Why reviews and surveys miss what's actually blocking your team

Last Friday I sat down to write the week's summary and got stuck on one line.

I run a team whose output is judgment. Not units shipped, not tickets closed. The work is deciding what to build, what to drop, who to talk to, and in what order. At the end of a week like that, the honest question isn't "what did we get done." It's "what actually blocked us."

Nobody could answer it. Including me.

That's an uncomfortable thing to admit in public, so let me be precise about what I mean. It isn't that the week went badly. It's that we finished it without a shared, specific answer to the one question that would have made the next week better.

The obvious fix, and why it kept failing#

You already know the standard responses, because you've tried them too.

Run a survey. Ask in the one-on-one. Put it in the review cycle. Set up a retro. Add a Friday form with three questions and a free-text box.

I've run versions of all of these. They come back thin. Not hostile, not dishonest, just thin. "Too many meetings." "Waiting on design." "Context switching." True sentences that don't tell you where to push.

The lazy explanation is that people aren't being candid. Sometimes that's true. But I stopped believing it was the main thing when the same thinness showed up in my own answers, on a form nobody else would ever read.

The part that reframed it for me#

Here's the sentence I keep coming back to.

A person who is blocked usually can't name the thing blocking them. If they could name it, they would already have said it.

That does more work than it looks like it does.

It means the failure isn't in the phrasing of your question. It's structural. The answer you want does not sit inside the reach of direct introspection. You're asking someone to report on a thing they cannot currently see. A better question, a safer setting, an anonymous form, none of those change the fact that the information isn't available to the person you're asking.

Which means an entire category of tooling is aimed at the wrong joint. Better survey design, smarter pulse cadence, more thoughtful one-on-one prompts. They're all refinements of "ask more skilfully." The problem is upstream of asking.

And the person doing the rating isn't a fixed point either#

If asking doesn't work, the fallback is judging. Someone senior looks at the work and forms a view.

Two findings are worth sitting with here.

The first: how people rate their own performance and how their manager rates it barely move together. Pooled across 115 studies and 37,752 people, the self-to-supervisor correlation lands at r = .22. That isn't zero, but it's nowhere near two people describing the same thing. Two honest observers, same work, largely different pictures.

The second is sharper. When researchers decomposed where the variance in a rating actually comes from, about 62% of it was idiosyncrasy of the rater and roughly 21% was the ratee's actual performance. The score tells you more about who's holding the pen than about the person being scored.

So the two available instruments both leak. Asking hits a limit of introspection. Rating mostly measures the rater.

And the people who own these systems are not under any illusion about them. Out of a panel of 135 Fortune 500 CHROs, only 2% strongly agree that their performance management system inspires employees to improve. That's a narrow question and I don't want to stretch it: it's about inspiring, not about measuring. But it tells you that the most senior people in the building, the ones who bought the system, don't believe it does the one thing it was sold to them for.

What I think the actual shape is#

If the blocker can't reliably be asked for, and can't reliably be judged from outside, then there's only one place left for it to come from.

It has to be surfaced by the person themselves, out of their own material, in their own words. Not from a form they fill in. From the traces their week already left behind.

That's a narrow claim and I want to keep it narrow. I'm not saying people are unknowable. I'm saying the accurate picture of what's blocking someone is the one they draw themselves, and the useful job is handing them the raw material and the space to draw it. See it, then clear it. The seeing is the hard part, and it's the part almost nothing on the market actually helps with.

That's the bet I'm building on, and it's worth saying which part of it exists. One person assembling their own picture, privately, out of their own week: that runs. The thing you probably want, which blockers keep recurring across your whole team, does not run yet.

When it does, it will only ever show patterns, never a view into any single person, and only once there are enough people for a pattern to mean anything. That's not a privacy gesture I'm adding for taste. It's load-bearing. The moment someone suspects their entry can be read back as a report on them, the entries go careful, and a careful entry is worth nothing to anybody. The restriction is what keeps the material real.

So I'd rather hand you the half that's true than describe the whole and hope.

One question back to you#

If your team's output is judgment, not units: when was the last time someone could actually name what was blocking them, before it turned into a resignation or a missed quarter?

I read every reply.