Failed deliveries are one of the most stubborn headaches in urban logistics. The courier arrives, nobody’s home, and the whole costly cycle begins again: a second attempt, extra fuel, more emissions, and a frustrated customer. A new study published in Research in Transportation Economics (Escudero-Santana et al., 2026) takes a fresh angle on this problem: what if the delivery algorithm simply knew more about when you’re actually at home?
Having a deeper understanding of user habits can significantly improve the efficiency and effectiveness of delivery planning and execution. Accurate user behavior data allows for better route optimization, reducing delivery times and costs. It also increases the likelihood of successful deliveries by aligning delivery schedules with users’ actual presence at the specified locations.
The Core Problem
The numbers are stark. According to the paper, 44% of failed delivery attempts happen simply because the customer wasn’t there when the courier knocked. With e-commerce sales projected to exceed $8 trillion globally by 2027, even a small reduction in that failure rate translates into enormous savings in cost, congestion, and carbon emissions. Most current systems ask customers to declare a preferred delivery window, say, “morning” or “3–6 PM”, but stated preferences don’t always align with reality. Life gets in the way.
Three Tiers of Customer Knowledge
The researchers from the University of Seville designed a simulation-optimization framework that tests three progressively richer levels of customer information. The first is the standard approach: customers declare their preferred delivery windows, and the algorithm takes them at face value. The second adds historical data — what percentage of past deliveries to this address, in this time slot, were actually successful? The third, and most novel, uses a full availability profile: a probabilistic curve built from anonymized geolocation data that shows the likelihood that a customer is home at any given moment throughout the day.
What the Experiments Found
Simply adding historical success rates to the declared time windows didn’t move the needle much. The reason is intuitive: customers tend to choose windows that already align with their routines, so the historical data largely confirms what was already known.
The real gains came from full availability profiles. When the routing algorithm could consult a dynamic probability curve (knowing, for example, that a particular customer has an 80% chance of being home between 5 and 7 PM on Mondays but drops to 20% by noon), performance improved substantially. Compared with conventional morning/afternoon slots, the profile-based strategy reduced failed delivery rates for home deliveries by more than 45%. Costs fell by around 17% for tight 60-minute windows. Crucially, these gains came without sacrificing computational efficiency.
The study also confirms a well-known trade-off: narrower time windows reduce failed attempts but push up route costs, since couriers have less flexibility in sequencing stops. Availability profiles offer a way to break this trade-off, delivering precision without forcing customers into inconveniently tight slots.
Convenience Points Add Another Layer
When alternative delivery locations (parcel lockers, local shops) were added alongside home options, both failure rates and costs fell further. The best-performing scenarios gave each order three options: a personal location, a locker, and a flexible third choice. This flexibility lets the routing system adapt to real-world conditions on the day of delivery.
Privacy and Practicality
The team acknowledges the elephant in the room: this approach requires customers to share location data. They developed a prototype mobile app that periodically queries proximity to delivery addresses, without storing raw location history. Whether enough customers will opt in remains an open question, but for those who do, the logistics benefits are clear.
The Takeaway for City Logistics
This research offers urban logistics operators a concrete, data-driven path toward fewer wasted journeys, lower emissions, and better customer experience. As cities grow more congested and sustainability pressures intensify, smarter use of behavioral data (with appropriate privacy safeguards) may prove to be one of the most practical levers available.