Big Data and Smart Cities: Last Mile Delivery and Urban Logistics
Connecting city data to retail fulfilment: the OMS view

✍️ Translate this blog into Hindi, Tamil or Telugu
Introduction
Last mile in retail is the final leg of fulfilling any order. This is when you see Amazon, UPS, or FedEx trucks in your town. Trucks aren't the only final-mile delivery option. Sometimes it's people wearing the company vests in their own cars. When any of these methods fails to deliver the package unharmed or to the right location, the delivery has to be attempted again, which usually means another trip through congested streets and traffic. The World Economic Forum estimates that demand for urban last-mile delivery will grow 78% by 2030, putting 36% more delivery vehicles on the streets of the world's top 100 cities and raising last-mile emissions by more than 30% if nothing changes (World Economic Forum, 2020).
This report argues that big data is the main lever that lets retailers, carriers and cities meet faster delivery expectations without overwhelming urban infrastructure. It examines the data sources involved, a reference architecture, real implementations at Amazon, UPS, FedEx, Walmart and Target, and city side smart loading zone and parcel locker pilots. It closes with the challenges that remain and the role the order management system (OMS) plays in turning city data into fulfillment decisions.
🛍️✨ From the author · ByteStoreWhat is meant by enterprise ready?₹500View offer →Why is the last mile a big data problem?
Last-mile logistics fits the definition of big data in all three dimensions: volume, velocity and variety.
UPS’s routing system draws on more than 250 million address data points (Sustainable Brands, 2013) and FedEx scans every package roughly 20 times before it is delivered (FedEx, 2024). This is just one example of volume.
When it comes to velocity, traffic, weather and driver locations change every minute. FedEx's SenseAware ID sensors report a package's location every two seconds (Parcel and Postal Technology International, 2024).
When we talk about variety, a single delivery decision may combine order data, store inventory, carrier scans, vehicle telematics, map and traffic feeds, curb availability, and locker capacity, just to name a few, each in a different format and owned by a different system.
In a smart city, much of this data is public or semi-public, so its value depends on how well it is shared across organizational boundaries.
A reference architecture
A typical big data platform for urban last-mile delivery has five layers, shown left to right below.

Figure 1. Five-layer big data platform for urban last-mile delivery
Table 1. Layers of the platform
| Layer | Function | Technology |
|---|---|---|
| Ingest | Streams order events, carrier scans, telematics, IoT sensors and city open data feeds | Apache Kafka, MQTT, REST APIs, Curb Data Specification |
| Store | Keeps raw and curated data for history and model training | Data lake or lakehouse on cloud object storage |
| Process | Cleans, joins and aggregates data in batch and in streams | Apache Spark, Spark Structured Streaming, Flink |
| Model | Predicts demand, travel times and delivery success; optimizes routes | Machine learning (ETA prediction, demand forecasting), operations research solvers |
| Act | Pushes decisions to systems people use | OMS sourcing rules, delivery date promises, driver apps, locker assignment, city dashboards |
How do the leaders use big data in the last mile?
UPS: route optimization at fleet scale
UPS's ORION (On Road Integrated Optimization and Navigation) system looks at daily packages, pickup times and historic route data to optimize the sequence of the driver’s stops. UPS reported ORION saved 6 to 8 miles per route during the rollout (UPS, 2015). After the full-scale deployment, ORION was credited with saving about 100 million miles, 10 million gallons of fuel and 100,000 metric tons of carbon dioxide per year, worth an estimated $300 to $400 million annually (INFORMS, 2016). UPS has noted that removing just one mile per driver per day is worth up to $50 million a year, which shows why small data-driven improvements matter at scale.
Amazon: Regional networks and machine learning inventory placement
Amazon's national fulfillment network was disassembled into eight interconnected regions. They use machine learning to predict demand, decide what items to stock and in which facility. This makes sure that the product is shipped from a warehouse closer to the customer (Retail Dive, 2024). The smaller same-day sites cut the average time from picking to packing and placing the package on the outbound dock to 11 minutes. In 2025, U.S. Prime members received more than 8 billion items the same or next day, and same-day item volume grew 70% year over year (Retail TouchPoints, 2026). Amazon has also opened its data to researchers: the 2021 Last Mile Routing Research Challenge, run with MIT, released 9,184 real delivery routes on the AWS Registry of Open Data so academics could train models that learn from experienced drivers' choices (Merchán et al., 2022).
FedEx: Real time visibility and predictive intervention
Microsoft Azure was used to build FedEx Surround, which combines scanned data, IoT sensor data and external conditions like severe weather, natural disasters and other delays. Surround lets teams intervene before a shipment is late (Chain Store Age, 2020). FedEx Dataworks Group is the company’s analytics division that turns the network's data into near real time decisions (FedEx, 2025).
Walmart and Target: Stores as fulfillment nodes
Walmart reports it can reach 93% of U.S. households with same-day delivery from its stores (RetailWire, 2025), and more than a third of store-shipped orders arrive in under three hours (SupplyChain247, 2026).
Target fulfilled more than 95% of its sales through stores and added local sortation centers, so store associates no longer sort and batch packages for carriers (FreightWaves, 2022). Its curbside Drive Up service costs about 90% less to fulfill than shipping from a warehouse (NACS, 2022). These models depend on accurate, real-time store inventory, which is exactly the data an order management system manages.
The City Side: Smart Curbs and Shared Lockers
Cities are now collecting their own data. Philadelphia's 2022 Smart Loading Zones pilot converted 21 Center City loading zones into reservable zones and used the Open Mobility Foundation's Curb Data Specification to measure who used the curb and for how long (City of Philadelphia, 2022). Similar pilots in Aspen, Omaha and Pittsburgh gave cities their first reliable data on loading demand by time and day.
Shared lockers show measurable results. In a 2018 pilot at the 62 floor Seattle Municipal Tower, a common carrier locker open to UPS and USPS cut total delivery time by 78% and reduced failed first deliveries to zero (Urban Freight Lab, 2018). A later controlled study in a Seattle residential building confirmed a statistically significant drop in the time drivers spent inside the building (Ranjbari et al., 2023). The WEF modeled that combining electric vehicles, off-peak delivery, multi brand lockers and dynamic rerouting could cut urban delivery emissions and congestion by 30% and delivery costs by 25% compared with doing nothing (World Economic Forum, 2020).
Connecting City Data to Retail Fulfilment: The OMS View
Most published work treats routing (carrier side) and urban planning (city side) separately. The decision that actually creates a delivery trip, however, is made earlier, inside the retailer's OMS: which node fills the order, what delivery date or window is promised, and whether the customer is offered home delivery, BOPIS (buy online, pick up in store), or a locker.
Published retail cases show where these decisions sit. In the Walmart and Target examples above, the store chosen to fill an order sets the neighborhood a delivery trip starts from, and the promise shown at checkout depends on store inventory being accurate in real time. Amazon acts even earlier, using machine learning to forecast demand and place inventory in the regional facility closest to expected orders (Retail Dive, 2024). When a promise is at risk, FedEx Surround combines scan, sensor and weather data to predict delays so teams can intervene before the customer is affected (Chain Store Age, 2020). Together, these cases place the key fulfillment decisions upstream of routing, inside order and inventory systems.
Big data makes these decisions city aware:
Smarter promising. Delivery windows can be calculated from predicted travel times, current traffic and carrier capacity rather than fixed transit tables.
City aware sourcing. The OMS can favor a store whose route avoids congestion or restricted delivery hours, not just the nearest node with stock.
Steering to consolidated options. When predicted delivery success at an address is low (for example, high rise buildings with past failed attempts), checkout can offer a nearby locker or BOPIS first.
Closing the loop. Delivery outcomes stream back as events and retrain forecasting and ETA models.
Challenges and Limitations
Data sharing. Carriers treat route data as a competitive asset, and cities lack standards to receive it. Open formats such as the Curb Data Specification help but are not yet widely adopted.
Privacy. Delivery data reveals where people live and when they are home. Platforms need security and clear governance.
🛍️✨ From the author · ByteStoreResume review within 48 hours₹200View offer →Data quality. Bad addresses and missing building access details still cause failed deliveries regardless of model quality.
Latency and cost. Streaming platforms and real-time models are expensive to run, and the benefit must justify the cost per order.
Integration. As UPS found with ORION, algorithms that work in the lab must be adopted by drivers and fit legacy systems (INFORMS, 2016).
Conclusion
The evidence from UPS, Amazon, FedEx, Walmart, Target and city pilots shows that big data already cuts miles, failed deliveries and emissions in the last mile. The next step is integration: connecting city-level data on traffic, curbs and lockers with retailer-level decisions on sourcing and promising. Retailers that treat the OMS as the point where city data turns into fulfillment decisions will be able to offer faster delivery while helping cities manage congestion rather than adding to it.
https://www.linkedin.com/in/arunrktech/
References
Chain Store Age. (2020). FedEx provides near real time view into shipments. https://chainstoreage.com/fedex-provides-near-real-time-view-shipments
City of Philadelphia. (2022, October 3). City announces new SmartCityPHL pilot project focused on loading zones in Center City. https://www.phila.gov/2022-10-03-city-announces-new-smartcityphl-pilot-project-focused-on-loading-zones-in-center-city/
FedEx. (2024). How big data is delivering the future. https://www.fedex.com/en-sg/business-insights/tech-innovation/how-big-data-is-delivering-the-future.html
FedEx. (2025). Proactive monitoring and real time visibility: FedEx Surround helps Polish businesses protect critical shipments. https://newsroom.fedex.com/newsroom/europe-english/proactive-monitoring-and-real-time-visibility-fedex-surround-helps-polish-businesses-protect-critical-shipments
FreightWaves. (2022). Target bets on same day delivery with new sortation hubs in Chicago, Denver. https://www.freightwaves.com/news/target-bets-on-same-day-delivery-with-new-sortation-hubs-in-chicago-denver
INFORMS. (2016). UPS On Road Integrated Optimization and Navigation (ORION) project. https://www.informs.org/Impact/O.R.-Analytics-Success-Stories/UPS-On-Road-Integrated-Optimization-and-Navigation-ORION-Project
Merchán, D., Arora, J., Pachon, J., Konduri, K., Winkenbach, M., Parks, S., & Noszek, J. (2022). 2021 Amazon Last Mile Routing Research Challenge: Data set. Transportation Science. https://doi.org/10.1287/trsc.2022.1173
NACS. (2022, July 26). Target expands ecommerce operations with new delivery hubs. https://www.convenience.org/Media/Daily/2022/Jul/26/4-Target-Expands-E-Commerce-Operations_Operations
Parcel and Postal Technology International. (2024). FedEx Surround launches in Asia Pacific with SenseAware ID. https://www.parcelandpostaltechnologyinternational.com/?p=28602
Ranjbari, A., Diehl, C., Dalla Chiara, G., & Goodchild, A. (2023). Do parcel lockers reduce delivery times? Evidence from the field. Transportation Research Part E: Logistics and Transportation Review, 172. https://www.urbanfreightlab.com/?p=3430
Retail Dive. (2024). Amazon reaches fastest Prime fulfillment speeds. https://www.retaildive.com/news/amazon-fastest-prime-fulfillment-speeds-doug-herrington/689444/
Retail TouchPoints. (2026, February 4). Amazon same and next day deliveries increased 30% in 2025. https://www.retailtouchpoints.com/topics/fulfillment-last-mile/amazon-same-and-next-day-deliveries-increased-30-in-2025
RetailWire. (2025, February 26). Walmart expands same day delivery, reaching 93% of households. https://retailwire.com/walmart-expands-same-day-delivery/
SupplyChain247. (2026). Walmart chooses Dallas to test storing marketplace inventory. https://www.supplychain247.com/article/walmart-tests-in-store-marketplace-inventory-delivery
Sustainable Brands. (2013). UPS cuts GHGs by 14,000 tons using big data. https://sustainablebrands.com/read/ups-cuts-ghgs-by4-000-tons-using-big-data
UPS. (2015, March 3). UPS accelerates use of routing optimization software to reduce 100 million miles driven [Press release]. https://app.3blmedia.com/news/ups-accelerates-use-routing-optimization-software-reduce-100-million-miles-driven
Urban Freight Lab. (2018). The final 50 feet of the urban goods delivery system: Common carrier locker pilot test at the Seattle Municipal Tower. University of Washington. https://www.urbanfreightlab.com/?p=3581
World Economic Forum. (2020). The future of the last mile ecosystem. https://www3.weforum.org/docs/WEF_Future_of_the_last_mile_ecosystem.pdf
✅ Quick check
3 questions to see what stuck.
0 votes · Sign in to vote
- chainstoreage.com/fedex-provides-near-real-time-view-shipments
- phila.gov/2022-10-03-city-announces-new-smartcityphl-pilot-project-focused-on-loading-zones-in-center-city/
- fedex.com/en-sg/business-insights/tech-innovation/how-big-data-is-delivering-the-future.html
- newsroom.fedex.com/newsroom/europe-english/proactive-monitoring-and-real-time-visibility-fedex-surround-helps-polish-businesses-protect-critical-shipments
- freightwaves.com/news/target-bets-on-same-day-delivery-with-new-sortation-hubs-in-chicago-denver
- informs.org/Impact/O.R.-Analytics-Success-Stories/UPS-On-Road-Integrated-Optimization-and-Navigation-ORION-Project
- doi.org/10.1287/trsc.2022.1173
- convenience.org/Media/Daily/2022/Jul/26/4-Target-Expands-E-Commerce-Operations_Operations
- parcelandpostaltechnologyinternational.com/?p=28602
- urbanfreightlab.com/?p=3430
❓ Questions readers ask
Why is the last mile a big data problem?
Last-mile logistics fits the definition of big data in all three dimensions: volume, velocity and variety.
How does city data support smarter promising?
Delivery windows can be calculated from predicted travel times, current traffic and carrier capacity rather than fixed transit tables.
What closes the feedback loop?
Delivery outcomes stream back as events and retrain forecasting and ETA models.
🛍️ Arun Kumar's shelf
Courses, cohorts and more from the authorGet Arun Kumar's next blog by email
One email a week when they publish. Add the weekly Best of ByteBlog too. No spam, unsubscribe with one click.
We email you once to confirm. By subscribing you agree to the privacy policy.
More from Arun Kumar

Two Trucks: Explaining VMs and Containers
VMs and Containers

When AI Found a Decade Old Bug in 70% of the Web?
What PHP2Shell Teaches New Developers ?

What If You Knew the weather before it started changing?
A new AI tool can now spot major storms and weather shifts weeks ahead, giving retailers and supply chains time to stock the right products and keep goods moving.

💬 Comments (0)
Top · NewestSign in to comment