Why Skip-Levels Were Overrated
Eighteen months of monthly skip-levels. Most of what they surfaced was already sitting in three Slack channels I had stopped reading.
I ran monthly skip-levels for 18 months with every IC two layers down. Call it 45 thirty-minute conversations a quarter. The practice was sold as the cure for manager filtering, the way to hear what the team thought, unfiltered. By month 18, most of what those conversations surfaced had been sitting in three Slack channels the whole time.
The engineering retro notes. The customer-feedback channel where CSMs forwarded complaints. The on-call channel where the load on senior engineers showed up as raw message volume. Each held, in writing, most of the signal the skip-levels produced out loud. I had not been reading them closely.
The skip-level had become a stand-in for the harder work of reading the signals already in front of me.
Most management rituals drift the same way. The ritual is the visible artifact, and the artifact gets mistaken for the intent. The intent is to hear what is happening. The calendar entry only signals that you care to. A founder who reads three channels for ten minutes a day learns more in a week than one who runs five skip-levels in the same week. I made the same mistake with one-on-ones, which I wrote about in how the one-on-one becomes a status meeting.
Skip-levels do earn their slot in specific cases. A new executive who has not built trust needs direct conversations to compress months of it. A team coming out of a crisis, a layoff, an outage, a senior departure, needs the leader present in an unmediated way. A specific worry about a manager needs checking from below. I unwound my own standing cadence for the reasons in the skip-level you stopped doing.
Outside those cases, the standing monthly skip-level substitutes for signal the leader could already see. It feels productive. It is also weight on a calendar, paid against the time you would have spent reading the retros. The cheapest management hygiene is to read what the team already writes down. The next cheapest is to run a skip-level only when it will surface something the channels will not.
The standing version is the default. It is also the version the team quietly stopped bringing anything real to.