Amazon split its US fulfillment network into eight largely self-sufficient regions. Site-to-customer distance fell 15% and cost per unit dropped over $0.45, by Amazon's count.
A four-warehouse distributor has the same problem. OpEra builds that model in weeks.
26
Transvision plans up to 15,000 subsidized Dutch taxi rides a day with a model that pools them into shared routes. It cut 15M driving km a year, by the team's conservative estimate.
A 40-vehicle shuttle operator has the same problem. OpEra builds that model in weeks.
19
DHL Supply Chain built a model that decides which shipments go on its own trucks, which go to carriers, and how they're routed. By DHL's estimate it saved $117M in 2.5 years.
A 25-truck distributor has the same problem. OpEra builds that model in weeks.
12
Swedish home care teams replaced manual visit planning with a model that assigns each visit and routes each carer. Efficiency rose 10–15%, worth €20–30M a year, by the team's count.
A 20-technician service team has the same problem. OpEra builds that model in weeks.
8
MISO uses optimization to decide which power plants run and how much each produces, every five minutes. The region saved $2.1B–$3.0B from 2007 to 2010, by MISO's estimate.
A site with its own generators has the same problem. OpEra builds that model in weeks.
1
19
Canadian Pacific moved its freight trains from "leave when full" to a schedule built by optimization models. It saved C$300M from mid-1999 to autumn 2000, by CP's count.
A regional carrier has the same problem. OpEra builds that model in weeks, on your own data.
1
20
In 2021 Lyft changed its matching to count where each driver would be needed next. Requests with no driver found fell 13%. Lyft's estimate: over $30M a year in revenue.
A dispatch desk with 40 vehicles has the same problem. OpEra builds that model in weeks, on your own data.
1
13
Baosteel replaced manual planning at its Shanghai steel plant with four optimization systems. Its estimate: $76.81M in benefit from 2007 to 2012 and 585,770 fewer tons of CO2 a year.
Planning from experience at a smaller plant? OpEra builds that model in weeks, on your own data.
11
Tata Steel let an optimization model reset its annealing furnace every minute. Strip within 5°C of target rose from 30% to 50% and fuel per ton fell 8%, by its count.
A mid-size heat treater has the same problem. OpEra builds that model in weeks, on your own data.
1
39
Merck Animal Health built models that decide how each vaccine batch is run, including when to harvest. It added $200M in revenue with no new equipment, by its own count.
One fermentation plant has the same problem. OpEra builds that model in weeks, on your own data.
1
17
Samsung cut the time to make a memory chip from 80+ days to under 30 with a model that sets work-in-process targets at every step. The authors estimate $1B in extra sales.
A three-line plant has the same problem. OpEra builds that model in weeks, on your own data.
11
Motorola built a model that decides which supplier wins which part, within capacity limits. It saved more than $600M on $16B of spend, by Motorola's count.
A plant with 40 suppliers has the same problem. OpEra builds that model in weeks, on your own data.
1
19
Same planes, same seats. American Airlines estimated $1.4B in extra revenue over three years from deciding, with a model, which seats to sell at which fare.
Trucks, lines, slots: any fixed capacity has that choice. OpEra builds that model in weeks, on your own data.
11
In 1997 CSX put in a model that decides which empty railcar goes to which customer order. CSX's 2009 estimate: over $51M saved a year and $1.4B in railcars never bought.
Trailers, containers, rental gear: same problem. OpEra builds that model in weeks, on your own data.
10
Alibaba promises groceries in 30 minutes, so its routing models get seconds to plan picking and delivery. The authors count about $55 million a year saved across five businesses.
Routes that change all day, at your size? OpEra builds that model in weeks, on your own data.
1
42
Chile feeds 1.3 million schoolchildren a day. Since 1997 an optimization model has decided which caterer serves which region: prices came in about 22% below trend.
Awarding lanes or contracts by judgment? OpEra builds that model in weeks, on the data you already have.
1
10
Hewlett-Packard had too many products. A model showed which ones covered most of the revenue, and which to launch. HP counts more than $500M in added profit since 2005.
Too many SKUs for your lines and warehouses? OpEra builds that model in weeks, on your own data.
14
Sears routed 12,500 service vans and 1,000 delivery trucks with a model that built tomorrow's routes overnight. Savings by Sears's count: over $42M a year.
It works for 30 vans too. OpEra builds that model in weeks, on the data you already have.
1
24
Netherlands Railways threw out a timetable from 1970 and rebuilt 5,500 daily trains with optimization models. Its estimate: €40M a year in extra profit on the same tracks.
Is your schedule still built on an old structure? OpEra builds that model in weeks, on your own data.
18




















