Thirty-two years inside China's life insurance agency model

Mandate

Rebuild HR for a listed life insurer whose agency was beginning to contract, and give the board people figures it could rely on.

Outcomes

• Replaced allowance-led agent recruitment with selection; thirteen-month retention of new agents rose from 9% to 31%.
• Removed forced ranking for 4,700 verification and control staff after falsified return-visit calls; genuine-contact sampling rose from 83% to 98%.
• Grew qualified actuaries from 22 to 39 with six external hires, cutting external actuarial support from RMB 31m to RMB 6m a year.
• Introduced a six-year tenure limit for provincial general managers; ten rotated by mid-2026 and expense findings fell from 7.6 to 2.2 per audit in their branches.
• Created a salaried training-lead track in 2023 for 640 former agency team leaders.

Decisions in context

I joined Songbai weeks before the agency contraction became visible in our figures. My first task was ordinary HR modernisation: one grading structure for head office and branches, consistent labour contracts and a single HR system. By the end of 2020 the agency had fallen by almost a third, and the president asked me to work with the head of the agency channel on what kind of agency we wanted. Since then the work has widened. I led the move from recruitment to selection, the response to the return-visit audit in 2021 and the end of forced ranking in verification roles, the actuarial programme that accompanied IFRS 17 and the solvency changes, and the tenure limit for provincial general managers. I prepare papers for the remuneration and nomination committees and attend their meetings except closed sessions. I am not a director of Songbai.

Reason for the move

Current appointment; completing the team-leader career path and the second round of branch rotations.

Mandate

Professionalise HR for a growing joint venture and build a selection-led agency in four new provinces.

Outcomes

• Opened four provincial branches with agencies built through selection; thirteen-month agent retention of 36% against 14% in the older branches.
• Reduced salaried staff attrition from 24% to 15% through a published grading structure and internal vacancy rules.
• Introduced deferral of 40% of senior managers' annual bonuses over three years, with recovery terms approved by both shareholders.
• Prepared the board remuneration committee's papers from 2016 to 2020.

Decisions in context

I left a state-owned national insurer for a joint venture because I wanted to see whether an agency built on selection could work in China at all. Brennmark Xinghe Life had two equal shareholders whose instincts on pay, hierarchy and risk differed, and every significant HR proposal needed both to agree. I learnt to write one-page papers that set out the decision, the alternatives and the objection each shareholder was likely to raise. When the company opened branches in four new provinces, the general manager agreed that we would recruit fewer agents, pay for a longer preparation period and measure each new branch by the retention of its intakes. The results were good enough to convince me, and small enough that I was not sure they would hold at the scale of a large insurer. Songbai offered the chance to find out.

Reason for the move

Recruited by Songbai to lead HR at a larger listed insurer.

Mandate

Set recruitment and training standards for the agency across 31 provincial branches and build a development path for agency managers.

Outcomes

• Built a three-level manager-development programme completed by about 41,000 team leaders.
• Introduced a national induction curriculum with a module on explaining participating policies.
• Helped lead recruitment campaigns that grew the agency from 480,000 to 760,000, with thirteen-month retention below 15%, a result I now count against the work.
• Commissioned the company's first cohort study of agent income and departure.

Decisions in context

Moving to head office put me inside the engine of the agency model at its fastest. My division set the standards each province used to recruit and train agents and ran the programmes that turned successful agents into managers. The manager-development programme was the best of that work: it gave tens of thousands of team leaders a structured grounding in coaching, compliance and running a team. The recruitment campaigns were the part I now regard differently. Every province was measured on headcount growth, and my division supplied the materials, events and allowance rules that drove it. I commissioned a cohort study in 2013 because the thirteen-month retention figures troubled me, and its findings matched my MBA dissertation: most agents left because they could not earn a living. The company noted the study and kept the campaigns. That gap between evidence and practice is why I left.

Reason for the move

Recruited by Brennmark Xinghe Life to lead HR in a joint venture building its agency on selection.

Mandate

Establish a professional HR function for a large provincial branch after the company's conversion into a joint-stock company.

Outcomes

• Moved 3,600 employees onto consistent written labour contracts and a single grading structure.
• Reduced first-year attrition among branch service staff from 31% to 18%.
• Coordinated the deployment of 140 staff into earthquake-affected counties in 2008 and support for colleagues' families.
• Developed eleven city-office managers later promoted to provincial leadership roles.

Decisions in context

This was my move from agency training into general HR, and the scale was a shock. The branch had recently become part of a joint-stock company, and its employment practices still reflected older arrangements: contracts that varied by city office, grades that meant different things in different places and promotion decisions recorded nowhere. I spent my first two years making those things consistent, working with the employee representative congress on each change. I also learnt how far a provincial branch's people decisions depend on its general manager, a lesson that later informed my view of tenure. In May 2008 the earthquake in the north-west of the province changed the work for a year. We deployed staff from unaffected cities to support claims and service, arranged leave and help for colleagues who had lost relatives, and learnt how much a temporary posting asks of a family.

Reason for the move

Promoted to head office to lead agency talent and training nationally.

Mandate

Build agent training for a fast-growing provincial agency force.

Outcomes

• Rewrote the induction course with experienced agents and a claims manager, adding sessions on participating policies and declined claims.
• Trained about 26,000 new agents over five years.
• Introduced a three-month field-coaching period for new agents in Hefei and Wuhu.
• Became head of agency recruitment and training for the branch in 2002.

Decisions in context

I came to agent training from statistics, with no experience of selling, and spent my first months accompanying agents on appointments in cities and villages across Anhui. What I saw changed the induction course. Customers asked most often about what a participating policy would pay and what would happen if they needed to claim, and our course taught neither well. I rebuilt it with two senior agents and a claims manager, added a session in which new agents heard from policyholders whose claims had been declined, and introduced field coaching so that a new agent's first appointments were observed. In 2002 I took on recruitment as well as training, which meant running the branch's January recruitment events. I ran them as I had been taught to, and the retention figures I collected afterwards began the questions I still work on.

Reason for the move

Appointed director of human resources for the Sichuan branch.

Mandate

Prepare the branch's statistical returns and, from 1996, measure the new agency force.

Outcomes

• Defined three agency counts, contracted, trained and active, for the branch's monthly report to head office.
• Built the branch's first agent-level persistency report in 1998.
• Showed that 60% of agents contracted in 1997 had made no sale by the end of that year.

Decisions in context

My first job was arithmetic: premium, claims and expense returns for a provincial branch of a state-owned insurer, prepared by hand and then on the branch's first personal computers. In 1996 the branch launched an individual agency force, and measuring it fell to me. The question of who counted as an agent turned out to be contested, because the branch's standing with head office depended on the answer. I proposed reporting three figures side by side, contracted, trained and active, and head office eventually adopted the format for the province. In 1998 I built a report linking each policy's persistency to the agent who sold it, which showed for the first time how differently policies behaved depending on how long their agent had been working. That report was the reason the branch asked me to run training, and it is the reason I still trust cohort data more than totals.

Reason for the move

Asked to run agent training for the Anhui branch.