Attrition Diagnostics: Finding the Real Reasons People Leave

"Our attrition is 24%." As a sentence, this contains almost no usable information. Is that regretted or total? Concentrated or spread? Early-tenure or veteran? Among your best people or your weakest? The headline number is a symptom report. Diagnosis starts when you cut it open.

Most retention efforts fail because they treat attrition as one problem. It is usually three or four different problems wearing one number.

First, fix your definitions

• Regretted vs non-regretted. Losing a bottom performer you were managing out is not the same problem as losing a star. Mixing them poisons every insight downstream.
• Voluntary vs involuntary. Terminations and redundancies belong in a different analysis entirely.
• Annualised consistently. Decide your formula and keep it stable, or trends become fiction.

Then segment ruthlessly

The patterns that matter live in the cuts:

• By tenure. A spike in 0–12 month exits points at hiring or onboarding. Exits clustered at the 3–4 year mark point at stalled growth paths.
• By manager. Attrition is rarely uniform. A handful of managers usually account for a disproportionate share of regretted exits. This is the single most actionable cut — and the most politically avoided.
• By performance tier. If your high performers leave at a higher rate than average, you have an emergency regardless of the overall number. If your low performers never leave, you have a different one.
• By function, location, and compensation position. People paid well below band leave for money; the cut tells you where benchmarking is overdue.

Triangulate the "why"

Numbers tell you where; they rarely tell you why. Combine:

• Coded exit interview data — consistent categories, aggregated quarterly (see our note on exit interviews as strategy).
• Engagement survey verbatims from the affected segments, often written 6–12 months before the exits they predicted.
• Stay interviews with the people most similar to those who left — same team, same tenure band — who have not resigned yet. They will tell you what is on the fence.

When all three sources point at the same cause, act with confidence. When they conflict, the exit interviews are usually the most honest of the three.

From diagnosis to intervention

Match the fix to the actual finding, not the fashionable one:

• Manager-driven attrition → targeted coaching or, sometimes, a hard conversation about whether the person should manage at all. Our leadership development work most often starts here.
• Early-tenure attrition → onboarding redesign and hiring-bar review.
• Growth-stagnation attrition → internal mobility, career frameworks, stretch roles.
• Compensation attrition → fix the specific bands that are off market. Across-the-board corrections are expensive and usually unnecessary.

Beware the universal reflex of throwing engagement events or pay hikes at every attrition problem. Pizza does not fix a bad manager, and a 10% correction does not fix a dead-end role.

Make it a rhythm, not a project

The companies that manage attrition well review a segmented dashboard quarterly, hold managers accountable for regretted exits in their teams, and predict — not just report — risk. None of this requires data science; it requires discipline and a willingness to look at the manager cut honestly.

If your attrition number is climbing and the explanations feel generic, a structured diagnostic usually takes three to four weeks. Contact us to scope one.