Chapter 7 · The Hard-to-Replace Professional

Extended content

Liquid part: this text keeps growing after the printed edition.

The Friday call

Open your calendar for the past six months and find the times someone called you because something had gone sideways. Three is enough. Next to each, write what that person asked you for, exactly. Rarely was it information. Almost always it was a reading: "Is this serious?", "What do we touch first?", "Do we tell the client today or tomorrow?"

That list is the real inventory of why you are hard to replace, and it tends to resemble your résumé very little. Add a second column: how many hours last week went to that kind of work. The distance between the two columns is the afternoon's finding.

A case: Diego and the clean scripts

Diego coordinates data integrations at a mid-sized insurer. His team has used a model to generate the scripts that move billing from an old system to a new one. They come out tidy, commented, with tests. The cutover to production is set for the last day of the month.

Diego stops it. He couldn't yet say why. He has been looking at the calendar for a while and something rings a bell: an old month-end, a spreadsheet of surcharges that someone in sales updated by hand every Friday and on which the price of one product depended. He calls Paco, who has held that job for fifteen years. Paco still does it. The sheet appears on no diagram, and the scripts, correct line by line, would have left it with nowhere to go.

Diego didn't beat the machine on speed. He exercised three of the integrator's functions at once: filtering, because out of everything the model delivered he knew what deserved suspicion; validation, because a correct script is not the same as a correct cutover; calibration, because he trusted the model for the syntax and himself for the calendar. Then he wrote the finding down, so that Paco stops being the only place it lives.

Writing a scar in one line

Pick a decision from your past whose consequences you now know. Note what you decided, what happened, and what signal was there at the time, whether you saw it or let it pass. Close with a one-line rule: "When I see this, I ask that."

Do it with three decisions. They are your first reflexes put into words. You will be able to teach them to a junior and, if you like, add them as a check when reviewing an output. Not all of your tacit knowledge fits in one line, and that's as it should be. What does fit frees up attention for what doesn't.

Questions to work with your matrix

• Which function in the upper-right quadrant do I perform without its appearing in my job description?
• Which hour of my week, in the lower-left quadrant, could I hand to a model starting tomorrow, with my review?
• What do I know about this system that only one other person knows, and what happens the day they retire?
• If I'm called in the middle of a crisis, would a model ask my first question too?
• Who is learning from me what I learned the hard way?

Twenty minutes with a veteran

Ask the person with the most years on your team for twenty minutes and a single question: what have you seen fail that isn't written down anywhere? Take notes on what they smelled, more than on what happened. Whoever has been inside a system for a long time usually carries half a dozen stories like this, and almost nobody has asked for them.

With three conversations of this kind you have the draft of a channel for tacit knowledge. It is the part of the organization no manual covers and no model can inherit from the manual. If a year from now those notes are still in a drawer, the exercise hasn't worked; if someone used them to decide something, it is already functioning as mentorship.

Where to next

The matrix tells you where your value sits. What remains is what it is worth out there: the energy that keeps the tools running, the interests of the people who sell them, and the conditions of the board you play on. That is the subject of Chapter 8.