What NOT to Document

Most organizations document defensively and starve the documentation that actually protects people — the performance conversation, the commitment, the disagreement. The paper that creates real risk isn't the paper you're missing. It's the paper you didn't need.

You're writing a memo. Nobody asked for it. Nobody will read it unless something goes wrong, and if something goes wrong, that's exactly the point: you're building the record now so it exists later. You file the acknowledgment form the employee signed but never read. You send the email that summarizes a conversation in language more careful than the conversation itself, because the email is the version that counts.

This is going to read, for a few paragraphs, like an argument for doing less of that. It isn't. The problem was never documentation. The problem is what gets documented and what doesn't, and most organizations have those two things backward.

What gets over-documented

The retroactive file is the clearest case. A decision gets made about someone's employment, and only after it's made does anyone start writing down the performance concerns that are supposed to justify it. The dates on the documents say one thing. The actual sequence of events says another. Anyone who reads the file closely enough can tell the difference, which means the file exists to be read closely by exactly the wrong person at exactly the wrong time.

Then there's the paper that exists to be filed, not read. The training acknowledgment nobody absorbed. The policy sign-off collected in the first week and never referenced again. The compliance form that proves someone clicked a button, not that anyone changed what they do. This kind of documentation isn't dishonest. It's just aimed at the wrong target. It protects the organization's ability to say a process existed. It does nothing to confirm the process worked.

The CYA email belongs in the same category, even though it feels different in the moment. It's written to create a record, not to communicate. The tone is careful in a way normal correspondence isn't. The audience isn't really the recipient. It's whoever reads this later, if it comes to that. Everyone involved can usually tell the difference, which means the email accomplishes exactly what it was written for and almost nothing else.

What all three have in common is that they're written for the moment after something has already gone wrong, not the moment when it could still have been prevented. That's not a moral failing. It's a reasonable response to real risk. But it means the organization's documentation muscle gets built almost entirely around defense, and a muscle that's only ever exercised one way atrophies everywhere else.

Where documentation becomes the exposure

Some documentation doesn't just fail to help. It creates the very risk it was meant to guard against. Three areas where this shows up most often, each for a different reason.

ADA. Writing down a suspicion of disability before the employee has disclosed anything turns an informal impression into a record. "Seems like something might be going on with him" in a performance file, written before any conversation, is now evidence that the organization regarded the person as disabled, whether or not that was ever discussed openly. The same problem shows up when performance concerns land in someone's file the same week as an accommodation request. Nobody has to prove intent when the timeline does that work by itself. And medical detail belongs in the separate confidential file the ADA requires, not folded into a general personnel record where it's visible to people who have no reason to see it. The exposure isn't always the decision itself. It's what happens once that detail is sitting somewhere the wrong person can find it.

Medical and personal situations. The reason someone is out, whether it's a diagnosis, a family emergency, a divorce, tends to travel further than it needs to: into a shared calendar entry, a Slack message meant to be helpful, a manager's casual note. None of that is malicious. It's usually someone trying to give context or cover for a colleague. But personal circumstances don't need to be visible to people who don't need to know them, and once they're written down somewhere, they don't stay contained. The same instinct shows up in performance reviews, where a manager adds sympathetic context, "understandably distracted given everything going on," meaning well and creating a permanent record that ties someone's health or personal life to a formal evaluation of their work.

Whistleblowing. A complaint gets made, and the identity of the person who made it circulates more widely than the investigation actually requires, sometimes just because more people were on the email than needed to be. Performance concerns about that person start appearing in writing for the first time shortly after the complaint, with no earlier written history to show the concerns predate it. Internal debate about "what to do" gets conducted over email, in detail, with the person's name attached, when the substance could have been worked through without naming them at all. None of this requires bad faith. It requires ordinary carelessness about where information goes once it's written down.

The pattern across all three is the same. The risk isn't the underlying fact. It's the artifact. A suspicion, a diagnosis, an identity: none of these are dangerous until they're written somewhere they didn't need to be, at which point they become discoverable, permanent, and impossible to un-write.

What gets under-documented

Now the other side, which is where most organizations are actually thin.

The performance conversation happens. It's real, specific, sometimes difficult, and then it evaporates. Nobody writes down what was said. Six months later, one person remembers a clear warning and the other remembers a passing comment, and there's no way to settle which version is accurate because neither was ever recorded.

A commitment gets made out loud in a meeting. A budget, a timeline, a promise about headcount. Everyone in the room nods. Nobody sends the follow-up that confirms what was agreed, so six weeks later the commitment either quietly dissolves or turns into a dispute about what was actually said.

A disagreement gets raised and then smoothed over in the moment, treated as resolved because the conversation ended calmly, when nothing was actually resolved. It just stopped being discussed. There's no record that the disagreement happened, which means there's no record that it was never actually settled.

These are the moments that most need a written trace, and they're the ones organizations are least likely to produce one for.

Why the gap exists

The asymmetry isn't an accident and it isn't laziness. Defensive documentation is easy to produce because it doesn't require anything real to have happened first. You can write a training acknowledgment without training anyone. You can send a careful email without having a hard conversation. The paper can exist independent of the practice it claims to represent.

Documenting a real performance conversation requires having had it. Documenting a commitment requires standing behind it in writing, where it can be pointed to later. Documenting a disagreement requires admitting, on the record, that something wasn't resolved. Each of these carries exposure the defensive version doesn't: the exposure of being wrong, of being held to something, of having a disagreement that didn't get fixed as fast as it looked like it did.

If you've felt reluctant to write one of these down, that reluctance makes sense. Putting a hard conversation into writing can feel more confrontational than the conversation itself did. Naming a disagreement as unresolved can feel like giving up on a good-faith effort to move past it. Committing a promise to paper before you're fully sure it will hold can feel premature. None of that is a character flaw. It's a normal response to putting something real on the record.

But documenting is an action, and feeling uncertain is not one. You can feel unsure how a conversation landed and still write down what was said. You can feel uneasy naming a disagreement and still write that it wasn't resolved. The feeling doesn't have to resolve before the action does, and it usually won't. The leaders who are best at this aren't the ones who feel more certain. They're the ones who've learned to write the sentence anyway, the same way they'd send the defensive email without waiting to feel entirely justified in it.

This isn't optional practice for people who happen to be comfortable with conflict. Having the hard conversation is part of what leading people actually is, not an unfortunate duty attached to it. The discomfort doesn't excuse the responsibility. It's built into it.

So most organizations default to the documentation that costs less and only pays off if something goes wrong later, and they underinvest in the documentation that would have made it less likely to go wrong in the first place. The gap isn't a mistake anyone made on purpose. It's the easier choice, repeated often enough to become the pattern.

What would help

1. Write down the performance conversation the same day, in plain language, and share it with the person it's about. Stick to what was said and observed, not speculation about why. Not a formal write-up. A short, honest account both people can point back to.
2. Document disagreements as disagreements. If something wasn't resolved, the record should say that, not describe a resolution that never happened.
3. Document commitments in a form the person who made them can see. A follow-up email after the meeting, confirming what was agreed, addressed to the person who agreed to it.
4. Keep ADA, medical, and whistleblowing material inside the narrow channel it belongs in. Not wider, not more visible, not folded into records that don't need it.

None of this is complicated. It's also not what most organizations default to, because the easier version of documentation and the useful version of documentation usually aren't the same document.

The point

An organization that documents well isn't the one with the thickest file. It's the one whose file matches what actually happened, in both directions: the conversations that occurred are on record, and the ones that didn't need to be written down never were. Most organizations have that backward. They can produce a file for anything that might end up in a dispute and almost nothing for the conversation that actually mattered. The fix isn't writing less. It's writing down the right things, before they're needed rather than after, and being willing to put your name next to them even when you're not entirely sure how they'll read later.