December 16, 2025
By: Dr. Tim Judkins, Principal Engineer, Flybuy
At Flybuy, location precision and accuracy are always at the core of our platform. This year, the team expanded our ETAi® solution with the development of a new probabilistic event triggering model, designed to deliver just-in-time events during the final moments of a customer’s journey.
After analyzing millions of pickup trips, we uncovered a surprising pattern: traditional map provider ETAs are often 60–90 seconds behind a customer’s actual arrival. Small factors like store layouts, traffic lights, or parking configurations can throw off timing in big ways.
So, we set out to solve it. Our engineers built proprietary AI models that continuously learn from real Flybuy Pickup data, fine-tuning arrival predictions and making each guest interaction smarter, smoother, and more precise.
How It Works
The enhanced ETAi® model leverages extensive historical location data to create a site-specific probability field (or “heat map”) to model the likelihood of a customer being within a specified time (e.g., 30 sec) from the site.
Image A shows an example of a heatmap trained for 30 seconds prior to arrival. Note how this approach not only takes into account customers arriving via the main road, but also accounts for customers cutting through the parking lot. The model also takes into account when the customer arrives on premises as opposed to a “pin” that traditional maps provide.
Flybuy’s enhanced ETAi® triggers events for future orders when we are confident that a customer will arrive within that time period. We tested ETAi® as a prearrival event trigger and found that ETAi® significantly improves the timing of events compared to ETA-based triggering.
For example, Image B represents data from a site with a target prearrival of 30 sec always triggered on arrival with ETA-based timing whereas ETAi®-based timing prior to arrival. Note this initial implementation triggers when we have a high confidence of arriving within 30 sec hence the range for ETAi® trigger times between 0-30 sec. We continue to tune the model for further improvements.
Why it Matters
When staff have time-critical tasks (like bagging, making a milkshake, firing fries, etc) for orders within 2-3 minutes of a customer arriving, ETAi® provides significantly more accurate timing compared to ETA alone.
The precision in predicting customer arrival times allows personnel to optimize resource allocation and expedite service handoffs, resulting in greater customer satisfaction.
Get Started
Contact us today to learn how Flybuy can improve your operations and guest experience.