Chief Executive Officer / India
The choices behind a defensible service
- Company colleagues
- 620
- March 2026; three Indian delivery cities
- FY 2026 revenue
- ₹156 Cr
- Company revenue; ₹18 crore operating profit
- Client organisations
- 42
- India, UK and UAE; scoped testing services
Decisions & consequences
01Removing a reward for the easiest findings
A review of nine engagements showed a mismatch between our recognition system and the questions clients needed answered. Team reports celebrated large finding totals, but the underlying records included many minor observations. Investigating a difficult failure path could take several days and produce one consequential finding, while a quick sweep produced a more impressive count. The measure was easy to administer and increasingly poor at describing useful work.
Some delivery managers wanted to retain the count-based component because it appeared objective. They worried that replacing it with professional judgement would create inconsistent recognition and longer review meetings. I accepted that concern but asked them to compare the evidence behind the nine reports. We found examples where important agreed risks had not been examined deeply enough, despite strong totals. I did not turn that small review into a claim about every engagement or the safety of any released system.
We removed raw finding count from that component of recognition and introduced an engagement review based on agreed risk coverage, the quality of supporting evidence and completed retesting. The delivery head owned the criteria; technical reviewers examined the work, and client-facing managers explained the basis to sponsors. I approved the additional review time and accepted that a simple cross-team league table would disappear. The replacement required more thought and could itself be gamed, so we retained periodic examination of actual records.
The immediate result was a different decision process for recognition, not a headline reduction in defects. Teams had a clearer reason to spend time on difficult paths even when the final count was small. We also made unexamined areas visible instead of allowing a large total to imply broad coverage. My lesson was that a CEO cannot ask for rigorous evidence while rewarding a convenient proxy that points elsewhere. The cost of a more demanding review belongs in the operating model, rather than being quietly absorbed by managers expected to meet unchanged delivery commitments.
Funding apprentices before promising their hours
Our proposed apprenticeship intake included thirty-six people. The initial operating plan assumed that most would move quickly into billable assignments, which made the cost look modest. Delivery leaders were less confident: several assignments required judgement that could not be acquired through classroom completion alone, and experienced reviewers already had full calendars. Recruiting the cohort without protected mentoring would create people on the payroll without a credible route to useful responsibility.
The finance lead asked whether a smaller intake would reduce the risk. Commercial colleagues wanted to preserve the original capacity forecast because new engagements were being discussed. I asked the team to cost the learning plan separately from the sales plan. We approved a ₹1.9 crore twelve-month learning and support allocation, including participant support and protected senior mentoring, with six months of supervised practice before readiness decisions. We removed the assumption that an offer accepted was equivalent to a billable place available.
This meant accepting a lower near-term capacity forecast and assigning experienced people to teaching rather than fully charging their time to clients. The programme lead and delivery managers reviewed observed work before approving scoped assignments. At the twelve-month review, thirty-one people had completed the programme and five had left. Twenty-four had entered defined billable work; the other seven completers still needed suitable placements or further supervised experience. We reported all four numbers rather than describing thirty-six recruits as thirty-six new delivery resources.
The programme did not confer an external professional certification, and the placement result did not prove readiness for every type of test assignment. What it gave us was a funded and observable route from recruitment to responsibility. I would assess a later cohort using retention, demonstrated capability, mentoring demand and client fit together. The CEO decision was to bear the real cost of development and correct the commercial forecast, instead of making a promising recruitment announcement that delivery colleagues would have to make true through overtime.
Sharing the benefit when repeat work becomes faster
One recurring regression engagement had been billed at ₹800 an hour. A typical release window required 1,200 repeat-execution hours, producing a ₹9.6 lakh bill. Our automation work reduced that execution effort to 420 hours. Maintenance, environment preparation, investigation and reserved specialist review still mattered, but they were poorly represented by a contract centred on the repeat hours. We faced a choice between allowing the invoice to collapse without a sustainable service model and obscuring the benefit of work we had improved.
The account manager proposed retaining the old effort estimate because the client was accustomed to the amount. I rejected that basis. The delivery head instead separated the continuing obligations from repeat execution and described the release windows for which capacity had to be held. Finance tested the cost assumptions. We offered a fixed-scope service at ₹7.8 lakh per window, including the agreed regression coverage, maintenance allowance and reserved review capacity. Changes to the application or requested coverage beyond that scope required a separate discussion.
The client initially argued that 420 hours at the old rate should determine the whole price. We explained that the earlier hourly model had bundled several obligations into an execution-led estimate and that the new offer needed to state them openly. After negotiation, the client accepted the defined-scope service. Its price was ₹1.8 lakh lower per window than the former typical bill, while we had a funded basis for the continuing work. We did not promise that every future window would fit an unchanged scope or that all automation savings would follow the same proportion.
The decision changed our commercial conversation. The company could benefit from a better method without pretending that fewer hours had not occurred, and the client could see a reduction alongside a more explicit service commitment. The trade-off was a lower headline invoice and greater responsibility for estimating the work we retained. I would not generalise the economics from this single engagement. I would generalise the need to confront a charging model that makes an improvement commercially awkward before it encourages anyone to hide the improvement.
Keeping a qualification when payment depended on it
Near the end of a software-testing engagement, a client sponsor asked for a release endorsement broader than the work we had performed. Our team had examined agreed transaction paths in a defined environment. Several integrations and operating conditions remained outside scope. The sponsor wanted a short positive statement for an internal release meeting and linked payment of a ₹42 lakh milestone to receiving it. The technical lead would support the findings but not the broader conclusion.
The account director argued that the statement would be read in context and that a qualification could damage an important relationship. I had personally sponsored the account, which made my own position part of the problem. I asked another executive to lead the commercial discussion and kept the technical review with the specialists. We issued a report describing the examined conditions, unresolved findings and exclusions. We also offered additional scoped work, with no implication that payment for it would secure a particular conclusion.
The board reviewed the cash consequence and supported retaining the qualified report. The ₹42 lakh payment arrived seven weeks later, after the client's internal discussion and acceptance of the contracted deliverable. We did not claim that our refusal prevented a failed release; the client retained its decision, and we did not have evidence for that counterfactual. Nor was the delay presented as costless. It required treasury planning and management time, while the account relationship became harder to manage.
Afterwards, we revised escalation arrangements so that a CEO-sponsored account did not make the CEO the final arbiter of disputed technical wording. A commercially powerful person should not be able to turn an evidence discussion into a negotiation over loyalty. My responsibility was to support the specialists, make the financial effect visible and improve the process that had allowed the pressure to become concentrated at the end. The experience reinforced a practical boundary: our service can inform a release decision without supplying the authority or certainty a sponsor wishes the report contained.