It’s a common saying that first you get good, then you get fast. Such a sequential approach to operations, unfortunately, is a luxury couriers can’t afford. To have success in the fast lane you’ve got to be good and fast – at the same time.

The courier business must contend with the dual realities of dealing with millions of transactions that generate relatively low dollars per transaction and having to support a high cost base due to customer service expectations. Changes in the business climate and competitive pressures point to a trend toward premium service at a low cost. An increasing demand for pre-10:30 a.m. deliveries, escalating daily and seasonal volume fluctuations and larger, lighter packages are all driving this trend.

Key to balancing the need to meet customer service expectations with the requirement to manage costs in a low-revenue per transaction industry is the logistics behind how the shipment moves from the dock to the customer.

Consider the numbers:

– In urban centres, 40-50% of a courier’s day is spent on driving versus 30-40% on customer contact time

– up to 30% of a courier’s pickup stops are called-in on the same day

– a 1% improvement in courier productivity translates into a savings of several million dollars.

Yet when Purolator, one of the largest players in the North American courier sector, looked to design and develop more efficient courier routes, it quickly concluded the traditional methods of route planning – essentially riffling through piles of route sheets and paper maps -were no longer sufficient to incorporate the increasingly dynamic demands of customers. This led to the search for sophisticated computer-based tools that could help with this process and what came to be known as ‘Route Optimization’.

“Finding ways to better structure routes became a key initiative to both increasing productivity and creating competitive advantage,” explained Mark Tilden, vice president engineering at Purolator.

SOFTWARE SEARCH AND SELECTION

The first step towards developing improved courier routes was to determine what routing tools were available and which would best suit the needs of the company, which in a typical day in Toronto completes over 50,000 deliveries and pickups.

The industrial engineering team at Purolator was tasked with determining both the requirements for the system and creating a viable business case.

“With a decentralized management structure and a wide-range of facility volumes, we needed scalability and flexibility,” recalled Bogdan Coroblea, transportation systems manager, who led the design team. “There was also a requirement to handle large quantities of data as some facilities complete up to 8,000 delivery and pickup stops per day. Lastly, we required full mapping capabilities for all provinces in Canada down to the level of individual street addresses, even in rural areas. We were asking a lot from this system.”

After an initial screening process, four specialized software vendors were contacted and detailed discussions on their offerings were held over a period of three months. After formal presentations, two firms were selected to participate in a series of tests. These tests were to validate the performance of their systems on key areas such as geocoding accuracy. The test results showed one firm most capable of meeting the requirements.

GEOCOMtms, based in Quebec City, is a small firm looking to expand into new markets. As part of the discussions, it agreed to work with the engineering team to run a pilot site and test the capabilities of the system.

“We were interested in expanding the capabilities of our system,” said Francois Poitevin of GEOCOMtms. “It knew it would be a challenge to implement, but we felt it would be a win-win project.”

PILOT SITE

The Purolator Hamilton facility was selected as the pilot site as it represented a typical-size operation while not being too unwieldy in terms of data demands. “We also chose this facility because we had recently completed a manual re-route and wanted to see whether the technical solution could work better,” said Tilden.

The next step was to determine the best process for extracting and converting the large quantities of customer data into geographic data. By selecting the actual stop data from a representative day’s operation, the route structure could be evaluated to identify areas for improvement. However, without longitude and latitude coordinates, all the existing data had to be geocoded so that the mapping system could plot the stops. This entailed extensive editing of the data so that it followed a consistent pattern. “After much effort and several passes, we achieved a geocoding success rate of 97.5% which was excellent for the pilot site. The remaining 2.5% had to be plotted manually”, said Bogdan. “Once we had the process defined, implementation at other sites would only increase this rate.”

There were also some challenges encountered in the pilot site. As the system uses map data from multiple sources, there were often conflicts between postal code data and municipality data. A related issue was the recent amalgamation of smaller cities, such as Stoney Creek, into the city of Hamilton. These issues led to the refinement of the mapping development process to better integrate map updates.

One of the outcomes of the testing was to develop a “grid” that overlays the digital maps. This grid was customized for the facility and was a network of relatively square shapes, or “cells”, that encompassed a selected number of stops. As stops are usually grouped in clusters, such as industrial areas or residential neighborhoods, the cells represented logical portions of work.

Explaining the reasoning behind this, Bogdan said, “The grid was necessary so that we could develop route territories rather than specific travel paths. This also allowed the frontline managers to get creative and try new route structures which is the true power of the system.”

The Hamilton pilot was a great success. By creating logical and flexible routes, courier productivity improved by a further 8% over the manual route plan.

Service levels improved and the couriers were also pleased with the results citing a more coordinated approach to volume fluctuations. An added bonus was that the frontline managers now had a tool to adjust route structures and see the impact before implementing changes.

IMPLEMENTATION

After the pilot site and validating the route optimization process, the next phase was to implement at other locations. Rather than a regional or geographic approach, large facilities were prioritized over medium and small facilities. This was expected to generate the greatest improvements in the shortest time possible.

The overall plan was split into four phases with Phase 1 focusing on simultaneous implementation at 11 key facilities across the country. Emphasis was placed on the Metro Toronto region as it contained a high concentration of large facilities. Jim MacIntosh, general manager for the region saw this as a huge opportunity: “We were going to dramatically change the operations at six key facilities in the Toronto area and we were a little nervous about that. But the results from the pilot confirmed we were on the right track to help each facility achieve higher performance levels in a short time frame.”

The engineering team set up a training and operations centre at the company’s Mississauga facility and installed six workstations networked to a dedicated server at head office. Five additional workstations were connected from Richmond and Burnaby B.C., Calgary Alta., Montreal, Que., and Halifax N.S. An issues log listed all the problems encountered and the corresponding solutions so that anyone could quickly look up the answer to common problems. Once each manager had developed an optimum route structure, they then worked with the engineers to build a detailed operations plan. This included creating the route details for the couriers, reprogramming the barcode scanners, retraining package sorters and even drawing new truck parking plans. For the Toronto locations, the workstation was then moved to their base location so that the local couriers and managers could make use of the system.

RESULTS

The results of Phase 1 were remarkable. Of the 11 facilities, nine were implemented within a three-month period and typically exceeded pre-implementation performance levels in 60 days. Productivity improvements ranged from 3% to 12% with the average running close to 6%. This compared to an average improvement at non-optimized facilities of about 2% during the same time period.

Currently, 29 facilities have completed the route optimization process with another seven in progress. Some facilities are now looking at repeating the process to fine tune their structure and reap additional benefits.

“There is a huge amount of untapped potential in this system ranging from daily workload planning to identifying low stop-density areas for the sales group to realigning service territories,” commented Tilden. “This is just the beginning of several phases in optimization which will later include dynamic routing of dispatch calls and delivery routes.”