Five pieces this week, and every one of them turned out to be about an instrument. Not a broken instrument either, which would be an easy problem, but a working instrument aimed at the artifact instead of the thing that produces it.
Reuters published its investigation into Meta's Project OT on Wednesday, and buried in it is the cleanest illustration of this week's problem that anyone could have handed me. Organization Transformation started at a January leadership retreat and set out to test whether AI agents could absorb the daily work of thousands of employees, leaving smaller pods of human builders to supervise them. Executives explored shrinking many teams by as much as sixty percent, in two waves.
Then the internal numbers arrived. Code changes to the internal software platforms and infrastructure Meta employees use on the job were up two hundred and twenty percent year over year, according to a June post by the company's chief technology officer. Changes that actually reached users as new or upgraded features were up thirty-six percent. Major technical and security incidents, including service disruptions and possible data leaks, rose forty percent, and the time staff spent firefighting them rose seventy percent. Amid an open employee revolt, Meta's half-year Pulse survey put favourable sentiment at fifty-five percent against seventy-four before it. Hours before the first layoff wave in May, Zuckerberg called off planning for the second.
Every one of those numbers is real. Only one of them was ever the point. Code changes is an activity count, and activity counts are the easiest thing in any organization to move, which is exactly why they keep getting chosen. Multiply the count by three, ship a third more, and if the count is what reaches the executive dashboard then the operation looks like it is working right up until somebody reads the incident line.
That is the habit I circled all week from five directions, in a policy document, a review cycle, a pay grid, an adoption statistic and a culture score. Meta had instrumentation covering every one of them. The instruments were working. They were pointed at the half of the machine that is easy to count.
On buying honest counsel from someone you pay
Spend a morning with a team and then an afternoon with the executive who signs the invoice, and you will notice the two conversations pulling in different directions. In the morning people tell you what is actually happening. In the afternoon you are asked how it is going, by somebody who has already told you what they believe is happening, and who is paying for the answer.
Almost every executive coach, consultant and advisor works inside that arrangement. It is not a character test and I would rather not pretend it is one. It is a structure, and structures win. The person you hired to tell you the truth sits on both ends of your hierarchy, often in the same week, and nothing about how they get paid rewards them for handing you the harder version of what they heard downstairs.
A graduate module on influence put a principle in front of me this year that I have not been able to put back down. Do not say something up the hierarchy that differs from what you would say down it, because when people find the inconsistency, and they do find it, what you lose is not the argument. It is the standing to make any argument at all. The principle is aimed at the person climbing. It applies with more force to the person selling.
I felt the pull of it the first time somebody senior asked me a question I had a genuinely unwelcome answer to, and I noticed how many softer versions arrived in my head before the accurate one did. The softer versions are not lies. That is what makes them dangerous. Each one is defensible, each one is a fair reading of some part of what you saw, and the collection of them adds up to a picture that will not survive contact with the organization. The moment I hand a leader that version, I have stopped being a coach and become a courtier.
What makes it hard to catch from the buyer's side is that a courtier is pleasant to work with. The engagement feels productive. The reports are good and the meetings run on time. Nothing appears to be wrong until a problem you were never told about arrives at a size that makes it somebody else's news, and at that point the advisor is usually gone and the invoices are paid.
There is a symbolic layer sitting on top of the money that does more damage than the money does. Whose office does the coaching happen in. Who walks the advisor into the building on the first day and how they introduce them. Whether the organization describes that person as the chief executive's coach or as the team's. Every one of those is read by everybody, immediately, and they decide what people will risk saying in front of the advisor. Get them wrong and you have bought an expensive instrument that only reads what you already believe, which is the same failure I have spent this week describing in engagement surveys and pay grids.
So the useful question is not whether your advisor is honest. Assume they are, and assume the structure will grind them down anyway. The question is what you have done to make honesty affordable for the person you are paying.
A few things actually help. Say out loud, early and in front of other people, that you expect to hear things you will not enjoy, and then let the first uncomfortable finding land without defending yourself, because the room is watching that moment more carefully than anything you said. Pay for a defined piece of work rather than an open-ended relationship, so that continuing the engagement is not the advisor's main incentive. Put somebody other than you in the room for the findings. And ask, plainly, what they are hearing that they have not told you, then sit through the silence rather than filling it.
None of that guarantees candour. What it does is stop you being the reason candour was expensive.
I am aware of the position this puts me in, writing it. I am one of the people being described, and the argument runs against my own commercial interest, which is roughly the point. If you are paying somebody for judgement and every incentive around them rewards agreement, the fault when you get agreement is not really theirs. So: when did the person you pay for honest counsel last tell you something that genuinely annoyed you, and what happened in the room right after they did?
This week's rotation carried Otis Florence, Episode 5, the one we called The Business He Sold at a Bus Stop. Otis has been working for himself since he was eighteen, and the deal that redirected his whole career started with a stranger who had just stepped off a city bus and struck up a conversation on the corner. He is blunt about what most sellers get wrong, and it starts earlier than anyone expects.
Listen →
I don't just talk about Agile and leadership, I put it into practice in my Skool, building and shipping real products with Claude every Thursday at 12:30pm EST.
Join us →Thursday's piece on automation paying only where the steps touch each other has a cousin in chapter three of Agile Sucks! (When You Do It Wrong). James and I walk through a media technology client that spent over a million dollars on a way of working nobody had laid out properly, including a hundred and forty thousand dollars burned on planning events inside a single week, and seven product owners rotated through in two and a half years. Buying the practice was never the hard part.
Get the book →A short read every Saturday morning. No spam, unsubscribe anytime.
Subscribe