Chief Human Resources Officer / India
Count the unfinished work beside the decision
- General-insurance employer
- 520 employees
- Eighteen branches; ₹1,120 crore FY26 gross written premium
- People function
- 28 HR colleagues
- Four heads; ₹4.9 crore FY2026–27 function expense
- Preparing for a first appointment
- 0 board seats
- Executive presentations are not statutory directorships
Decisions & consequences
01Fourteen specialists progressed without acquiring artificial teams
In 2023, managers submitted 24 roles for progression review. Our grade framework expected direct reports above a particular level. Several managers therefore proposed moving colleagues beneath strong technical contributors, although the work did not require a new supervisory layer. I had supported that framework as head of people. Its apparent consistency was now creating a choice between holding back a specialist and altering the organisation chart to justify recognition.
Irene Zolmavi argued for clearer specialist progression, while finance wanted to avoid an uncontrolled second route to higher pay. Both concerns were reasonable. A specialist title without defined contribution could become a reward for tenure, and parallel routes could leave managers feeling their responsibilities were undervalued. We asked technical leaders to describe the judgement, complexity and contribution expected at each proposed level, then examined the roles individually. HR facilitated the process; it did not confer underwriting or claims authority through a grade decision.
The review cost ₹16 lakh in role-design and assessment work. Fourteen roles moved into specialist appointments, six followed genuine management routes and four remained in development. The annualised pay movement was ₹68 lakh. We did not describe that recurring expense as an immediate productivity saving. Managers also had to explain why some colleagues progressed and others did not, without publishing private assessment material or implying that management work was less valuable.
The arrangement removed the need to manufacture direct reports for the fourteen specialist appointments. It did not show that all retention risks had disappeared or that each person would meet the expectations of the new role indefinitely. Technical permissions remained separately governed, and performance discussion continued. The four development cases were an important part of the result: a more suitable route still required evidence of readiness. My own lesson was that a framework can reward the wrong organisational behaviour even when its stated purpose is fairness. I would now examine the work people must invent to comply with a rule before presenting consistency as sufficient proof that the rule is sound.
An unanswered employer was not a finding against the candidate
A 2024 recruitment review identified 42 candidates whose employment checks could not be completed through the former-employer contact normally used. Some organisations had closed, records had moved or the available contact could not confirm the relevant period. The original process held every case at the same incomplete stage. Managers wanted vacancies filled, while the verification team worried that changing the route would weaken the standard. Candidates received repeated requests without a clear explanation of what fact was still missing.
I asked the team to separate the fact required from the familiar method of obtaining it. With appropriate advice and consent, we identified alternative employment records and corroboration suitable for the particular gap. Essential qualifications and other role requirements were not waived. We also kept verification apart from the broader selection decision: establishing that someone had worked somewhere did not prove that the person was suitable for our vacancy. The review required ₹14 lakh in additional verification and process work.
Of the 42 cases, 28 were resolved through the accepted evidence route, nine remained unresolved and five candidates withdrew. Seven of the 28 were subsequently appointed after the other selection steps; the remaining 21 were not selected for those vacancies. Those counts were reported together. Claiming 28 hires would have confused a completed check with an employment outcome, while treating the nine unresolved cases as dishonesty would have asserted more than the evidence supported. We retained the uncertainty and explained the next available process to affected candidates.
The revised approach required more judgement than repeatedly sending the same request. It also exposed differences between cases that a single status label had hidden. Managers could not simply instruct the team to mark a difficult file complete, and candidates could not be promised an offer in exchange for supplying another document. The useful result was a more relevant account of what had been established. I would defend the standard of evidence while remaining willing to change the route used to obtain it, especially when the original route depended on a third party no longer able to respond.
The useful filter also made a colleague recognisable
Our 2025 employee-feedback exercise received 368 responses from 520 eligible employees. Managers asked for team, location, tenure and role views so they could act on specific concerns. I supported additional filters because broad averages could conceal a difficult local experience. The first report review showed a problem with my request: combining otherwise acceptable views could identify a small group or make a particular respondent recognisable. Removing names did not remove that inference.
There were 46 teams in the reporting structure. Eighteen had fewer than five responses, so we suppressed their team-level reports. The remaining 28 were not automatically released without further examination. We reviewed combined filters, distinctive circumstances and comments that could reveal an individual. Wider aggregation and fewer views were used where needed. Identifiable free text was excluded from dashboards, and sensitive individual concerns needed an appropriate separate route rather than being exposed through a manager's survey view.
The reporting and privacy review cost ₹22 lakh. Some managers objected that a less detailed report made it harder to know what to change. That was a real cost, not evidence that the concern was unimportant. We supported broader discussions about working conditions and asked managers to distinguish an invitation to improve from an attempt to discover who had answered. The survey explanation also needed to describe the limits of confidentiality honestly. We did not make an absolute anonymity claim simply because one suppression rule had been applied.
The result was a smaller set of reports and a more cautious use of employee comments. It did not prove that every respondent felt safe or that no inference could ever be made. Future exercises would need renewed review as teams and available data changed. I took responsibility for requesting detail before considering how the views interacted. The case changed my definition of an actionable report: it should support a responsible decision, not provide every possible clue about the person who raised the concern. More precise data is not automatically a more appropriate management instrument.
Returning to the payroll still required returning to meaningful work
In 2026, nine employees were preparing to return after approved extended leave. Their circumstances differed, and the employment arrangements required individual discussion. In several teams, duties had been redistributed or temporary cover had become part of the operating routine. A return date alone did not establish the assignment, the manager or the capacity needed to support a workable transition. I asked the business to describe the actual work available rather than report the cases as complete once an administrative status changed.
We worked with each employee, relevant managers and appropriate advisers, keeping private circumstances out of general staffing presentations. The review considered duties, grade, location and the terms of any temporary arrangement. It did not presume a health condition or give managers permission to seek unnecessary personal history. Applicable obligations required proper advice; we did not use a generic HR process as a substitute. Qualified insurance leaders retained responsibility for technical duties and any permissions needed to perform them.
We funded ₹36 lakh of overlap and management capacity. Six colleagues moved into comparable-grade assignments, while three remained in temporary arrangements under review at the October 2026 reporting date. Those three cases continued to have named owners and individual discussions. The cost allowed existing teams to meet their commitments while transitions were arranged, but it was not evidence that every placement issue had been resolved. Nor did we claim improved health or long-term retention from the immediate assignment outcome.
The competing pressure was to avoid paying twice for work while temporary cover remained. I understood that concern and asked for the overlap to have a defined purpose, rather than assume additional cost was always the right answer. In these cases, refusing any transition capacity would have made a credible assignment harder to establish. The board received the six completed assignments, three unfinished arrangements and full cost together. My lesson was to report a return through the work actually provided and the uncertainty still carried, instead of allowing an administrative milestone to stand in for a person's practical experience.