The Algorithm That Stopped UPS Turning Left — and Saved $400 Million a Year

The Left Turn Problem

This sounds like the set-up to a joke. It isn't.

Left turns — or right turns in left-hand-drive countries like the UK — are dangerous, time-consuming, and fuel-inefficient. To turn across oncoming traffic, a driver must wait. In heavy traffic, that wait can be significant. Over millions of deliveries, those seconds compound into hours, then days, then years.

UPS, the global parcel delivery company, had known this intuitively for decades. Drivers would informally avoid left turns when they could. But intuition isn't scalable across 100,000 vehicles and tens of millions of daily deliveries.

So they built an algorithm to do it systematically.

ORION: On-Road Integrated Optimisation and Navigation

ORION — On-Road Integrated Optimisation and Navigation — is UPS's proprietary route optimisation system. It was developed over a decade, beginning in the early 2000s, and rolled out across the US fleet from 2012 onwards.

The system considers more than 200,000 variables to calculate the optimal route for each driver each day. These include traffic patterns, delivery time windows, package weight, truck loading sequence, and — centrally — the preference for avoiding turns across traffic.

The core insight is counterintuitive: the shortest route is not always the fastest or most fuel-efficient route. A route that adds three miles of driving but avoids six left turns might use less fuel, take less time, and have a lower accident probability.

The Numbers

UPS has published various estimates of ORION's impact over the years. The headline figures are striking:

• **Approximately 100 million miles saved** per year across the US fleet
• **10 million gallons of fuel saved** annually
• **100,000 metric tonnes of CO2 reduced** per year
• **Cost saving of approximately $300–400 million** annually

For context: UPS employs around 125,000 drivers in the US. The saving is the equivalent of eliminating the annual fuel consumption of a small city.

ORION also has a safety impact. Fewer left turns across oncoming traffic means fewer accidents. The actuarial savings are harder to quantify but are real and significant.

Why This Matters Beyond Logistics

The UPS story is often cited as a logistics case study, and it is. But the underlying lesson is more broadly applicable.

The best optimisations are often invisible. Left turns feel like a trivial detail. They are, until you multiply them by 100 million daily deliveries. The most impactful software improvements frequently address problems that look small in isolation.

Human intuition doesn't scale. Experienced UPS drivers knew to avoid left turns when possible. But human judgment can only optimise a single route on a single day. An algorithm can optimise every route simultaneously, accounting for variables no human could hold in mind at once.

The investment horizon matters. ORION took a decade to develop and a further several years to roll out fully. The business case required patience. But the return — $400 million per year — is an investment that pays for itself many times over annually.

Data is the prerequisite. ORION only works because UPS had years of data on delivery patterns, traffic, and driver behaviour. The algorithm is the application of that data. Without the data foundation, there's nothing to optimise.

The Smaller Version of This Story

Most businesses won't build an algorithm with 200,000 variables. But most businesses have their version of the left turn problem — a process that, when optimised, would compound significant savings across thousands of repetitions.

It might be the order in which customer service tickets are assigned. The sequence in which products are picked from a warehouse. The routing of field service engineers. The prioritisation of sales leads. The scheduling of maintenance visits.

These problems look operational. They are. But they're also software problems — specifically, the kind of software problems where a well-designed system can deliver measurably better outcomes than human judgment alone.

We work with businesses at exactly this point: where a process is complex enough to benefit from systematic optimisation, but where nobody has yet built the system to do it.

The UPS story is an extreme example of what's possible. The principle — that software can optimise what human judgment can't scale — applies at every size.

The question isn't whether your business has a left turn problem. The question is whether you've looked for it. Our [consulting and strategy service](/services) is often where this conversation begins — [get in touch](/contact) if you'd like to explore what optimisation might look like for your operations.