Root Cause Drill

Trace recurring problems to their true root cause.

Root Cause Drill

What this fixes

A problem keeps coming back. Each time, you find a plausible cause, patch it, and move on - and a few weeks later, it returns, in a slightly different form.

That’s the trap. The first cause you land on is usually the comfortable one. It explains the symptom, patching it feels productive, so the drilling stops. But a cause you can patch isn’t always the one driving the problem. The first cause is often where curiosity stops, not where the problem starts.

This tool makes you keep drilling - and, crucially, keep proving. You trace the problem down one evidenced step at a time, and you’re not allowed to move to the next cause until the current one is backed by evidence. The chain ends at a root you can actually remove.

How this works

Imagine this: your team’s releases keep slipping. The obvious cause is “we underestimate the work”. You could stop there and add buffer time - and slip again next release.

You build a chain, one link at a time. Start with the symptom, then lay the first cause beneath it. Before you can add the next cause down, you have to show evidence that the current one is real - a log, a number, an example, not a guess. Then ask what caused that proven cause, add the next link, and prove it. Repeat, going down, until removing the bottom cause would stop the recurrence.

The evidence is the whole point. The moment you can’t back a cause, the chain stops - and that’s usually where the room realises it’s been guessing, not drilling. Unlike Five Whys, every answer must be backed by evidence before you move on.

You’ve reached the root when removing it would stop the recurrence - not just reduce this instance. Then you read the finished chain back from the root up to the symptom: the proof that this cause really does drive the problem.

Team: one person acts as the facilitator - they run the activity using the facilitation guide while everyone else takes part.

Time required

Solo ✓   Team ✓   Solo: about 20–30 minutes. Team: 40–60 minutes.

The time required is indicative and will depend on how many layers sit beneath the symptom.

The facilitator should read the short facilitation guide first - about 6–8 minutes. It is a downloadable PDF on this page.

When to use

Bring one real problem that has come back after being “fixed” at least once - a live recurrence, not a hypothetical.

Use it whenever:

  • the same problem keeps returning after you thought you’d solved it
  • you’re about to apply the obvious fix for the second or third time
  • a “solution” only ever seems to hold for a few weeks
  • different people blame different things and none of it is evidenced

What you get

  • A named root cause - the one that, if removed, stops the recurrence
  • The evidenced chain that reached it, from symptom to root
  • One thing to change, and a way to check whether it worked

You’ve done it well when: every link in your chain has evidence behind it, and the root is something you could actually remove - not another symptom to patch.

What you’ll notice

  • How early you’d normally have stopped
  • How many of the “causes” were really just guesses
  • That the real root is usually two or three layers below the obvious one

Why this works

Recurring problems survive because people stop at the first plausible cause. It feels productive to patch it, so the investigation ends too early - even though the information to go deeper is usually right there. Requiring evidence at every link keeps the team drilling until the real cause is found, instead of the first comfortable explanation.

What you need

One recurring problem. A specific problem with a real symptom you can point to. Choose one that has already been returned after being fixed at least once. Bring a live one, not a hypothetical.

The capture sheet (supplied). The linked cause boxes where the chain is built: symptom at the top, then a cause and its evidence at each step down, ending in the named root. An unfinished chain is meant to look unfinished. Leave space to add more links if the chain grows.

Evidence to hand. Logs, numbers, timelines, screenshots, examples - whatever lets you back each cause. Part of the activity is noticing where you don’t have it.

Roles - Team: a facilitator (runs the flow, holds the evidence gate, writes the chain) and participants (propose causes, bring evidence, decide the root). Solo: you run it yourself from the capture sheet.

Steps

  1. Set up (2 min). Team: appoint a facilitator first - they run the activity from the facilitation guide. Solo: just begin. Write the recurring symptom in the top box.
  2. Name the symptom precisely (3–5 min). Describe the symptom before anyone suggests causes. Write what actually keeps happening, in observable terms - what you see, how often, since when. Not the cause yet; the recurrence itself.
  3. Lay the first cause (3 min). Write the first cause you’d normally fix directly beneath the symptom. This is the one you’d usually stop at.
  4. Prove it (3–5 min). Write the evidence for that cause - a log, a number, an example. No evidence? You cannot go further until you have it. Go and get it, or mark the link unproven and stop there, honestly.
  5. Drill again (repeat). Ask what causes the proven cause, write it as the next link down, and prove that one too. Repeat - cause, evidence, cause, evidence - going down, never sideways. Follow one chain only; don’t branch into separate possibilities.
  6. Name the root (3 min). Stop when removing the bottom cause would end the recurrence, not just reduce it. Circle it as the root.
  7. Read the chain back (2 min). Read the chain from the root up to the symptom, out loud. If every link still makes sense when read from root to symptom, you’ve found the root. Write the one change you’ll make and how you’ll check it worked.

Output / Takeaways

You leave with:

  • The evidence chain from symptom to root
  • The named root cause - the one worth removing
  • One change to make, and the check that tells you it worked

Important rules

  • No evidence, no next link - you cannot drill past a cause you can’t prove
  • Drill down, not sideways - one chain, not a list of separate causes
  • A cause you can only patch isn’t the root - keep going
  • Test each link, don’t just ask “why” - this isn’t five questions in a row

Planning to facilitate this activity? Access the resources below to get started.

MIndn
Execution must be smooth. Not chaotic or stressful.
Built by Witmer.