cost concerns. However, many machines do not have the horsepower to take full advantage of AI. In the kiosk world, this realization is now front and center, particularly for some workflows such as identity—the latency in connecting to the cloud as well as cloud and connectivity costs may not justify the Return on Investment (ROI). Some vendors see this as an opportunity. Esper, a device lifecycle management solutions provider, has helped clients manage AI workloads using an on-premises gateway. Offered via its Pando Edge AI Solution Accelerator, Esper is using local AI gateways to provide the right balance of local processing and memory for AI inferencing, while avoiding the delay and costs of real-time connectivity to the cloud.

Remote Assistance

The trend of driving more efficiency in operations using a smaller workforce combined with technology places kiosks as the perfect testing ground for remote assistance in two areas. The first is live agent services. Currently, telehealth kiosks from vendors such as REDYREF and Olea include video conferencing capabilities to enable remote patient visits. Automated Teller Machines (ATMs), although not considered a kiosk, are offering remote assistance for services such as loan applications. This same approach can be applied to wayfinding, price checking, and travel kiosks where location and price information can prompt product/service recommendations that are both AI generated and via remote assistance. The second area is remote support capabilities. As kiosks become more modularized, customers can self-diagnose and maintain their kiosks with the help of remote services. And using AI and Machine Learning (ML) to monitor kiosk functions, customers can be alerted to potential operational issues and through remote care services self-diagnose and repair machines. Some kiosk OEMs such as Diebold Nixdorf are already on this path, offering more services through their customer kiosk engagement and enabling end-to-end support that drives revenue and lowers costs.