The skip-level is a standard practice: a manager meets the people who report to their direct reports, without those managers present. The stated purpose is to hear what's really happening without filtering.

It's a reasonable idea and it very frequently produces nothing useful. Understanding why says something about how information moves in organisations.

The incentive problem

Consider the position of the employee. They're being asked, by their manager's manager, whether anything is wrong.

If they say yes and identify a problem involving their manager, several things could follow. The senior person might act, in which case the manager will likely work out where it came from. Or they might not act, in which case the employee has taken a risk for nothing and their manager may still find out.

The expected value of candour is poor unless the employee has considerable trust in the senior person, which usually requires a relationship that a quarterly meeting doesn't build.

So the rational response is to say things are broadly fine, offer a minor and safe observation, and leave. Which is what happens in most of these meetings.

What gets said instead

There's a recognisable genre of skip-level content: problems that are real, safe to raise, and outside anyone present's control.

The office is too cold. A tool the company uses is annoying. Another department is slow to respond. Process X takes too long.

These are genuine irritations and raising them costs nothing. They also aren't the information the meeting was designed to surface, and a senior manager who leaves with a list of these and concludes things are healthy has been misled without anyone lying.

The manager's position

Worth considering the middle manager too, because they're frequently treated as the obstacle when they're in an awkward position.

Their reports are meeting their boss without them. Whatever's said, they won't hear it directly and may hear a version of it later. They have no ability to provide context on anything raised.

That's genuinely uncomfortable, and managers who become defensive about skip-levels are often responding rationally rather than hiding something. Where the practice is introduced without explanation it reads as a loyalty test.

When it does work

Skip-levels can be valuable, and the conditions are fairly specific.

Frequency and normality. If they happen regularly with everyone, they stop being an event. A meeting that occurs once when something is suspected is an investigation; one that occurs quarterly with the whole team is a routine.

A clear and limited purpose. "I want to understand what you're working on and what's getting in your way" is answerable. "Is everything alright?" invites the safe answer.

Explicit boundaries on what happens next. Stating clearly what will and won't be repeated, and then honouring it exactly. One breach ends the usefulness of the practice permanently.

Visible action on the small things. If a minor irritation raised in one meeting is fixed and the fix is attributed, the next meeting gets better information. Trust is built through demonstrated response, not through assurance.

The manager knowing in advance. Announcing skip-levels as routine, with the manager informed, removes the sense of going behind someone.

Better questions

The framing of questions matters more than anything else, and generic openings produce generic answers.

Things that tend to work: "What took longer last month than it should have?" "What do you spend time on that you think is pointless?" "If you were running this team, what's the first thing you'd change?" "What did we get wrong recently?"

These are specific, they invite concrete answers, and they don't require criticising a named person.

The last one is particularly useful because it invites criticism of the organisation rather than of an individual, which is far safer to give and frequently surfaces the same underlying issues.

The alternatives

Several mechanisms produce better information than skip-levels and are used less.

Anonymous surveys, if genuinely anonymous and if results are published including the uncomfortable parts. The publication is what makes them credible; a survey whose results are never shared teaches people not to bother.

Exit interviews, taken seriously and aggregated. People leaving have the least to lose and consistently say the most useful things, and the information is routinely filed and ignored.

Watching what people do rather than asking. Where do requests pile up? Which processes get worked around? Which meetings do people skip? Behaviour is more honest than testimony.

And simply being around. Managers who are present regularly hear things incidentally that no scheduled meeting would produce. That's harder to systematise and it's where most real information actually comes from.

The scale limit

One practical constraint that gets ignored when these programmes are rolled out. Skip-levels only work at a scale where the senior person can hold the relationships in their head.

A director meeting thirty people quarterly is conducting a hundred and twenty short conversations a year with people they do not otherwise interact with. That is not a relationship, and without a relationship the candour never arrives.

Above a certain span, the honest options are to sample rather than cover everybody, to accept that the meetings are ceremonial, or to invest in other mechanisms entirely. What does not work is scaling the format past the point where it can function and then wondering why the information stopped flowing.