She interviews for the crisis better than anyone in the building. In the talent review she is the name three people say without being prompted. Composed. Structured. Asks the second question before the first one is finished. The calibration panel does not debate her rating. They confirm it and move on, because everyone in that room has watched her handle a hard conversation, and watched her handle it well.
Then a supplier fails at 6:40 on a Friday, the kind of failure that becomes a headline by Monday if it is not contained tonight, and the composed operator the panel described is not the person who picks up the phone. The person who picks up the phone protects herself first. Before the customer. Before the team. Before the strategy she articulated so cleanly in the review. Nobody trained her to do that. Nobody has ever seen her do it, because until 6:40 on that Friday there had never been a moment fast enough and real enough to get underneath the version of her that everyone had spent two years assessing.
The Self Your Instruments Can See.
Consider what your organisation actually knows about its leaders. It knows how they present in a 360, an instrument they are fully aware is running. It knows how they answer in an interview, a performance conversation, a talent panel, a leadership programme with a facilitator taking notes. Every one of these is a mirror the leader can see. And a leader who can see the mirror does the most natural thing in the world.
They perform. Not dishonestly. They perform an accurate, well-lit version of their best self, because that is what the situation rewards, and they did not reach a senior seat by misreading situations. This is the part most assessment gets wrong about its own limits. The problem was never that leaders lie to the instrument. The best of them tell it something true. The problem is that an observed instrument can only ever collect the truth about the observed self, and the observed self is not the one who shows up when the pressure is real and the room is empty of anyone worth performing for.
Nobody watching does not require literal solitude. The crisis at 6:40 was full of people. What it lacked was the slack a leader needs to build and hold a performance. Under enough load, arriving fast enough, the performing self simply cannot keep up, and a more primitive operator takes the controls. The one who decides in the first ninety seconds. The one who chooses, without deliberation, what to protect and what to let burn. That operator has been running quietly the whole time. It has never been measured, because measurement always arrives with a mirror attached.
Why More Instruments Do Not Help.
The instinct, once an organisation senses this gap, is to add instruments. More raters on the 360. A psychometric. A structured behavioural interview with a crisis vignette bolted onto the end. It is worth understanding precisely why this does not close the gap, because a great deal of budget dies here.
Every new instrument inherits the same property as the old ones. The crisis vignette is still a question the leader can see coming. And a question you can see coming is a question you can answer as your best self, calmly, with the part of your mind that is never available during an actual emergency. You can install ten mirrors in a room. You still cannot use any of them to see the person who only appears once the mirrors are gone. The observer effect is not a flaw in a particular tool. It is a property of being observed, and you cannot out-observe it by observing harder.
You have spent years refining your view of the one self a crisis will never call on. You have never once seen the other one.
What the Gap Costs at Scale.
The cost is not abstract, and it is not evenly distributed. Organisations promote the observed self and are then governed, in the exact moments that matter most, by the unobserved one. The leader who reviews as collaborative centralises everything the instant it turns dangerous. The one who reviews as decisive freezes when the decision is genuinely irreversible. The one everyone calls calm transmits panic downward through a hundred small signals nobody can name afterward, only feel.
None of this appears on the succession slate. All of it appears in the outage, the recall, the quarter that quietly goes sideways, the negotiation that collapses in the last hour. The organisation discovers who its leaders actually are at precisely the moment it can least afford the education, and it pays the tuition in customers, in trust, in the eighteen months it takes to recover from a decision that was made in ninety seconds by a person the review process had never met.
Multiply that across a leadership population and you are running a company on a map of your people that is accurate for the calm and blank for the storm. The storm is not the exception. It is the part of the job you are actually paying senior leaders to hold.
The Turn.
Here is the uncomfortable part, and it is not the one most executives expect. Your instruments are not broken. Your 360 is well designed. Your calibration is rigorous. Your interviewers are trained and your talent reviews are honest. It does not matter, because every one of them shares a single structural property that no amount of rigor can remove. They are watched. You have not been measuring your leaders badly. You have been measuring, with great care, the one version of them the next real crisis is guaranteed not to summon. That is not carelessness. It is the natural result of never having built a condition that could not be performed to, which turns out to be a very specific and deliberate thing to build.
What Rebuilding Looks Like.
Surfacing the unobserved self is not a matter of watching harder. It is a matter of engineering a situation the performing self cannot survive. Real enough that the body treats it as real. Fast enough that there is no room to compose the narrative. Consequential enough that the choice of what to protect actually costs something. Under those conditions the mirror goes dark on its own, and the operator who runs your real emergencies steps forward while, for the first time, someone is deliberately watching.
This is the entire reason the SSUNDAR Organizational Crisis Simulation exists. It does not ask a leader how they would handle pressure, because that question interviews the observed self and receives its polished, sincere, useless answer. It puts them inside five cascading crises with the clock running and every decision compounding into the next, and it reads what they actually do when the situation moves faster than any performance can. What surfaces is frequently not what the 360 predicted. It is, with unnerving reliability, exactly what the next real crisis was always going to get.
The leaders who see the gap between their two selves are not diminished by it. They are relieved by it. For most of them it is the first time the operator who runs their hardest moments has been made visible, including to themselves, and you cannot develop what you have never been able to observe. The work is not to shame the unobserved self into hiding. It is to bring it into the light long enough to build it, so that the person your organisation promotes and the person who runs your worst day slowly become the same person.
Your leaders have a self you promote and a self that runs your worst day. So far, you have only met one of them.