#370 2004 · United Parcel Service (UPS) · Logistics / package delivery
UPS makes its trucks drive farther on purpose, because turning left costs more than driving extra blocks
the problem
The shortest path on a map isn't the same as the fastest, safest or cheapest path to actually drive
background
Route planning for a delivery fleet naturally optimizes for the shortest driving distance between stops, and turning left across oncoming traffic to reach the next stop on a map is often the shortest path available. The hidden cost of that turn doesn't show up in a distance calculation: a driver waiting in a left-turn lane for a gap in oncoming traffic burns idle fuel, loses time that compounds across dozens of stops a day, and faces a meaningfully higher accident risk than a right turn, since a left turn crosses the path of oncoming vehicles.
UPS runs roughly 55,000 delivery routes a day in the US alone, at a scale where a cost that looks trivial on any single turn — a few dozen seconds idling, a fractional increase in accident risk — compounds into a real operating expense once repeated across an entire fleet's daily mileage. Company accounts describe informal predecessors to a no-left-turn approach going back decades in UPS's loop-dispatch route planning, but the practice wasn't studied, quantified and made formal company policy until UPS began instrumenting its fleet with better vehicle-tracking data in the early 2000s.
what everyone would do
Keep optimizing routes for shortest total driving distance, the standard metric for delivery-fleet planning. It fails because distance isn't the same as cost: idle time waiting for a gap in oncoming traffic and the elevated accident risk of crossing that traffic never show up in a distance calculation at all, so minimizing miles driven can actually maximize the real cost of running the fleet.
what they saw
UPS saw that the metric everyone was optimizing, shortest distance, wasn't the thing that actually mattered, lowest total cost. A left turn's true cost — idle fuel burned waiting for traffic, time lost that compounds across dozens of stops, meaningfully higher collision risk — was invisible to a distance-based route planner but very real once repeated across tens of thousands of routes a day. A longer route with more right turns could be cheaper in aggregate than a shorter one with more left turns, because the hidden per-turn cost, multiplied by scale, outweighed the extra mileage.
the move
UPS formally announced its policy of routing delivery trucks to avoid left-hand turns across oncoming traffic in 2004, after data from its vehicle-tracking systems confirmed that eliminating left turns — even when it meant a truck drove farther and made more right turns to reach the same destination — reduced total fuel burn, idle time and accident exposure enough to outweigh the extra distance. Left turns now make up less than 10% of all turns UPS trucks make.
why it works
A left turn across oncoming traffic requires waiting for a gap on every single occurrence, burning idle fuel and losing time; at the scale of roughly 55,000 routes a day, even a cost of a few dozen seconds and a small increase in accident risk per turn compounds into a massive aggregate expense across the fleet's daily operations. Routing to minimize left turns instead of distance trades a small, guaranteed increase in mileage for the elimination of a much larger cost that was previously invisible to route planners, and because the marginal cost of extra distance is smaller than the eliminated idle-time and accident-risk cost per turn, the net effect is lower total cost even though the route looks less efficient on a map — confirmed by roughly 10 million fewer gallons of fuel burned a year after the policy took effect.
the payoff
UPS has reported using roughly 10 million fewer gallons of fuel a year as a result of the policy, cutting on the order of 20,000 tonnes of carbon dioxide emissions annually and delivering hundreds of thousands more packages with the fuel and time saved — an estimated $20 million or more in annual fuel-cost savings alone at typical diesel prices, before accounting for the reduction in accident-related costs from avoiding left-turn collisions.
where it breaks
The approach only pays off at a scale large enough for the per-instance saving to compound into something significant — for a single delivery or a small fleet, the aggregate benefit from avoiding left turns may not outweigh the guaranteed mileage increase. It also depends on the local road network actually making left turns costlier than right-turn detours; an intersection with a protected left-turn signal, or a layout where right-turn workarounds require dramatically longer detours, would flip the calculation back toward optimizing for distance directly. And it requires real instrumentation to verify the hidden cost and confirm the tradeoff — without vehicle-tracking data quantifying the actual fuel, time, and accident numbers, the policy is an assumption rather than a validated optimization, and could be net negative if the underlying cost estimate turns out wrong for a specific market.
what came after
The no-left-turn policy became one of the most widely cited real-world examples in operations-research and route-optimization teaching of a counterintuitive efficiency gain — a policy that increases distance traveled but decreases total cost — and was later reinforced and extended by UPS's ORION route-optimization software, rolled out fleet-wide by 2016, which validated and refined the same underlying insight algorithmically across the company's full delivery network.
references
- [1]UPS Trucks Don't Turn Left and Neither Should YouTIME, 2017time.com
- [2]'No Left Turns' (self-interested conservation at UPS)MasterResource, 2017masterresource.org