Three Interventions That Change
What Gets Written Down.

I want you to picture a form.

Not a meeting, not a difficult conversation. A form. The one your organization opens at review time. There is a rating on it, and underneath the rating there is a box where you type a paragraph about a person.

That box is the subject of this episode. And if you are somewhere you can actually pull one up while you listen, do that, because everything I am going to ask you to do today happens on a form or an agenda you already have. By the end of this I want you to have decided to change one thing about one of them, before Monday.

So let me tell you why that box matters more than it looks like it does.

When you type a sentence about someone into it, something happens to that sentence that does not happen when you say the same thing out loud. Out loud, it is your read. Somebody in the room can push back on it. It can be wrong, and get corrected, and be forgotten. On the form, it stops being your read and becomes a fact about that person that other people act on. The next person who opens that file was not in the room where you formed the impression. They cannot check it. And reasonably, they treat what is written there as the record of what happened.

Then they add to it. They write their own interpretation on top of yours. And here is the part I think is genuinely under-appreciated: the second interpretation is not independent of the first. The person writing it has already read yours. So it is not a second data point. It is an echo of the first one, sitting on the page looking like corroboration.

Two cycles in, that file holds a layered account of how somebody comes across, and there is nothing underneath it to check the account against.

I want to say the next part plainly, because it is not an abstraction and I am not going to soften it. Many neurodivergent people have lived on the wrong end of exactly this. It affects whether you get hired. Whether you keep the job. Whether you progress. Whether you are treated with dignity at work. Every cycle an organization runs without changing that box puts more of those sentences into more of those files, and they turn into consequential decisions about people's lives. That is what is actually at stake in a piece of software most organizations think of as an HR formality.

Now, your organization already trains people not to do this. Almost every leader listening has sat in a feedback workshop that taught the difference between an observation and an evaluation. That training is good, and it is worth what you paid for it. I will come back at the end to where it comes from and why it stops working, because that part is worth knowing.

But here is the short version. The training asks a leader to hold that split in their own head. Under time pressure. At the end of a quarter. In a room where the other people may never have had the same training, or had it four years ago. And while the form in front of them asks only for a rating and a paragraph.

So the training is doing its work in one place and the artifact is doing its work in another, and the artifact is the one that gets saved.

Which brings me to the three changes. My Cognitive Translation Protocol sets nine structural interventions against the six failure dimensions I walked through last time, and those nine come in three lanes. Today is the first lane. Three changes, each one to an artifact you already own, and any one of them fits inside a single review cycle.

The first one is interpretation guardrails.

Here is the change. You add a field, and that field sits before the assessment field, and it asks what was said and done.

That is the whole intervention.

Let me show you why it works, because it sounds almost too small to matter.

Go and read the last set of reviews you signed off on, and read the comments rather than the ratings. You will find sentences that look like descriptions and are really conclusions. Struggles to influence senior stakeholders. Comes across as defensive when challenged. Needs more executive presence. Not one of those tells you what the person said, or what they were asked, or what happened next. And every one of them gets read by the next person as though it were a record of an event.

The form invited those sentences, because it asked for an assessment and gave nowhere to put the thing the assessment was based on.

Now here is what that same feedback looks like when the observation field exists.

The observation field says: in the March steering review she was asked whether the timeline held, and she said the build was on track, integration testing was the open risk, and gave a date to watch. The assessment field says: I came away thinking she would not commit.

Read those two together. That is a completely different document, because the second sentence is now visibly a conclusion sitting next to the thing it was drawn from. Anyone opening that file can see the gap between what she said and what the room heard, and can decide for themselves whether the second follows from the first. Six months later somebody can look at that and think, hold on, she did answer the question.

In a talent or calibration meeting, the same intervention is a rule instead of a field. A claim about a person does not go into the record until the person making the claim has said what they observed, and the note taker writes down both.

What does it cost you? Almost nothing to build, and running it is close to the time you already spend. The box usually exists on your review platform already, as an open comment field nobody ever defined a job for. The work is in what you require of that field, and what you refuse to accept while it is empty.

The real expense is the first two cycles. People will write "seemed disengaged" in the observation field, because that is what twenty years of these forms trained them to write, and they have to be sent back to write what they actually saw and heard. That is uncomfortable the first few times. It gets easier faster than you would expect, and the reason it gets easier is that writing down what you actually saw turns out to be less work than defending an impression.

And what stops. A direct answer stops being recorded as reluctance, because the answer now sits on the page next to the conclusion somebody drew from it. The state of mind comes out of the record too, because "came across as frustrated" does not survive a line that asks what was said. And when an instruction has to go into the observation field exactly as it was given, that instruction stops arriving in the file as the other person's failure to follow it.

Second change. Context equalization.

Think about the last decision meeting where the discussion got worse the longer it went on. A proposal goes up. Around that table, three people have read the analysis from six weeks ago and four people have not, and nobody says which group they are in.

The three evaluate the proposal against the analysis. The four evaluate it against the last plan they saw. And every objection each group raises is correct from where that group is sitting.

What the room experiences is disagreement about the proposal. What is actually happening is that half the exchange is people reasoning from different starting facts. And nobody names it, because nobody in the room can see it. From the inside it just feels like a hard meeting with a difficult group.

The agenda created that. The agenda went straight to the proposal without establishing what the room already knew.

So the change is a written baseline that exists before the evaluation starts, and that everybody has actually read. Amazon's six-page memo and the silent reading period at the top of the meeting is the version of this most leaders have heard of, and what makes it work is the shared baseline, not the prose.

You do not need six pages. Let me tell you what you actually need, because this is one you could write tomorrow.

At the top of the material, three short blocks. What is known: the facts this decision rests on, with dates on them. What is assumed: the things you are treating as true and have not verified, said out loud as assumptions. And what has changed since this group last looked at the question, which is usually the one that surprises the room.

Then the first item on the agenda is not the proposal. It is confirming that block. Does anyone have a different number. Does anyone think one of those assumptions is wrong. Two or three minutes, and you have either confirmed everybody is starting from the same place, or surfaced the disagreement that was going to take the next forty minutes anyway.

This one takes real effort, and that is exactly why most rooms skip it. Somebody has to own writing the block, and it is real work every single time. The first few meetings run long, because the gaps come out into the open instead of staying in people's heads.

Although I would push back on that a little. Those meetings were already running long. The time was going into an argument with a fact gap underneath it, and nobody got to see the fact gap, so nobody could close it, and it came back the following month as a different argument about the same numbers.

What stops: a challenge that comes out of a fact gap stops being recorded as resistance, because the baseline is on the page and anyone in the room can see which fact the challenge is actually about. The person is no longer difficult. They are working from row three of the context block, and now everyone can see it.

Third change, and this is the one that reaches furthest. Outcome-based evaluation.

Go back to that same review file and look for the field that records what the person was right about.

In most forms, there is not one.

Your organization can tell you that a director did not carry the room in March. It cannot tell you that she was the only person in that room who said the migration date would not hold, and that it did not hold.

The form has a place for the delivery and no place for the result. So the delivery gets scored, and the result never enters the record at all.

The change is a field that asks what this person's calls turned out to be worth. You fill it in with the calls, the dates, and what happened after them. The reads that were right and were acted on. The reads that were right and were not acted on, which is the most expensive category in your organization and the one nobody counts. And the reads that were wrong, because the point is an accurate record of judgment rather than a list of wins.

And it sits alongside your competency ratings rather than replacing them. How someone works with other people is real information and you are not giving that up. You are adding the other half of a picture most organizations have simply never held.

This is the most expensive of the three, and the expense is keeping a record across the year rather than money. Somebody has to write the call down when it gets made, and that is the whole discipline, because in December nobody remembers March.

Most organizations already have somewhere it could live. A decision log, if you keep one. The meeting minutes, if they are any good. The work is deciding that a named person's read goes in there with their name attached, at the moment they say it. And one line is enough. March eleventh, platform review, she flagged the migration date as the risk. That is the entire entry. In December that line is worth more to you than the whole paragraph somebody wrote about her presence.

And what stops. A read that turned out to be right stops being invisible, because something in your organization is holding it. A person's worth stops getting estimated from how they present, because there is now something else on the page to weigh it against. And the reads you were getting for free, from people your reviews were quietly marking down, stop walking out of the building with them.

That last one is the business case, honestly. Those people leave. They usually leave for somewhere that noticed. And the thing they were carrying leaves with them, and because you never wrote it down anywhere, you do not even find out what you lost.

Alright. Operating Brief. Four parts, same as always.

The moment is this. A sentence about a named person is about to get typed into something your organization keeps. A review comment, a calibration note, a succession slate, a handover document. That is the moment, and it is happening somewhere in your organization this week.

The vibe-led default is that you change nothing on the artifact, because the artifact is not yours and the cycle is already open and you have a job to do. So the interpretation goes in with nothing underneath it, and the next person inherits it as established and writes on top of it, and after two cycles the file holds an account of how somebody comes across, with no record of what was said and no record of what they turned out to be right about. By then there is nothing left in the file to check it against, and the account has become the person as far as your organization is concerned.

The redesign is to pick one of the three and change one artifact this cycle. Add the observation field ahead of the assessment field. Put a written context block at the top of one recurring meeting's material. Add one line to the review that asks what this person's calls turned out to be worth. One of them. Not a program, not a rollout, not a working group. One field, on one form.

And I should be straight with you about the boundary, because I do not want to oversell this. These three decide where an interpretation is allowed to enter your record. They do nothing about timing, and they do nothing about sensory load. The person who needs until tomorrow morning to give you the best answer in the room still loses the question to a fast meeting. The all-hands still costs what it costs. Those are the next two lanes, and the next two episodes.

The Monday test is the smallest one I have given you yet. Open the artifact you picked, make the change on it, and then say out loud in your next leadership meeting which field changed and what it now requires before anything goes into it. Saying it out loud is not decoration. It is what turns a change you made into a rule the room knows about.

Before I let you go, the part I promised you at the top.

That workshop training goes back to Marshall Rosenberg. He was a clinical psychologist, and he founded the Center for Nonviolent Communication in 1984. His model runs in four steps, and the very first one is separating what you observed from any evaluation you made of it. The Center for Creative Leadership teaches leaders the same separation as situation, behavior, impact.

The distinction is correct, and it is worth every penny you spent on it. An observation can be checked. An evaluation cannot be checked without context, so without the split made explicit, an evaluation is an unsupported argument about who somebody is.

But Rosenberg built a personal discipline, and he built it for a context where the individual was the right unit. Two people, in a room, working on how they speak to each other. A personal discipline holds for the leaders who are good at it, on the days they have the attention for it, in the rooms where the other people were trained too. Which makes the benefit to your organization real, uneven, and almost impossible to trace.

If you have been following the work, you know I take this back to Edgar Schein. Culture is measured by the alignment between artifacts, espoused values and tacit assumptions. You already train the observation and evaluation split at the level of espoused values. The tacit assumption underneath never got touched. And the form is where that assumption lives, because the form is set up to rate an individual and attribute an evaluation to them, and it has been set up that way for so long that nobody experiences it as a choice anymore.

That is the artifact barrier. And an intervention is what you do about it.

So if you take one thing from today, take this. You do not have to become better at feedback than you already are. You have to change what the artifact asks for, and what it asks for first.

One change. One artifact. This cycle.

I will see you next time, in the process lane.