HR · Exit Interview Theme Analysis
Upload your exit interview transcript and let the AI analyze recurring concerns, severity, and actionable recommendations.
Describe the task
Upload your exit interview transcript and let the AI analyze recurring concerns, severity, and actionable recommendations.
Persona: People Analytics lead reviewing exit-interview transcripts.
Here is an exit interview transcript. Analyse it and return: 1. The top 3 recurring concerns Jordan raised, ranked by how often / how strongly they came up. 2. For each concern, give 2 verbatim quotes from the transcript that justify it. 3. For each concern, give a severity rating (low / medium / high) with a one-line reason. 4. For each concern, suggest one concrete action the People team could take. Be concise. Use the transcript only — do not invent quotes.
Before you start
- Open
exit-interview-transcript.txtand either attach the file to the chat or copy the entire contents into the prompt. - Decide what kind of output you want — themes, quotes, and recommendations (the prompt above asks for all three).
- Start a new chat to ensure there is no leftover context from previous conversations.
Sample Files
exit-interview-transcript.txt
31-minute transcript of an exit interview between People Ops and a departing senior engineer (Jordan Reyes), containing signals on promotion paths, management style, and compensation.
What RedbrickAI does
When you paste this prompt with the transcript, the AI performs a structured linguistic analysis to identify and rank the top three recurring concerns. It extracts verbatim quotes to ground each theme in evidence, assigns severity ratings with a one-line rationale, and synthesises the findings into concrete, actionable recommendations for the People team.
Example output
1) Lack of clear career growth / promotion path
Jordan said his growth “just stopped,” described unclear staff-engineer progression, and repeated that the answer kept being vague or inconsistent. This was the strongest and most repeated concern in the transcript.
- Quotes: “Honestly, the growth just stopped for me.”
- “Every time I asked about it in a one-on-one I got something vague like ‘let's revisit next quarter.’”
Severity: High — Jordan says a concrete promotion plan or comp adjustment “would have kept” him, and the lack of clarity was the hardest part.
Action: Have managers give senior engineers a documented promotion roadmap with specific milestones and review dates.
2) Weak manager ownership of career conversations / one-on-ones
Jordan repeatedly said his one-on-ones became status updates rather than meaningful career discussions, and that attempts to redirect them did not stick.
- Quotes: “They were status updates, mostly.”
- “Make managers actually own career development.”
Severity: High — Jordan explicitly linked this to losing his next step at the company and said the company would lose fewer people if it changed.
Action: Require managers to use regular one-on-ones for career development with a tracked agenda and follow-up actions.
3) Compensation not keeping pace with market / internal progression
Jordan said pay was a factor, noted very small annual raises, and said the company was no longer paying market rate for senior engineers in his region.
- Quotes: “I don't think we're paying market rate for senior engineers in this region anymore.”
- “The last two annual reviews, my raises were 3 and 3.5 percent, which is essentially flat after inflation.”
Severity: Medium — Jordan said compensation was “not the biggest” reason, but it materially contributed and there was no counter-offer.
Action: Run a senior-engineer market pay review and adjust comp bands or salaries where they are below market.
What a good answer looks like
| Fact | Expected wording |
|---|---|
| Theme 1 — Career growth & promotion path | High severity. Key signals: "Growth just stopped", "no clear path anymore", "let's revisit next quarter". Brought up 4–5 times, contradicted by other promotions in March. |
| Theme 2 — One-on-ones used for status updates | High (or Medium) severity. Key signals: "They were status updates, mostly", "I gave up trying to redirect it". Raised once, drifted back. |
| Theme 3 — Compensation lagging market | Medium severity. Key signals: "Not paying market rate anymore", 28% jump at new role, 3 / 3.5% raises = flat after inflation. |
Pass
Identifies all 3 main themes, quotes are real (in the transcript), severities make sense, recommendations are concrete and actionable.
Partial pass
Gets 2/3 themes, or quotes paraphrased instead of verbatim, or recommendations too vague ("communicate better").
Fail
Misses promotion/comp themes, invents quotes not in the transcript, or refuses the structured format.
Try follow-up prompts
Follow-up prompts let the AI build on the conversation context it already has — you don't need to repeat the setup. This tests multi-turn memory and shows how the system handles nuanced edge cases.
Initial prompt
“Are there any minor signals worth flagging that didn't make the top 3?”
AI response
Yes — a few secondary signals came up that did not make the top 3 but are still worth flagging. Jordan mentioned out-of-date runbooks and a heavy on-call rotation as stress points, and he also said the CI pipeline is very slow, describing it as “death by a thousand cuts.”
Another minor signal is that Jordan said he had two other engineers on his team who had similar recruiter conversations and were at least open to leaving, which suggests the compensation concern may not be isolated.
He also noted that he still believed in the company’s product and would recommend the company to other engineers, so the transcript contains a positive retention signal alongside the concerns.
Follow-up prompt
“Draft a short, anonymised summary I can include in the quarterly People dashboard — 3 sentences max.”
Updated AI response
One exit interview highlighted recurring concerns around limited career progression clarity, inconsistent manager ownership of development conversations, and compensation lagging market expectations for senior engineers.
Secondary signals included frustration with operational friction such as outdated runbooks, a heavy on-call load, and a slow CI pipeline, described as cumulative “death by a thousand cuts.”
There was also a broader retention risk signal, as the employee noted peers had similar recruiter conversations and were open to leaving, although they still spoke positively about the company’s product.