An accurate record can carry the wrong conclusion

The result looked like a price rise until the composition of the two groups became visible. In our selected 2023 example, the older and newer vehicle bands retained means of ₹8 lakh and ₹16 lakh. Yet changing the mix from fifteen older and five newer vehicles to ten of each moved the overall average from ₹10 lakh to ₹12 lakh. I wanted the publication to explain that arithmetic without claiming the example had solved every quality difference in the market.

I grew up in Kochi and completed a Bachelor of Science in Statistics at the fictional Nerpilvi Statistical College in May 1993. The following month I joined Pernavli Data Studies as an assistant in an eight-person research group. Senior colleagues approved the studies. My job was to preserve what the source actually recorded, including the distinction between a quoted amount, an observed transaction and an outcome that had not been established.

In April 2001 I became a research analyst leading eleven colleagues within ₹1.3 crore annual practice expense. Clients often wanted a clear answer from records collected for a different purpose. I learned to ask what an absence meant before assigning it a commercial interpretation. A record disappearing from a source could be a useful observation without revealing why it disappeared. That lesson was available to me well before the later error in my own publication approval.

I joined Kershavi Vehicle Intelligence in April 2009 as data manager, responsible for seventeen colleagues and ₹2.6 crore annual function expense. The work connected vehicle listings, supplied transaction evidence and customer research questions. I could follow an interpretation through to the material a subscriber received. The challenge was to retain the distinction between the source event and the conclusion, even when a simple market headline was easier for a commercial colleague to explain.

I became head of data in April 2015, with twenty-three colleagues and ₹4.2 crore annual expense. In 2018 I approved an account treating thirty-six removed listings as completed sales. Farhan’s review found twelve confirmed transactions and twenty-four outcomes that remained unknown. We spent ₹33 lakh over eight weeks correcting the account and its approval basis. I could not defend the original conclusion by pointing to accurate removal records, or replace it with the equally unsupported claim that the other vehicles had not sold.

The CDO appointment followed in April 2021. Four heads now coordinate twenty-nine colleagues within ₹5.8 crore annual expense, including ₹1.6 crore methodology development. Anita Desvarki retains enterprise authority and Farhan Telnashi leads research methods. Their challenge is part of the account, because a data leader should not describe every correction as an insight personally discovered after others had already supplied the evidence needed to change it.

Our 2023 composition review cost ₹37 lakh over nine weeks. It explained why the selected group means differed, but did not produce a complete constant-quality index. Two years later, a ₹54 lakh, ten-week review separated sixteen single-vehicle deals from two bundled deals in an eighteen-deal account. One bundle contained twelve vehicles for ₹1.44 crore. An equal division of that total was a calculation, not twelve observed prices. We retained the bundle’s existence while excluding it from the single-vehicle comparison.

I hold no personal equity and no statutory appointment. An outside contribution would require further preparation and a company-specific assessment, with twenty-three proposed annual days. I would bring experience of asking how a correct calculation becomes an unsupported commercial conclusion. The work is useful only if the distinction remains visible to someone who did not prepare the analysis and has a real decision to make from it. I also want the reader to see what our evidence cannot resolve, because an honest unknown is more informative than a category assigned simply to make the report look complete.

I now ask the team to identify the person who will encounter the finding without its working papers. That may be a subscriber comparing periods or a commercial colleague explaining why a revised account is narrower. If that person cannot tell whether a figure is observed, modelled or allocated, our internal clarity has not reached the product. The final wording deserves the same care as the calculation because it determines which decision the evidence appears to support. That responsibility is one reason I remain interested in leading research inside an enterprise rather than treating publication as someone else’s final step.

Open to Board appointments · Advisory roles · Executive opportunities

Languages / English · Malayalam · Hindi