When Kat Cole took over Cinnabon, one of the first things she did was go on a listening tour of franchises. She asked frontline employees a simple question: what do customers ask for that we don’t have?
A clear theme came back — smaller cinnamon buns. That became MiniBons, rolled out across the brand, and a 12% increase in revenue.
That’s the skip-level meeting at its best: a senior leader skipping the layers between them and the frontline, hearing the ground truth directly, and acting on it. It’s one of the most valuable things a leader can do.
It’s also one of the first things to break as a company grows.
What a skip-level meeting is (and why it works)
A skip-level meeting is a conversation between a leader and the people who report to their direct reports — or further down. The point isn’t status updates. It’s to hear, unfiltered, what’s actually happening: what’s working, what’s slowing people down, what customers are saying, and what the leader can’t see from where they sit.
Done well, skip-levels compress the time between someone on the frontline having a thought and leadership knowing about it. That’s a real competitive advantage.
Skip-level meeting questions that actually work
The best questions are open, specific to the business, and impossible to answer with “fine.” A few I’d start with:
- What do customers ask for that we don’t have?
- What’s the biggest thing slowing your team down right now?
- If you were in my seat, what’s one thing you’d change?
- What’s working well that we should do more of?
- What’s a risk you see that you think leadership might be missing?
- Where do you spend time that feels wasted?
- Who on your team deserves more recognition than they’re getting — and why?
- What should we stop doing?
Two habits matter as much as the questions. Follow up — the first answer is rarely the real one; ask for an example, what’s causing it, and what it’s costing. And if you’re the most senior person in the room, speak last, so you don’t over-influence everyone else.
Why skip-levels stop scaling
Here’s the math problem.
Say a CEO wants a 30-minute skip-level with every employee once a quarter. At 150 people, that’s 75 hours a quarter — almost two full weeks of the CEO’s time. At 1,000 people, it’s 500 hours. Nobody has that.
So skip-levels get rationed. A few teams, a few times a year, usually the ones that are easiest to reach. And even when they happen, four other problems creep in:
- Candor drops in the room. It’s hard to tell the CEO, face to face, that their favorite initiative isn’t working.
- Memory fades. What you heard in March is fuzzy by June, and it lives in one person’s notes.
- You can’t tell one voice from fifty. One passionate person can sound like a trend. A real trend can sound like one complaint.
- Nothing happens next. Insights from a skip-level rarely get attached to an owner or a goal, so they quietly die.
The result is the pattern I see over and over: as companies grow, skip-levels become increasingly impossible, the layers of the org chart slow down — or prevent — senior leaders from hearing the ground truth, and decisions get slower while signals get missed.
Why a pulse survey isn’t the answer either
The usual fallback is a pulse survey: five to fifteen rating questions, run monthly or quarterly. It’s a good tool for what it does well. It’s fast to answer, comparable round to round, and quick to read in a leadership meeting. If you need a reliable signal that something moved, and where, a pulse does that job.
But a score can’t tell you the three things leaders need to act:
- Why. A 3.6 on “I have what I need to do my job” doesn’t say whether the issue is tooling, staffing, a manager, or a customer that’s eating the team alive.
- The specifics. The open-text box usually gets one sentence: “Onboarding is slow.” Which customers? Which steps? Since when?
- The cost. “Onboarding is slow” and “Acme waited 41 days to go live and nearly churned” are very different inputs to a decision.
So the score kicks off a second, slower process — meetings, Slack threads, a manager asking around. Which is the skip-level problem all over again.
| Skip-level | Pulse survey | AI interview | |
|---|---|---|---|
| Reach | A few teams | Everyone | Everyone |
| Depth | Deep, if you probe | A rating, maybe a sentence | An example, a cause, & a cost |
| Follow-up questions | Yes | None | Automatic |
| Trend over time | No | Yes | Yes, with rating questions |
| Leader time | Very high | Low | Low |
What replaces them at scale
The answer isn’t to stop doing skip-levels. It’s to stop relying on them as the only channel.
What works at scale looks like an asynchronous skip-level with everyone, run continuously:
- Ask the same leadership questions to every team, on a cadence. Quarterly, monthly — whatever matches how fast your business moves.
- Let people answer the way they’d talk. Voice is several times faster than typing, and people say more when they can just talk. As one employee put it: “I like that I didn’t have to type anything out.”
- Ask thoughtful follow-up questions, like a sharp executive would. This is the part static surveys miss. In Elevea, when the AI asks follow-up questions, the average response is 80% longer — the difference between “onboarding is slow” and “Acme waited 41 days to go live and nearly churned.”
- Keep a few rating questions. They give you the comparable trend line a pulse is good at — and when a score drops, the reason is already sitting next to it, in the team’s own words.
- Protect candor. Real anonymity, enforced in the data, with small groups suppressed so nobody can be identified by subtraction.
- Count it. Group similar answers into themes so you can see whether one person or fifty raised it, and rank them by what leadership actually needs to know.
- Attach it to action. Turn findings into owned actions and goals, then ask again next round to see whether it changed. This is also the cure for survey fatigue, which is rarely about how often you ask. It’s a trust problem: if people see nothing change, they give shorter, safer answers next time — or none at all.
The bottom line
The best leaders lead first by listening. That doesn’t have to stop being possible at 500 or 5,000 people — it just has to stop depending on one leader’s calendar.
Keep the questions. Keep the curiosity. Let software handle the reach.
