Have you ever ordered food online, arrived at the quoted pickup time, and still waited an additional 10, 15, or 20 minutes for your food to be ready? What if the pickup and delivery times quoted by online ordering sites and apps were accurate and you trusted that your food would be hot and fresh the moment you arrived? This is the problem Fresh KDS is solving with an AI-powered order prediction system. Read on to learn more.

Navigating Uncertainty: Challenges in 2020

Everyone had a lot of things to be afraid of in March 2020 when a novel coronavirus of unknown virulence and transmissibility swept the globe. In addition to the obvious worries about their health (and the health of their loved ones), restaurant owners and workers had to worry about their livelihood as lockdowns shuttered in-person businesses.

The immediate impact on restaurants was stark: in the final week of March 2020 the number of year-over-year transactions dropped 33.85% for U.S. quick-service restaurants and 71.77% for U.S. full-service restaurants according to a report by the U.S. Department of Agriculture. The precipitous drop in year-over-year transactions continued into April before starting a slow recovery.

The saving grace for restaurants during this time was online ordering–through direct channels, third-party services, or both. Restaurants who had never offered online ordering before scrambled to get systems set up, and restaurants that previously only did a small percentage of orders "to go" suddenly found themselves doing 100% of them off-premise.

Off-premise orders, however, create a whole new set of challenges for restaurants to navigate. Most restaurants were built with a kitchen to match the size of the dining area. Historically, several natural throttles to control how many orders could be placed at a given time. For example, the number of tables in the restaurant, the host stand spacing seating, a cashier taking orders, or the number of cars that can get through a drive thru line all serve to regulate the demand placed on the kitchen. Online ordering bypasses these throttles. And although each ordering system has throttling rules, the average restaurant now accepts online orders from multiple platforms, which means it’s easy for orders to flood the kitchen during busy times.

Additionally, off-premise orders have a high threshold for accuracy. If a dine-in order is missing ketchup or was accidentally prepared with the wrong type of cheese, the restaurant staff can fix the problem easily and satisfy the guest. However, if the customer doesn’t discover the mistake until after they’ve left the restaurant it's much more difficult for the restaurant to recover and the damage to the relationship with the guest is greater.

Rise of Online Ordering: Shaping Survival

Now, three years after the pandemic forced restaurants to rapidly adopt new technology and adapt to changing consumer behavior, restaurant traffic has recovered to pre-pandemic levels, but restaurant operators are still feeling the lasting effects of the pandemic-triggered shifts. The percentage of online/off-premise orders has remained high for most restaurants. The typical Fresh KDS customer has seen a 3x increase in the number of online/off-premise orders compared to before the pandemic. Additionally, the impact of the pandemic on the labor market has made it harder than ever for restaurants to hire and retain employees. The National Restaurant Association’s state of the industry report for 2022 found that 7 of 10 restaurant operators said they didn’t have enough staff.

It is clear that the effects of increased online ordering are both beneficial and detrimental for restaurants. Likewise, the rise in online ordering has had both positive and negative impacts on restaurant customers. Millions of people enjoy the convenience of ordering online and having a wide range of food options delivered to their doors; however, when food isn't ready on time it's frustrating for consumers. Orders that take longer than expected leave hungry customers waiting for their food, while orders that are prepared too soon may be cold by the time customers receive them.

Off-Premise Challenges: Navigating New Realities

The ability to quote accurate times to customers and third-party delivery drivers would go a long way in solving both the kitchen operational challenges and providing a better experience to the guest. By quoting longer, but more accurate, wait times during busy periods restaurants would naturally smooth out the demand curve–instead of throttling online channels and losing orders, those orders would be pushed back to accommodate the current volume. Customers also benefit from accurate pickup times because it allows them to plan accordingly.

The solution seems clear, so why didn't anyone build this sooner? Research performed by Fresh KDS found that leading online ordering platforms are only accurate within 5 minutes of the actual ready time approximately 35 percent of the time, meaning the majority of customers are waiting more than 5 minutes for their orders or picking up orders that have been getting cold (or hot) for more than 5 minutes! There are two main reasons online ordering systems get quote times wrong:

1. Their static rules-based engines don't account for enough variables
2. They don't have insight into what's actually happening in the kitchen, a highly dynamic environment

Recovery and Reshaping: Three Years Later

Fresh KDS is uniquely positioned to accurately predict when orders will be ready. Since Fresh KDS receives orders from both dine-in and online sources it’s aware of orders from all channels. Additionally, Fresh KDS knows when orders are actually ready and complete based on status changes at the KDS screens and take out stations. Instead of making assumptions about how long items take to prepare and how long orders take to bag, Fresh KDS has actual real time data on every order that runs through the kitchen display system.

Using this real-time order bump data from the kitchen display system, Fresh KDS has built an AI model that can predict when an order will be ready. Current models are accurate within 5 minutes more than 85% of the time. These models will only get more accurate as Fresh KDS collects more data and feedback. The beauty of this approach is that it doesn’t require restaurant operators or kitchen staffs to do anything extra–there are no complicated rules to set up or guesses to make about capacity. All the kitchen has to do is use the KDS, bumping orders as they are prepared, and the model does the rest.

AI-Powered Precision: The Fresh KDS Approach

The order prediction system will have numerous applications. The first application that Fresh KDS is testing now is providing customers with accurate predictions via text message after the order is placed. In the future, these predictions will be incorporated into the checkout page of online ordering platforms and used to provide alerts and modifications to the KDS so kitchens are able to perform better. Contact our team to learn more about how Fresh KDS can benefit your business.