Notes

Working notes. Short, opinionated, occasionally wrong.

Written after engagements, mostly on the train back from Chakan. We publish them whole rather than as a newsletter with a signup wall, and there is no subscribe form on this page.

  • Note 07 · July 2026

    Read your last twelve exit conversations in one sitting

    Not one a month as they arrive, which is how they get filed and forgotten. Print twelve, strip the names, sit down for an hour with nothing else open. The patterns arrive in about twenty minutes and they will not be the ones your engagement survey flagged. In our casework they point at three or four specific managers, one broken handover between functions, or a single promotion decision that everybody watched and drew conclusions from. Do this yourself before you hire anyone to do it for you.

  • Note 06 · June 2026

    Nineteen behaviours became four, and now they get used

    We spent a Thursday reading a client's competency framework aloud to their engineering managers, asking for one real example per behaviour from the last quarter, with a name. They managed four out of nineteen. The framework is now four behaviours long and appears in every calibration meeting, which the nineteen-behaviour version never did. The test is cheap and anyone can run it. Most frameworks fail it and most HR teams already suspect this.

  • Note 05 · May 2026

    Write the governance before you touch the appraisal form

    Two paragraphs: how a rating gets decided, and who can overturn it. Most companies design the form first, discover mid-calibration that nobody agreed who breaks a tie between a delivery head and a practice lead, and lose the cycle to a four-hour argument in a conference room in January. If your founder can overrule the committee, write that down. Pretending otherwise teaches every manager in the room that the stated process is decorative.

  • Note 04 · April 2026

    Skip-levels are cheaper than any retention bonus

    The manager is usually the trigger for a resignation. What turns friction into a letter is that there was no second person to talk to. A skip-level that actually happens — on a calendar, twice a year, forty-five minutes, with no agenda item called "feedback about your manager" — costs a leadership team maybe eight hours a quarter. Compare that to the cost of replacing one senior engineer in Kharadi at current notice periods and market rates. The maths is not close.

  • Note 03 · March 2026

    The feedback already exists; it is just badly routed

    When a company says feedback does not happen here, it happens constantly — in the corridor, in a WhatsApp group with no manager in it, in the car pool to Wakad. It never reaches the person it is about and it never reaches the record. Building a new tool on top of that changes nothing. The work is routing, and it starts by asking managers what they already know about their team that they have never said to that team member's face.

  • Note 02 · February 2026

    Against the annual survey, at companies under two hundred

    An opinion we hold briefly and repeat often: at under two hundred people, the annual engagement survey is a way of avoiding conversations that would take an afternoon. It converts a set of specific, nameable problems into a score that can be discussed in a leadership meeting without discomfort. That conversion is the product being purchased, whether or not anyone says so. If you must run one, at least commit in advance to naming a manager when the data points at a manager.

  • Note 01 · January 2026

    What manager training cannot do

    Six half-days will not fix a manager whose incentive is to hoard the good work, and it will not survive a boss who cancels one-on-ones when the quarter gets tight. We have watched two cohorts decay inside a quarter for exactly that reason, which is why the layer above now sits a session too. Be suspicious of anyone, including us, who describes manager training as an intervention rather than one input among several.

We write these because it keeps us honest about what we have actually seen, versus what sounds true in a proposal.

Notes appear roughly monthly, usually after an engagement closes and a claim we made turned out to be only half right. If you want to argue with one, the email below reaches a person and not a queue.