I checked the arithmetic and missed the implication

I signed off the first version of the callback paper because its total reconciled. The combined rate had moved from approximately 75.4% to 73.3%, and the proposed heading described a deterioration in service. Pradeep Qelmavi asked to see the two channel populations separately. Both conditional rates had improved. What had changed sharply was the share of requests using the assisted channel, whose rate remained lower. The total was a valid description of the observed mix. It was not a sufficient explanation of what had happened inside either channel, and it certainly did not identify the cause of a change in performance.

I grew up in Kanpur and completed statistics at the fictional Vorzemi Statistical College in June 1997. In July I joined Nelqavi Market Research as a field statistician. The company employed seventy people and recorded ₹9 crore revenue for FY ending March 1998. I worked in a six-person research group with no direct reports or spending authority. Supervisors checked the sampling plan and interpretation. The work involved people who answered questions, people who did not and commercial teams who wanted a clear conclusion from both. I learned that a denominator was a decision about whose experience a number described.

In June 2003 I became sampling manager. Nelqavi then had 150 employees and ₹24 crore revenue for FY ending March 2004. I led ten colleagues through two supervisors within ₹72 lakh annual function expense. The role made me responsible for explaining what a study could reasonably support before its result became a business forecast. I could improve field discipline and investigate a response pattern, but I could not make an unobserved preference known by filling a blank. A clean dataset was useful only when the meaning of its entries remained faithful to the evidence behind them.

I moved to Talqeri in September 2009 as head of customer-data stewardship. The insurer had 310 employees, eight branches and ₹420 crore gross written premium for FY ending March 2010. Fourteen colleagues reported through two leads within ₹1.4 crore annual expense. I wanted to work with an account that had to remain useful after a project presentation was over. Underwriters, service colleagues and finance teams could legitimately need different views of the same business. My task was to connect those views without pretending that a shared data office acquired their professional decision rights.

The enterprise-measurement director appointment began in July 2015. It was a functional title without a statutory directorship. Talqeri had 590 employees, thirteen branches and ₹980 crore gross written premium for FY ending March 2016. I led twenty-six data colleagues through three heads with ₹4.1 crore annual expense. I became more involved in the way management measures influenced a conversation: a branch comparison, a vendor purchase or a description of progress. The difficult part was often an implication added to a correct number rather than a defect in the number itself.

I became CDO in January 2021. My present remit covers forty-three employees through four heads, with ₹8.6 crore FY2026–27 operating expense and a separate ₹2.4 crore data-improvement programme. The insurer now has nineteen branches, 920 employees and ₹1,860 crore FY26 gross written premium. CEO Saira Dazquri is my reporting line, and Pradeep leads data-quality stewardship. Underwriting, actuarial, claims, finance, technology and legal colleagues remain accountable for their respective decisions. A data recommendation may inform a premium or service question without giving me authority to decide it.

The callback correction cost ₹52 lakh. We kept both the channel account and the actual combined experience, including the requests that missed their promised slot. A later ₹48 lakh exercise distinguished 420 missing turnover entries from ninety explicit zero declarations within 2,400 reviewed files. It obtained confirmation for 312 missing entries and left 108 unresolved. That was evidence about what was recorded and subsequently confirmed, not a finding that every business declaration was accurate or an instruction to accept or decline cover.

Two other decisions changed what the company would distribute or purchase. We withheld a proposed partner table after review showed that exact totals could reveal a suppressed cell. A ₹34 lakh redesign produced a narrower authorised release; it did not establish that every future combination would be safe. We also spent ₹27 lakh evaluating a proposed external business dataset and declined its ₹1.6 crore annual contract. Only 36 of 480 reviewed records met the agreed criteria for a new field useful in the current workflow. The vendor might still have supported another purpose. It did not follow that this purpose justified this purchase.

I have no statutory board seat and no personal shareholding in Talqeri. My fictional IICA record shows registration in August 2026, with assessment preparation unfinished and no exemption claimed. Twenty-two days is a provisional annual allowance for one outside appointment, subject to employer consent and the actual calendar. I work in Hindi and English, with working Kannada developed in Mangaluru. I would bring experience in examining the inference a management number invites. I would also bring the memory that I approved a misleading service heading myself. A first board role would require learning how to ask for that explanation without taking over the data team's work.

Open to Board appointments

Languages / Hindi · English · Kannada — working