Have you ever ordered a burger at a self-ordering kiosk and seen a suggestion to add fries or a drink? At first glance, the feature seems simple.
Behind that offer, however, is technology that can analyze purchasing habits and suggest relevant products. AI-powered upselling on restaurant kiosks can help businesses increase order value without making customers feel pressured to spend more.
Imagine visiting a fast-food restaurant and choosing a burger as your main dish. Before checkout, the kiosk recommends a meal deal that includes fries and a drink.
You can accept the offer or continue with your original order. The process is quick, and the decision remains yours. For customers, these suggestions can make ordering more convenient.
For restaurant operators, the feature creates opportunities to introduce additional menu items and discover which products customers like to buy together. But how do these recommendations work, and how can restaurants measure their impact?
What Are Menu Recommendations on Self-Ordering Kiosks?
Menu recommendations on a self-ordering kiosk are suggestions for additional items shown while customers place orders through a digital screen.
These suggestions may include side dishes, drinks, desserts, or meal deals. Restaurants often use them as part of their upselling and cross-selling strategies.
Upselling encourages customers to choose a higher-value product or meal deal. Cross-selling, meanwhile, offers complementary items that go well with the main order.
For example, when a customer orders fried chicken, the kiosk might suggest rice, fries, or a drink. The goal is not simply to add more items to the cart, but to help customers find a suitable combination.
When recommendations are informed by data, AI can help restaurants identify purchasing patterns and present more relevant offers. To see how this works, consider the ordering experience from a customer’s perspective.
Read also: Mispricing Can Cost You Money! Know the 4 Key Factors in Setting Your Restaurant Menu Prices
From Choosing a Main Dish to Receiving Add-On Recommendations
Suppose you visit a restaurant to order a burger. You browse the kiosk menu, choose your burger, select the quantity, and move to the next step.
Before payment, the screen suggests fries and a drink. If the restaurant offers a meal deal, you may also see an option to upgrade your order to a set meal.
You can then decide whether to accept the suggestion. If you want a more complete meal, you can add the fries and drink. If you only want a burger, you can skip the offer and proceed to payment.
For customers, the experience is similar to receiving a menu suggestion from a cashier. The difference is that the entire process takes place on a digital screen.
For restaurant operators, each interaction can provide useful information. They can see which items are frequently ordered together, which suggestions customers accept, and when customers are most likely to add an item.
However, recommendations should not appear at random. The system needs to consider relevant information so that each offer makes sense.
What Data Can AI Systems Use for Menu Recommendations?
Menu recommendations can draw on operational data and transaction patterns. AI can analyze this information to identify connections between customers’ orders and other items they may want to add.
One important factor is the item a customer has already selected. When someone orders a burger, for example, the system may recommend fries or a drink as a complement.
If a customer orders coffee, the kiosk might suggest a pastry or a snack instead. The system can also learn which combinations customers frequently purchase together.
For example, transaction records may show that many customers buy fried chicken with rice and tea. The system can use that pattern to suggest a similar combination to other customers.
The time of day may also influence recommendations. At lunchtime, a restaurant might highlight items suited to a main meal. In the afternoon, snacks and drinks could be more relevant.
Outlet location can matter, too. Customers in different areas may have different preferences, so transaction data from each outlet can help restaurants understand local purchasing patterns.
Stock availability is another important factor. By taking inventory data into account, the system can help restaurants recommend items that are actually available, not just items that might sell well.
Recommendations Should Be Relevant and Reflect Stock Availability
Imagine selecting a burger only for the kiosk to recommend a side dish that is out of stock. You are interested, but then have to change your order because the item is unavailable. This can create unnecessary frustration.
To avoid this, recommendations should take the restaurant’s current menu availability into account. If fries are out of stock, for example, they should not appear as an add-on option.
Relevance matters just as much. Suggesting dessert after a main dish may be helpful, but displaying too many unrelated products can interrupt the ordering process.
Restaurants should decide which product categories are eligible for recommendations, which menu combinations make sense, and when offers should appear. This helps AI-powered recommendations support a smoother ordering experience.
Read also: What Is Grammage? The Secret to Accurate Measurements for Consistent Flavor!
Do Not Let Pop-Ups Slow Down Checkout
One advantage of self-ordering kiosks is that customers can manage their own orders. They can browse the menu, customize their selections, and pay without waiting for a cashier to enter each item.
Too many recommendations, however, can undermine that convenience. If a pop-up appears on every screen, customers must work through multiple offers before they can check out.
This can increase ordering time and lead to queues at the kiosks, especially during lunch rushes or on weekends.
Restaurants therefore need to choose when recommendations appear. They might show one offer after a customer selects a main dish and, when relevant, another before checkout.
Keep the message short, use clear product images, and give customers an easy way to accept or skip the suggestion. The ordering experience should remain the priority.
Once these recommendations are in place, restaurants still need to check whether they are making a measurable difference to sales.
Measure Recommendation Acceptance and Order Value
A kiosk showing more offers does not automatically mean the recommendation feature is working. Restaurants need to track how often customers actually accept those offers.
One useful metric is the recommendation acceptance rate: the number of accepted recommendations divided by the number of recommendations displayed, multiplied by 100%.
For example, suppose a kiosk displays a recommendation for fries 1,000 times and customers add fries to their orders 250 times. The acceptance rate is: 250 ÷ 1,000 × 100% = 25%.
This means customers accepted 25% of the recommendations shown. Restaurants should also monitor average order value (AOV), or the average amount spent per transaction.
Compare average order value before and after introducing recommendations. Also track the number of items per transaction and the sales contribution of add-on products.
However, an increase in AOV cannot automatically be attributed to AI recommendations. Promotions, price changes, seasonality, and customer traffic can also affect the results.
Test Menu Combinations to Find the Most Effective Offers
Every restaurant serves different customers. A combination that performs well at one outlet may not work as well at another. For example, a restaurant could test two add-on offers for a burger.
The first offer could suggest fries, while the second could offer fries and a drink as a meal deal. Restaurants can compare acceptance rates, order values, and the time customers take to complete their orders.
Similar tests can be run for other menu items. Customers ordering fried chicken may prefer rice and a drink, while coffee buyers may be more interested in adding a pastry.
Run tests over reasonably comparable periods and under similar conditions. Avoid changing every recommendation based on a single day of results; consistent data gives a more useful picture.
Why Human Input Still Matters in Menu Recommendations
AI can analyze transactions and identify combinations customers often buy together. However, restaurant teams still need to guide the system so that recommendations support business goals.
Marketing teams may choose which menu items to promote, while operations teams check stock levels and kitchen capacity. Finance teams can assess whether the recommended products offer suitable profit margins.
For instance, an item may have a high acceptance rate but a very low profit margin. Restaurants need to consider whether it is worthwhile to keep recommending that item.
The same applies when a restaurant introduces a new menu item. There may not be enough transaction data to reveal purchasing patterns yet. In that case, the team can set initial offers based on its marketing strategy and review the results later.
FAQs About AI Menu Recommendations on Self-Ordering Kiosks
1. What does AI do in kiosk menu recommendations?
AI can analyze transaction patterns and relationships between products to suggest relevant additional menu items to customers.
2. Can AI recommend items based on what a customer orders?
Yes. The system can use the item already selected to suggest complementary products, such as fries with a burger.
3. How can restaurants measure the success of AI recommendations?
Restaurants can monitor recommendation acceptance rates, changes in average order value, items per transaction, and sales generated by add-on items.
4. Does AI have to decide every recommendation?
No. Restaurant operators still need to set the strategy, choose suitable items, and make sure recommendations reflect available stock and operational capacity.
Conclusion
Menu recommendations on self-ordering kiosks help restaurants suggest additional items based on the current order and customer purchasing patterns. For customers, this can make it easier to find suitable combinations.
For restaurant operators, interaction data can help assess upselling performance. Success, however, should not be judged solely by how many recommendations appear on the screen.
Restaurants should also consider acceptance rates, average order value, product relevance, and how smoothly customers can check out. AI-powered menu recommendations are most useful when they help restaurants respond to customer needs.
Restaurants need self-ordering kiosks that make ordering easier while creating opportunities for additional sales. Discover ESB Kiosk to learn how self-service ordering technology can support restaurant operations and sales performance.
