Article   16. 2. 2021 – how continuous delivery software made possible to process 10 times more orders.

How did our planning program made possible to process 10 times more orders and saved two dispatching hours? Key application was developed to help the leading Czech e-shop with food.
from 22 questions
Cancel filter

When a new business is successful, sooner or later it will run into growth problems. When business is as successful as Tomáš Čupr’s Rohlí, growth becomes one big adventure. And not just for him.

In 2015, the two dispatchers of Rohlí were fully employed only in planning deliveries. With maximum commitment, they managed to compile 250-300 routes for 1,500 orders per day. No wonder they had their heeds full and at times didn’t catch up.

We got the task of designing a system that would automate the delivery plan, make it more efficient and free the service from outages and delays. It was also necessary to optimize the use of the vehicle fleet – not to block the delivery with a lack of vehicles or not to keep the cars on standby unnecessarily when there are fewer orders.

The need to process orders continuously and respond to them in real time led us to develop new software. What differs from classic planning programs is actually just a trifle: continuous delivery does not know the luxury of closing orders. However, the implications for the calculation are significant.

Routes are calculated and re-evaluated each minute

Our system examines the incoming order and pours it into a common pot. As time begins to run and the temperature rises, the algorithm goes through tens to hundreds of routes at the time to rank them and pick the best of them. If it can’t find a suitable route, it creates and quickly searches for other orders that might suit the current plan and order status better.

We carefully watch the ticking clock. As soon as the loading time approaches a certain order, our application closes its route and sends it to Rohlí, so that they can pick up, load and send orders to the world.

However, our system has learned to predict the order of picking orders before closing the routes. This helps the expedition reduce stress and error rates.

Storage and fleet challenges

It turned out that connecting planning with the fleet, which was initially available to Rohlí, will be a tough nut to crack. We cracked it by turning the problem around. We introduced the fiction of an unlimited fleet and stopped dealing with how many cars Rohlí needs for existing orders. Instead, we began to count how many orders he could take in order to distribute them in time with the available number of vehicles.

The transition to our logistic software freed dispatchers from the planning order and allowed them to pursue another useful agenda. At the same time, the daily capacity increased from the original 1,500 to today’s approximately 13,000 orders. Another 7,000 orders are generated by branches in Brno, Budapest and Vienna. At the same time, transport to more distant regions is also planned for the future development, which together significantly increases the demands on application performance.

Nevertheless, the number of orders on the route increased from five to almost three times the original number and the time to service the order was significantly reduced, which is the main indicator of savings for couriers and vehicles.