The 15-Minute Fact-Finding Conversation Every Founder Should Run Monthly

The single biggest mistake I see SMB owners make in skip-levels is treating them like a status check. By the time information has been filtered through three managers, what reaches you is a sanitized PowerPoint of reality. The fact-finding conversation exists to bypass that filter — without breaking trust with the managers who built it.

What is a fact-finding conversation?

It is a one-on-one between the owner of the company and an individual contributor two or more layers below. It is not a performance review, not a coaching conversation, and not a complaint channel. It is calibration.

Definition: Fact-finding conversation — a structured 15-minute monthly conversation between the owner and a non-managerial IC, designed to surface operational reality that hierarchy filters or distorts.

The owner is not trying to catch managers out. The owner is trying to triangulate. If every dashboard says "execution is healthy" and every fact-finding conversation says "we've been blocked on the same vendor for six weeks," one of those signals is wrong.

Why does ground truth not reach the owner via the hierarchy?

Three structural reasons, none of which involve bad managers.

First, every layer compresses. A team lead writes three bullets to a Head of, who writes one bullet to the founder. By the third compression, the texture is gone.

Second, every layer optimizes. Managers are scored on outcomes, so the upward narrative is shaped to make outcomes look defensible. This is not lying — it is rational behavior under incentives.

Third, the founder's questions condition the answers. When the founder asks "are we on track?", the answer is almost always "yes, with one minor concern." The conversation hasn't earned candor.

Definition: Hierarchy filter — the compounding effect of compression, optimization, and conditioning that reshapes ground-floor reality into the version that reaches the owner via management layers.

The fact-finding conversation neutralizes all three by going around the hierarchy, asking unconditioned questions, and removing the optimization incentive (no manager will ever read the answers).

How do you set it up without breaking trust with managers?

You explain it once, publicly, before you start.

The script: "I'm going to run a 15-minute fact-finding conversation with one IC per month, chosen at random. The point isn't to grade managers — it's to calibrate what I think is happening against what's actually happening on the floor. Managers will not see the notes. ICs will be told the same thing. If anything comes up that requires action, I will surface the issue to leadership without naming the source."

You do this in a leadership meeting. Once. Then you stick to it.

The mistake is doing fact-finding conversations covertly. Managers find out within two months, and the trust cost is far higher than any signal you got.

What are the six questions?

The questions are fixed. The order is fixed. You do not improvise.

Fact-finding conversation — 6 questions (15 minutes total):

1. WALK ME THROUGH YESTERDAY
   "Tell me what you actually worked on yesterday, in order."
   (Listen for: friction, waiting, rework, surprise priorities)

2. WHAT'S THE MOST FREQUENT BLOCKER YOU HIT IN A TYPICAL WEEK?
   (Listen for: tools, decisions waiting on others, unclear ownership)

3. WHAT'S A DECISION THAT GOT MADE RECENTLY THAT YOU'D HAVE MADE DIFFERENTLY?
   "You don't have to say why — just what."
   (Listen for: signals of misalignment that don't surface upward)

4. WHAT'S SOMETHING THE COMPANY THINKS IS WORKING THAT ISN'T?
   "Even something small."
   (Listen for: gaps between official narrative and shop floor)

5. WHEN'S THE LAST TIME YOU DID REWORK BECAUSE OF SOMEONE ELSE?
   "What was the trigger?"
   (Listen for: dependency-failure patterns hierarchy hides)

6. IF YOU HAD 30 FREE MINUTES OF MY ATTENTION ON ANYTHING, WHAT WOULD IT BE?
   (Open the room; respect what they say)

After the conversation: write 6 bullets in your private notes file.
Never share notes with the IC's manager.

Fifteen minutes. Six questions. Two pages of notes maximum. Done.

Tool tip (AIAdvisoryBoard.me): A fact-finding conversation is calibration — it tells you what's missing from the official Plan → Fact → Gap signal. After three or four of these in a row, founders typically discover the same two or three patterns repeating: a quiet rework loop, a decision queue stuck on one person, a vendor relationship that's been broken for weeks. That is exactly the kind of pattern a 7-day diagnostic catches at scale — automatically — so you can use fact-finding to validate the diagnostic rather than to discover problems one at a time. See how it works at https://aiadvisoryboard.me/?lang=en.

What's the rule about acting on what you hear?

You never act on a single conversation.

This is the discipline. One IC saying "the manager makes bad decisions" is data, not signal. Three ICs in three months independently saying the same thing about the same area is signal.

Definition: Triangulation rule — the discipline of treating any single fact-finding conversation as one data point, requiring at least two independent corroborations before acting.

If you act on a single conversation, three things happen. The IC gets identified as the source. The manager loses trust in you. And you make a bad call based on one perspective. The cost of restraint is one month of delay. The cost of impatience is your entire fact-finding program.

How do you choose who to talk to?

Three rules.

1. Random within constraint — pick from a different department each month. Rotate, don't cluster.
2. Not the same people — track who you've spoken with; don't repeat within a year unless something specific brought you back.
3. Skip new hires under 90 days — they don't yet have the calibration to give you signal.

Definition: Selection bias trap — the failure mode where the owner unconsciously picks ICs they already know or like, producing fact-finding conversations that confirm rather than calibrate.

The minimum: one per month. The maximum that stays useful: one per fortnight. More than that, you stop processing what you heard.

Manager scan (2-minute digest example)

A Plan → Fact → Gap digest where fact-finding conversations have surfaced patterns over three months at a 130-person services company:

• Customer Success — Plan: 6 renewals/week. Fact: 4. Gap: EXECUTION_DRIFT. Fact-finding: 2 of 3 CS reps mentioned waiting on engineering for "the same export bug" — never surfaced upward as a renewal blocker.
• Engineering — Plan: ship migration. Fact: shipped. Gap: none. Fact-finding: one IC mentioned a quiet 6-hour-per-week meeting that nobody on the leadership team knew existed.
• Sales — Plan: 30 demos. Fact: 28. Gap: minor. Fact-finding: two reps independently flagged that the CRM data quality has been broken for months — sales managers are aware, founders weren't.
• Marketing — Plan: campaign launch. Fact: launched. Gap: none. Fact-finding: no signal.
• Ops — Plan: vendor consolidation. Fact: behind. Gap: SCOPE_CHANGE. Fact-finding: IC flagged that vendor consolidation lost priority quietly after Q2.
• Finance — Plan: AR follow-up. Fact: done. Gap: none. Fact-finding: no signal.

Three signals worth acting on, surfaced over a quarter, without breaking a single manager's trust. The founder now has a conversation to have with the Heads-of about the CRM data and the export bug — without naming sources.

Micro-case (what changes after 7-14 days)

A 90-person SaaS founder ran fact-finding conversations for three months — one IC per month, six fixed questions, no exceptions. The first two surfaced nothing the founder didn't already know. The third surfaced a quiet rework loop in customer success: every week, three CS reps spent four hours each chasing the same data quality issue, but it had never been escalated because each rep assumed the others were handling it. Total recovered time once the issue was named and owned: roughly 36 hours per month, about a third of one CS headcount. The fix cost nothing — it was a process clarification, not a tool change. The founder kept the conversation cadence going.

Note on this case: This example is illustrative — based on typical patterns we observe with companies of 30-500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees.

Tool tip (AIAdvisoryBoard.me): Fact-finding conversations work because they catch what dashboards miss — but they don't scale. One per month at 90 people is calibration; at 300 people, you would need to triangulate signals automatically. That is the slot for a Plan → Fact → Gap diagnostic running across all teams: catch the patterns ICs mention informally, surface them as repeating Gap classifications, and feed the fact-finding sessions with targeted hypotheses to test. The diagnostic doesn't replace the human conversation — it makes the conversation higher-signal. See it at https://aiadvisoryboard.me/?lang=en.

FAQ

Won't my managers feel undermined? Only if you do it covertly. Announce it once publicly, stick to the rules (no notes shared, no acting on single conversations, no naming sources), and the managers will tolerate or welcome it. The ones who can't are the ones you most need to triangulate around.

What if an IC complains about their manager directly? Listen. Don't promise anything. Tell them you'll consider it as one data point among many. If two more ICs flag the same pattern within a quarter, then act — through your normal management process, not via the fact-finding channel.

Should I share the questions in advance? No. The point is unconditioned answers. Sharing the questions creates rehearsed responses. You can share the format ("six questions, fifteen minutes, calibration not review") without sharing the questions.

What if nothing useful comes up? That's still useful. After three or four sessions with no signal, you have a real-time read that the official narrative matches the floor. That's a healthy company. Most are not.

How is this different from an engagement survey? Surveys are aggregated and anonymous. Fact-finding is direct, unaggregated, and specific. They serve different purposes. A survey tells you whether morale is healthy; a fact-finding conversation tells you what's actually blocking work this week.

Conclusion

The fact-finding conversation is the cheapest calibration tool a founder has. Fifteen minutes a month. Six questions. No improvisation. No notes shared. No acting on single sessions. The discipline is what makes it work — break any of those rules and you turn a calibration tool into a political weapon.

Run it monthly for a year and you'll have surfaced more operational ground truth than any dashboard would have shown you.

If you want a system that surfaces the Plan → Fact → Gap automatically — every day, across the company — see how the 7-day diagnostic works at https://aiadvisoryboard.me/?lang=en.