From a bus depot to a freight network

Mandate

Make the network's data dependable enough for automated planning and safe enough for drivers, consignees and shippers to trust.

Outcomes

• Installed dimensioning scanners at fourteen hubs and lifted average trailer fill on 38 lines from 78 to 84 per cent
• Moved 17,400 driver records to a central verification desk and deleted 9,600 stale ones
• Introduced a field catalogue for every cross-border and foreign-shipper data request
• Published lane-level emissions statements showing the measured share of each figure

Decisions in context

The chief executive created this role after the Personal Information Protection Law made clear that our data was no longer only an operating matter. I lead four teams and set the standards that our hubs, outlets and commercial staff follow when they record or share data. Much of the work is unglamorous: measurement equipment, deletion audits, contract clauses with platforms. The more visible work is in planning models and customer reporting, where I insist that recommendations show their inputs. I present to the board and to its audit and risk committee twice a year, covering incidents, open risks and the reliability of figures we give customers. My measure of success is that a hub manager, an outlet owner and a shipper's auditor can each ask where a number came from and get a straight answer.

Reason for the move

Current appointment; exploring one compatible non-executive role.

Mandate

Replace outlet-by-outlet records with one waybill history and establish ownership of the company's data.

Outcomes

• Brought 1,600 outlets onto a single waybill record
• Completed the first company-wide personal-information inventory before November 2021
• Named an accountable owner for each of 41 core data sets

Decisions in context

When I took this role, outlets still ran different systems and a consignment could carry three numbers on its journey. The single waybill record took two years and a great deal of negotiation with franchise owners who liked their own screens. It gave us, for the first time, a history we could trust from collection to delivery. The personal-information inventory came next, prompted by the new law. It was meant to be a compliance list, but it became the most revealing piece of work in my career, because it showed how much we held about drivers and consignees without any clear purpose. I reported to the chief information officer, who supported the work generously, but data ownership needed a separate voice from systems delivery. The board agreed in early 2022, and I moved into the new role.

Reason for the move

Appointed chief data officer when the board separated data responsibility from systems delivery.

Mandate

Provide the analysis for a redesign of the hub network as volumes tripled.

Outcomes

• Supported a hub redesign that added 23 hubs and closed four
• Introduced arrival-scan data into linehaul scheduling
• Built the first trunk-line cost model shared by planning and finance

Decisions in context

These were the years of rapid growth, when online sellers began sending bulky goods through networks like ours. My team modelled where new hubs should sit and how trunk lines should connect them. The best analysis we did was often the least sophisticated: arrival scans showing when freight actually reached a hub, compared with the arrival times the schedule assumed. I also saw the limits of declared volume repeatedly, though at the time we lacked the equipment to replace it. Working beside finance on a shared trunk-line cost model taught me to present a recommendation with its uncertainty, because a hub decision ties up capital for years. The vice-president for network operations made the final calls; my team's job was to give him options he could challenge. By 2018 our planning depended on data from hundreds of outlets, and its quality had become a strategic question.

Reason for the move

Asked to lead the data platform once the network's planning depended on a single shared record.

Mandate

Turn the founder's hand-drawn trunk-line schedule into a planned network that could grow.

Outcomes

• Moved linehaul schedules from spreadsheets to a planning system
• Cut empty trailer kilometres on the busiest twelve lines by about 9 per cent
• Started recording dispatcher changes to the plan with reasons

Decisions in context

The founder had drawn the first trunk-line schedule himself, and it worked because he knew every hub manager. That knowledge did not scale. My six planners and I moved the schedule into a planning system and began to measure how often it was followed. It was followed far less often than anyone thought, usually for good reasons: a customer's late freight, a breakdown, a closed expressway. I started asking dispatchers to record why they changed the plan. Those notes were rough, but they showed which assumptions were wrong. The reduction in empty running on our busiest lines came as much from that record as from better optimisation. I joined Qianmoyun because consolidation freight seemed the hardest routing problem I could find, and three years in I had learnt that its hardest part was knowing what was really on the dock.

Reason for the move

Promoted to lead planning analytics for the redesign of the hub network.

Mandate

Make inbound component collection and outbound dealer delivery routes reliable enough for factory schedules.

Outcomes

• Recalibrated routing software with measured loading times at 86 supplier sites
• Raised on-time component arrivals at two assembly plants from 82 to 94 per cent
• Completed a part-time master's in applied statistics

Decisions in context

An assembly line stops if a component collection runs late, so routing for appliance factories had little tolerance for optimism. The routing software we bought produced plans that looked efficient and failed in practice, because it assumed loading at suppliers took a fraction of the real time. I spent weeks at loading bays with a stopwatch and a notebook, recording what happened at 86 sites. Recalibrating the software with those measurements made the plans longer and less impressive, and far more reliable. From 2007 I led four planners and dealt directly with the operations director on vehicle numbers and contracted capacity. The part-time statistics degree came from those arguments; I wanted to understand the methods well enough to question them. By 2010 I wanted a harder problem than milk runs, and less-than-truckload consolidation offered one.

Reason for the move

Joined a growing freight network to work on consolidation planning at a larger scale.

Mandate

Prepare timetables and rosters that matched passenger demand and labour rules.

Outcomes

• Rebuilt schedules for 23 routes using fare-card records from 2000
• Documented the collection date and gaps of every passenger count used in planning
• Reduced unplanned overtime on assigned routes through roster changes

Decisions in context

I joined the bus company straight from my engineering degree and stayed six years. The work was precise and public: a timetable error showed up at a bus stop within days. The overcrowded route of 1998, planned on counts taken before a housing district opened, taught me to record when and how every number was collected. When fare cards arrived in 2000, I helped turn their records into planning data, while noting what they missed, chiefly cash passengers and transfers. I worked under the operations planning manager and had no authority over service levels, which were set with the city. Rosters taught me something as well: a schedule that ignores drivers' rest and shift rules will be quietly rewritten by the people who must work it. By 2003 I wanted to learn routing in a commercial setting, where the customer could leave.

Reason for the move

Moved into contract logistics to work on freight routing.