What an old passenger count taught me

The timetable that taught me most was one I did not write. In 1998 a route in our city was rescheduled on passenger counts taken two years earlier, before a new housing district opened. For a month, buses arrived full at stops where people had been told to expect a seat. I was a junior scheduling analyst at Lanjiangpu Public Transport Company, and my task was to explain to the operations manager why the plan looked correct on paper. The arithmetic was fine. The counts were old, and nobody had written down when they were taken.

I had joined the bus company in August 1997, a month after finishing a degree in traffic engineering at Beichuanlu Institute of Communications. For six years I built running schedules and duty rosters and, from 2000, worked with the first fare-card records the company collected. Those records were more complete than manual counts, but they missed cash passengers and transfers. I learnt to describe what a source left out before presenting what it showed.

In 2003 I moved to Tuoxinmao Supply Chain Services, a contract-logistics firm that collected components for household-appliance factories and distributed finished goods to dealers. Routing software arrived soon after I did. It produced elegant milk-run plans that drivers could not complete, because loading times at suppliers were longer than the system assumed. I spent weeks at loading bays with a stopwatch. That experience made me wary of any optimisation that has never met the place it is optimising. I completed a part-time master's in applied statistics during those years, mostly so I could argue with the software vendor on equal terms.

Qianmoyun Freight hired me in March 2010 as network planning manager. The company then had eleven hubs and a little over RMB1 billion of revenue. Less-than-truckload freight is consolidation work: thousands of consignments of different sizes, gathered by outlets, sorted at hubs and combined into trailers on trunk lines. Every plan depends on knowing how much space and weight is coming. For years we relied on what outlets declared. I led the analytics behind a redesign of our hub network between 2013 and 2018, and I saw how often a declared cubic metre was a guess.

In 2018 I took responsibility for the data platform and its governance. We built a single waybill record across outlets that had previously kept their own systems, and I led our first inventory of personal information before the Personal Information Protection Law took effect in November 2021. That inventory changed my view of the job. Our records described not only freight but people: consignees, outlet staff and thousands of drivers, many of them self-employed owner-operators who had never been asked what we kept about them.

I became chief data officer in March 2022, reporting to the chief executive. I present to the board and to its audit and risk committee twice a year on data security, personal-information handling and the reliability of the figures we give customers. I have no directorship. My team builds models, pipelines and reports, but operating decisions stay with the people who run hubs and trucks, and legal judgements stay with our counsel. I lead through a few standards rather than a long rulebook: every recorded figure has an owner, every model shows its inputs, and every request to share personal data names its purpose. My managers can apply those standards at a hub in Urumqi or an outlet in Wenzhou without calling me first.

The cases on this site reflect the questions I now ask first. When a load model promised fewer departures, I asked how the volume had been measured. When outlets kept drivers' documents on their phones, I asked whether they needed them at all. When a foreign shipper wanted consignees' details abroad, I asked what each field was for and whether an event would serve instead. In each case someone I respect disagreed, and in each case we accepted a cost.

I am now interested in a board contribution. Many companies hold operational data about people who never chose to be in their systems: passengers, drivers, delivery recipients. I can help a board understand what that data can support and what it should not be asked to do. I would also have a great deal to learn about a different company's obligations, and I would expect the chair to tell me when my freight experience does not transfer.

Open to Board appointments

Languages / Mandarin · English