The behavioural round: going deeper

Competencies: scope, leadership, judgement.

"Routely had five delivery services owned by three organisations, and each wrote its own retry and duplicate-payment guard. Duplicate cash-on-delivery postings ran at about 120 a month across the five, from the finance team's monthly reconciliation report. Nobody owned the problem. I wrote a one-page proposal for a shared library and a single rule for retries, and asked one director to sponsor a six-week pilot on my own team's service. I rejected asking the CTO for a mandate, because it would have made the library the CTO's project, and I needed each organisation to own its change after the pilot ended. I had no authority over the other two organisations, so I made the cost to them small: I sent each a ready change for review and did the testing myself. What I gave up was my team's own roadmap item for that quarter. A year later, four of the five services use the library. I left the fifth alone on purpose, because it is due to be retired. Duplicates fell from about 120 to about 15 a month on the same report. What I would change: I counted postings, not rupees, and I never asked finance for the value."

What the notes say: scope named (three organisations, five services), an option rejected with its reason (the mandate), a cost of the choice, a before and after on a stated source, and a candid gap, which S28 (web) later shows him acting on. Probes: "Who said no first, and what did you do?" and "Why not the fifth service?"

Competencies: hiring, leadership, growth.

"I led a team of five with one open seat. I wrote the design question for our loop and a four-line rubric before we saw anyone. We interviewed eleven people. I argued against one candidate who solved the coding task fast but waved away the pairing exercise, because the seat needed someone who works in other people's code. We hired an engineer from a smaller services firm whose algorithms were weaker and whose debugging was the best I saw. I gave her a 90-day plan with two named milestones. A year later she owns the on-call tooling and led a postmortem last quarter, which my manager has since asked her to repeat for another team. What I got wrong: in her first three weeks I gave her too little, and she had to ask for harder work."

What the notes say: a bar defended before a name was known, a rejection with a reason, and a twelve-month outcome. Probes: "What did the rejected candidate do well?" and "What if she had not worked out?"

Competencies: people management, ownership, judgement.

"For eleven months I was acting manager of a team of six. One engineer had missed three consecutive commitments over two quarters, and two incidents traced to changes that nobody had reviewed. I spoke to him within a week of the second incident, with the specific examples written down, and asked what was in his way. Part of it was my fault: his expectations on one project had never been written. I fixed that. When the pattern held, I agreed a written eight-week plan with HR and my manager: three measurable goals, a weekly check, and his words on whether they were fair. By week six two of the three were missed. He left through an HR-led exit with an agreed handover, and he has since joined another company. The team met its sprint commitments in seven of nine sprints over the next two quarters, against four of nine before. What I would change: I waited about six weeks too long before the first conversation."

What the notes say: evidence before action, the person's view heard, process followed, the outcome for the person as well as the team, and a candid delay. Probes: "What did he say the problem was?" and "What would you have done had he improved?"

Competencies: ambiguity, judgement, scope.

"Leadership said: make deliveries predictable. There was no metric. I wrote a one-sentence reading and sent it to the director: for urban parcels, the promised date should be right within one day for nine in ten parcels. Our share was 62%. My riskiest assumption was that the misses came from the last-mile handover and not from warehouse dispatch. The cheapest test was two weeks of scan data for 8,000 parcels. It showed that dispatch delay explained most of the misses. That changed the plan: we spent the quarter on dispatch estimates, not on route planning. I set a revisit at week six, when the share was 71% on the same definition. A year later it is 78%, short of the 90% I wrote down, and that miss is mine. Most of the remaining misses were parcels handed to two partner carriers, and closing them meant reopening their contracts. I priced that and took it to the director; she chose to spend the next quarter elsewhere, and I agreed and wrote 78% down as the level to hold until then. What was wrong: I had assumed the handover, and I set 90% before I knew where the misses were."

What the notes say: a written assumption, a cheap test, a changed plan, a revisit date, the number labelled with its definition, and the result compared with its own target: the 12-point miss owned, with the reason he stopped and who chose. Probes: "What if the director had disagreed with your reading?", "Why stop at 78%?" and "What did you stop doing?"

Competencies: influence, conflict, judgement.

"A platform team in another organisation refused to add a field my service needed. Their roadmap was full, and their best argument was fair: each new field is another thing they are paged for. I looked for what they would gain. Our heavy lookup was about 30% of their API traffic on their own dashboard, and the field would let me remove it. I offered to write and maintain the change behind a flag. They still said no for the quarter. I told their lead before I went up: on Thursday I would send both directors a one-page note stating both positions and asking them to decide. I did, and the directors approved the change for the next quarter with my team carrying the on-call for the field. I accepted that. A year later their lead asked me to review their API redesign, and the lookup is gone. What I would change: I opened with my ask, not their traffic problem; I now start from the other side's problem."

What the notes say: the other side's argument stated fairly, a trade, notice given before escalating, a decision taken by the right level, and the relationship a year on. Probes: "What if the directors had sided with them?" and "Did anyone think you went around them?"

Competencies: ownership, growth, judgement.

"A migration I approved cost our finance team about six person-days and made a partner's invoice a week late. For four hours one morning about 2,300 settlements posted with a zero fee. Finance reversed them by hand over three days, about six person-days by my estimate, and one partner's invoice went out a week late. My part: I approved a schema change to the settlements table without rehearsing it on a copy. I had assumed our staging data was close enough, and I did not know the change dropped a column default. A finance analyst found the zero fees in her midday check, and I told finance the cause that afternoon. I rejected a quiet fix to my own review habit, because a habit fails the next time someone else approves. I changed the system instead: money-table migrations now need a rehearsal on a copy, checked by a second person, and a CI check blocks any migration that changes a default. In the next twelve months there were nine such migrations, all rehearsed, and none caused an incident."

What the notes say: the four bold lines told in order. Impact: 2,300 settlements, about six person-days labelled as an estimate, a late invoice. His part: the decision, what he assumed and what he did not know. Prevention: a change to the system, with the quiet fix rejected. Proof: nine migrations over twelve months. Detection and cause get a clause each; the rest waits for a probe. The one-minute cut near the end of the lesson keeps the same decisions and numbers. Probes: "What did you know when you approved it?", "How was the 2,300 counted?" and "Who objected to the new rule?"