Coaching Exit Interview Questions: Eight Useful Prompts

Coaching exit interview questions help a departing client describe the service they actually experienced. They can reveal a useful conversation, a mismatch in expectations or a practical delivery problem. The interview is optional feedback, not a final attempt to persuade the person to renew or a test of whether they are ready to leave.

Use the eight prompts below in a written form or conversation, choosing follow-ups that fit the answer. The worked example follows a wrong preparation route through correction and client confirmation, then tests a shorter worksheet without pretending one person validates the redesign.

Prepare for feedback you may not like

Explain the purpose, approximate time, reader and alternatives. Let the client skip questions or decline. Keep feedback separate from money owed, agreed access, ordinary service corrections, referrals and public testimonials. Someone need not give a positive evaluation to receive the closing arrangements already promised.

AAPOR’s survey-research guidance recommends clear, specific, neutral questions, manageable burden, appropriate missing-answer handling and pretesting. Those principles inform these original service-feedback prompts. They do not make this a validated coaching assessment or require a small practice to claim population-level research.

Choose an interview when clarification is useful and wanted, not because a conversation must always be more honest. A written route gives time to think but may leave ambiguity. Be clear about the limits of both, including the possibility that clients will avoid criticism when speaking to the person who coached them.

For a twenty-minute conversation, allow two minutes for choice and purpose, fourteen for the most relevant questions and four for checking understanding and next actions. The eight prompts are a menu, not eight compulsory answers squeezed into that time.

1. How did the service compare with what you expected?

Start with the buyer’s understanding of the offer. Invite matches and mismatches without assuming a transformation or failure. Ask which part of the description or earlier conversation shaped the expectation.

A useful follow-up is “What did you expect would happen when you sent that document?” It can distinguish a promised editing service, a coaching discussion about writing, and a resource the client thought included individual feedback. Check the actual promise rather than attributing the mismatch automatically to misunderstanding.

Write the reported expectation separately from your investigation. “I expected tracked edits” is the client’s account. Whether editing was sold, omitted or never offered requires checking the offer, agreement and communications.

The action may be to provide missing agreed work, correct the description or explain a boundary. It need not be adding editing to every future package.

2. Which part of the work, if any, was most useful?

Allow the client to name a question, rehearsal, practical action or relationship feature. “If any” leaves room for nothing standing out. Avoid demanding a highest financial return where the person never used a financial measure.

Ask “What made that useful in your circumstances?” An answer about a short rehearsal may identify its concrete output, such as a clearer request, without proving that it caused a later promotion or agreement.

If the useful work was repeatedly outside the paid scope, examine that too. A client can value an unscheduled document review while the practice needs to define whether that work can be provided sustainably and competently.

A recurring preference is a reason to investigate, not proof that one exercise belongs in every engagement. Preserve less useful and mixed responses alongside favorable ones.

3. Was there support you wanted that was unavailable?

Separate missing delivery from an additional need. Ask what was expected, requested and answered. A promised worksheet that could not be opened is different from wanting a new resource that was never part of the service.

A person might have wanted specialist legal advice. That can point toward a suitable professional resource, not an invitation for the coach to invent a legal-advice service. The current ICF Code of Ethics addresses competence, accurate representation and clear agreements.

Ask “What response did you receive when you raised it?” rather than assuming the practice ignored the request. The answer may reveal a sound referral, an unanswered message, or a boundary that was communicated unclearly.

Decide whether to correct delivery, clarify scope or explore another appropriate resource. More content is not automatically the right response.

4. How did the pace and preparation fit your circumstances?

Ask about session frequency and preparation separately. Meetings can be well spaced while a worksheet is too long. A client who never used a resource may need a not-applicable answer rather than a forced rating of its quality.

Follow with “Which activity took more effort than was useful?” or “What would have been a workable alternative?” Ask what happened without assuming missed preparation proves low confidence, avoidance or lack of motivation.

A shorter spoken reflection, a relevant optional question or clearer instructions may fit. Keep any estimate of completion time distinct from measured timing. Do not claim a new form saves fifteen minutes if nobody measured the old one.

Record the actual suggestion and its context. One person describing a twelve-question worksheet as burdensome does not establish that every client dislikes preparation.

5. How did the included between-session support work for you?

Use this prompt only where support was actually included. Distinguish the channel, purpose, response arrangement and usefulness instead of combining them into one unexplained score.

“Was it clear when to use messaging?” and “Did the reply arrive within the agreed window?” ask different things. If someone did not use the channel, ask whether that was because no support was needed, the route was unclear, or another reason they want to share.

Compare the welcome information with what the practice actually delivered. Do not respond to unclear expectations by improvising unlimited availability. A client asking for support beyond the coach’s role needs an appropriate boundary and resource conversation.

The useful change could be one clear purpose and response clock. A greater number of messages alone does not establish a better service.

6. What made booking, access or administration harder?

Invite a concrete step: a resource that could not be found, unclear timezone information, an inaccessible document or an unexplained charge. The client should not have to diagnose the technical cause.

“I could not find it on my phone” describes an experience. It does not establish whether access, instructions, the account or the file caused the problem. Check those possibilities without asking the client to prove they deserve help.

Request only necessary details. Passwords, complete card details and unrelated private screenshots are not an appropriate feedback requirement. If a service problem needs investigation, move it into the proper support or complaint process.

Correct a clear rights or delivery problem even if only one person reports it. A majority vote is not needed to fix an incorrect charge or an unavailable purchased item.

7. Would you feel comfortable recommending this service to someone it suits?

An open answer may be enough. Ask what influenced it without requiring an introduction. A person can value coaching and still keep their participation private.

If you use Net Promoter Score, Bain’s measurement method uses a zero-to-ten recommendation scale: nine or ten are promoter responses, seven or eight are passive responses, and zero to six are detractor responses. Subtract the detractor percentage from the promoter percentage. These are scoring categories, not diagnoses of clients’ attitudes or permission to seek referrals.

Use consistent wording and method if you want a time comparison. Do not treat a casual open question as an NPS rating, silently replace skipped answers with zero or combine a phone answer and the same person’s form as two respondents.

The numerical example below shows the denominator and its uncertainty. A recommendation score does not measure every aspect of satisfaction, coaching effectiveness or future purchasing behavior.

8. What would you want me to do differently as a coach?

Invite a specific observation about your behavior, without making the client responsible for your professional development. They can decline or say there is nothing they want to add.

“I wanted more time before you introduced an exercise” suggests checking whether a tool would help and leaving space for thinking. “Challenge me more” needs clarification about what useful, respectful challenge means to that person.

Listen and check your understanding before evaluating what to change. You need not adopt every suggestion, but you can explain an appropriate next action without debating the client into a more favorable account.

Keep later supervision or service-design review within the relevant confidentiality arrangements. A client’s critical comment is not automatically public teaching material.

Send a complete optional invitation

In this fictional example, coach Vera invites ten clients whose individual engagements ended in October. They already have the appropriate workspace accounts. On 2 November 2026 she deliberately grants them access to the optional feedback form and sends its private client-access link with this invitation. The saved form uses /q/optional-service-feedback on her workspace domain; she creates no new Program or meeting:

Subject: Optional feedback about your coaching experience

Hello Alex,

I would welcome optional feedback about the service. Open the private Optional Service Feedback link included in this email while signed into your existing client account, or reply to arrange a twenty-minute conversation. The written response window closes at 17:00 on 13 November, Europe/London.

The questions concern expectations, useful work, preparation and practical arrangements. You may skip any question or decline. I will read the named responses for service improvement. They are handled under the privacy information already supplied for your service; this is not an anonymous survey and I will not send individual answers to a sponsor or publish them.

This choice does not change your settled account or agreed resource access, and it does not give permission to send a new offer or publish a quotation. If there is a service concern, you can raise it in this same business-email thread without completing the form.

Thank you, Vera

Alex confirms receiving the invitation on 2 November and opens the assigned form through its link in his existing account. He submits it on 5 November and asks for a conversation to clarify two answers. They agree 12 November, 10:00–10:20 London time. This is optional service feedback, not a paid coaching session deducted from his entitlement. Vera does not record the conversation; she checks her short written account with Alex.

Keep the seven-response calculation honest

By the 13 November cutoff, seven of the ten clients have submitted one form each. All seven answer the optional numerical recommendation item. The three nonrespondents’ views remain unknown. Alex’s later conversation adds context to his existing response; it is not an eighth respondent or rating.

The actual numerical item asks, “How likely are you to recommend this coaching service to a friend or colleague?” Its scale runs from zero, not at all likely, to ten, extremely likely. The open interview prompt can explore the answer, but it does not replace that recorded numerical item. Vera keeps the original seven ratings unchanged after the interviews; a useful explanation is not permission to recode an eight as a nine.

The score is (4/7 − 1/7) × 100 = 42.857, rounded to +43. Seven forms from ten invitations is a separate 70% participation figure. Do not divide the NPS numerator by ten or classify the three missing ratings as detractors.

For a sensitivity illustration only, if one rating of nine had instead been eight, the counts would be three promoter, three passive and one detractor responses. The score would be (3 − 1) ÷ 7 × 100, approximately +29. That is not an observed change; it shows how much one response can move a small result.

Report the seven-response cutoff and recruitment context. This self-selected group does not establish a market benchmark, actual referral behavior or a reliable population estimate. Keep comments and concrete service problems visible beside the number rather than using +43 as a general endorsement.

Follow Alex’s feedback through two different actions

Clarify the experience before interpreting it

In the interview, Alex says the conversations were useful but a preparation worksheet was hard to locate on his phone. He also found its twelve prompts more extensive than he needed. Vera initially suggests that he lacked confidence about completing preparation. Alex rejects that interpretation: he could not follow the instructions and wanted a smaller relevant task.

Vera withdraws the confidence explanation and checks her written summary aloud. Alex confirms the two distinct issues: finding the existing resource and whether its length was useful. Neither is a diagnosis of why he left.

They use the first two minutes to confirm the optional purpose, fourteen for those priority answers and four to agree the summary and next actions. The conversation ends at 10:20. Alex permits one service-resolution email about the route and volunteers for one separate ten-minute test of a shorter candidate, scheduled for 20 November at 10:00.

Correct the actual preparation route

Vera promises to investigate by 16:00 on 13 November. At 15:30 she confirms that the original worksheet and its task were assigned correctly. The welcome guide wrongly told Alex to find it inside a Program, although this was a standalone resource. She uses its private client-access link; the standalone task is not guaranteed to appear in My Coaching’s What to do next.

She corrects that guide and at 15:45 emails Alex the revised instructions and the private link for the already assigned worksheet. She does not ask him to buy access again or treat staff Preview as evidence of his experience.

At 09:10 on 16 November Alex confirms that he can open the correct worksheet on his phone using the corrected route. The delivery concern has an investigation, correction, actual communication and client confirmation. This does not establish that every other client’s guide or device has been checked.

Test a shorter candidate without claiming validation

For the development question, Vera prepares a separate three-prompt candidate rather than silently changing Alex’s historical response:

1. Which decision or conversation do you want to prepare for?
2. What information or authority is still missing?
3. What help do you need before that conversation?

The planned 20 November test takes ten minutes. Alex spends seven minutes completing the candidate and three discussing it. He finds the third prompt ambiguous: he reads “help” as a promise that the former coach will provide more between-session coaching. That is not the intended service.

Vera records the problem and changes the draft third prompt to “Which information do you still need to prepare your next meeting?” Alex hears the proposed clarification, but the revised candidate is not completed or independently retested in this case. The test is a completed observation of the first candidate, not validation of the revised one.

The W3C forms tutorial emphasizes labels that describe the purpose of controls. Clear wording and instructions matter as well as a form being technically available. One successful phone opening or one participant’s discussion does not establish general accessibility.

Vera makes no time-saving claim: the original twelve-prompt completion time was not measured. The two calls total twenty plus ten, or thirty feedback-conversation minutes, excluding investigation, drafting, client receiving effort and other administration. No whole-work cost or efficiency result is calculated.

Record the improvement and its actual state

The immediate route repair is complete. The development work has produced a candidate and a known issue, not evidence that all preparation should now be shorter. Vera keeps the original responses and versions distinct so a later evaluation does not compare incompatible questions as one unchanged measure.

Alex receives the factual action update after the test, and acknowledges it that day. No further test, renewal discussion or marketing contact is agreed. A new request would need its own appropriate choice; the feedback does not enroll him as a permanent unpaid evaluator.

Organize optional feedback in Coachful

We build Coachful. Coachful is our #1 recommendation for organizing optional questions, resource delivery and appropriate follow-through. This is our editorial workflow recommendation, not independent validation of the interview or proof that the software produced a service improvement.

Use a form with the selected prompts and explain who reads its responses. Assign it deliberately through Clients > client > Worksheets & resources > Assign, then send its private /q client-access link. A URL does not itself grant access, and a staff preview is not proof a client opened it. Creating an optional dated task does not make a standalone form appear automatically in My Coaching’s What to do next; that overview has its own eligible preparation, Journey, Program and continuing-content items.

Keep the reported experience, investigation, action and actual acknowledgment in separate fields. An internal completed task does not prove the client received a correction. A new candidate should have its own version rather than replacing what an earlier response referred to.

Do not infer publication permission from private feedback. Within its U.S. advertising scope, the FTC’s endorsement guidance requires an accurate account of the endorser’s experience; consent would not substantiate an invented result. This fictional feedback is not published as a testimonial.

Use the offboarding checklist for administrative closure, the satisfaction-survey guide for another feedback format, and Coachful’s features for the supported record workflow.

Frequently asked questions

Should I interview an unhappy client?

Offer an optional suitable route and address a complaint through its proper process. They do not need to join a call to have a service concern considered.

Are written answers less useful than a conversation?

Not necessarily. Written answers provide reflection time; a conversation can clarify ambiguity. Neither format guarantees candor, and the same person’s follow-up is not another respondent.

Does a recommendation rating authorize referrals?

No. It is an answer to that question, not permission to request names, publish praise or send promotional messages. Alex’s rating of eight also does not erase his concrete delivery concern.

Can one interview justify a service correction?

Yes, where it identifies a real error, as the wrong guide did here. A broader redesign needs its own evidence; the revised short candidate remains untested.

What if the client has nothing critical to say?

Respect that answer. Do not keep reformulating the question until the person supplies a flaw, or treat a lack of criticism as proof that every service element worked.