Follow up.
Don't assume.

Edition 2.0 - reviewed September 18, 2026

A useful follow-up asks a clear question through an authorized channel and gives an unresolved concern a responsible human owner. It does not diagnose by text, invent a service interval or promise that every customer will return. This complete free workbook includes the source ledger, finished messages, a last-minute withdrawal, outcome reconciliation and all ten answers.

North Quay Auto and every record are fictional. The exercise sends nothing and uses no actual customer or vehicle data. It covers communication operations, not diagnosis, emergency response, warranty entitlement or a declaration that a vehicle is safe. No essential worksheet or answer is withheld behind another purchase.

For service-desk staff or an owner, allow an estimated 30-45 minutes. You need only the supplied packet, paper or a spreadsheet and a calculator. Optional Kimi drafting is not necessary; the rehearsal requires no software purchase, while optional tools may offer paid plans. Adaptation to a real shop needs approved contact purposes, routing and escalation procedures, not just copied text.

1. Separate a service check-in from a promotion

The review time is September 24, 2026 at 10:00 UTC. A manager has approved one optional service check-in for completed visits whose customer accepted that specific contact purpose and route. It contains no promotion, review link or maintenance-due claim. These are supplied exercise rules, not universal legal classifications of messages.

Each ID represents one verified customer and visit. Private routes are known to authorized staff but are deliberately absent here. The raw import has eight rows: the seven rows below and an identical repeat of F02. The repeated row does not create a second permission or a second message.

The routine initial-check-in set is F01 and F02. F03 is withdrawn, F04 is not completed, F05 lacks the specified permission, F06 goes to the concern owner rather than a routine sequence, and F07 already has its message. Do not choose recipients based on whether staff think they were happy.

The September dates identify past visits, not maintenance intervals. The FTC's consumer guidance points to the manufacturer's schedule for the vehicle and driving conditions. This workbook supplies no mileage, due-date calculation or technical recommendation. Never turn "visited three months ago" into "your brakes are due."

2. Use complete messages with bounded claims

Draft A, initial check-in for F01:

"Hello from North Quay Auto. You agreed to one follow-up about your completed visit. Do you have a question about the paperwork or a concern you would like our service team to review? Please reply through this existing private channel, without sending payment-card information or identity documents. We cannot assess vehicle safety or diagnose a problem in this message. If you do not want further optional follow-up messages, let us know and we will update that preference."

This is deliberately not "everything still running perfectly?" A question that assumes success can discourage useful information. Nor does silence mean satisfaction, successful repair or permission for another sequence. The one-message rule remains in force unless a separate approved contact purpose applies.

Draft B, concern acknowledgement for F06:

"Thank you for telling us there is a concern after your visit. We have assigned the record to our service lead for review through this private channel. This reply does not diagnose the issue, confirm warranty coverage, promise a free repair or say the vehicle is safe to drive. If you believe there is an immediate danger, seek appropriate emergency or roadside assistance rather than waiting for a routine message. Our next administrative update is due by 12:00 UTC today; that is an update commitment, not a repair completion time."

The packet supplies service lead Mira as the concern owner and the 12:00 update commitment, approval H06. No technician finding or warranty decision is supplied. Keep F06 out of automatic reminder or review-request queues while Mira handles the concern. This is a routing hold, not a reason to suppress a customer's honest public feedback.

Draft C, acknowledgment of withdrawal:

"We have recorded your request not to receive optional follow-up messages in this channel. Your preference has been updated, and this optional sequence is stopped. This does not cancel an existing appointment or resolve an open service concern; our team will review any necessary communication separately under your actual arrangements. Thank you for letting us know."

Use that statement only after the preference update is recorded. A draft acknowledgment without an actual update must not claim the action is complete. An unsubscribe also must not silently erase unresolved service records that staff still need to handle responsibly.

3. Recheck the send list and reconcile outcomes

At 10:04, F02 withdraws the check-in permission under W02. The final send decision is at 10:05. No other record changes. The eligible set becomes F01 only. Cancel F02's queued optional message and record W02. Do not send first and process preferences later.

The worked outcome branch assumes the manager approved F01's current copy and verified route. It then supplies these events; they are fictional evidence, not results from a live provider.

Three rows describe two unique events and one message. There is one delivered message, not two deliveries to count as separate customer contact. At 10:10 a customer reply R01 asks whether an appointment next Monday is available. This is one reply and one enquiry, not an accepted booking. No time, scope, resource check or acceptance has been supplied.

Reply draft: "We can ask the scheduling team to review next Monday's availability. We have not booked a time. Please continue through this private channel so the team can check the required scope and resources before proposing an appointment. No payment or additional work has been authorized by this enquiry."

If a provider accepts M01 but no delivered event appears, classify delivery as unknown. Reconcile the original message key before retrying or switching channel. If the preference register cannot be read immediately before dispatch, hold the message; an old permission record is not current evidence.

4. Report outcomes without claiming retention gains

Use a separate supplied cohort to practice reporting. Six distinct customers completed an initial visit during September 1-7. Every one remains in the denominator through the September 30 cutoff. A repeat means at least one additional completed paid visit in that period, not a reply, enquiry or reserved appointment.

The observed repeat proportion is 2/6 = 33.3%, rounded to one decimal. There is no comparison group, spending record or causal study. The figure does not establish that follow-up created those visits, that C05 was lost, or that C06 is dissatisfied. Report supplied states rather than guessing motives.

Finished owner summary: "The separate September cohort recorded two additional completed paid visits among six original customers, an observed repeat proportion of 33.3% by September 30. One enquiry and one future booking were not counted as completed returns. This packet cannot establish an improvement, revenue effect or universal retention rate. Review unresolved concerns independently of marketing metrics."

If someone proposes a free remedy or a service-due reminder, request the actual supporting record. A manufacturer's schedule, qualified assessment or written coverage may be relevant; none is replaceable with a universal interval generated from a guide title. A helpful communication system knows when it lacks the facts to continue.

5. Reuse the handoff and challenge the exclusions

Independent challenge starts from the original 10:00 ledger, before W02. F05 now explicitly agrees to the same optional check-in under P05, while F01 reports a concern requiring human review. All other facts stay unchanged. The routine set becomes F02 and F05. The total remains two, but the identities differ; a count-only check would miss the error.

Complete handoff: "F01 moved from routine follow-up to concern review. F05 now has P05 and may enter the initial-check-in list. F02 remains eligible in this independent branch; W02 has not occurred here. Keep F06 with Mira. Verify the exact two records before any real dispatch, not merely a target count of two."

Optional Kimi prompt: "Use only the synthetic seven-row F01-F07 ledger and the identical F02 replay. Compute initial eligibility, apply W02 at 10:04, and reconcile D01/D02 plus the replay. Classify R01 as a question, not a booking. Separately compute the six-person cohort repeat proportion and solve the independent F05-permission/F01-concern challenge. Return exact IDs and reasons. Do not recommend service intervals, diagnose, promise a remedy, infer satisfaction or send anything."

Supply the full tables with the prompt. Check every returned ID manually. Provider availability, account settings and response quality are outside this exercise; no AI execution is claimed.

6. Ten checked follow-up answers

1. How many distinct visits are in the first ledger? Seven from eight imported rows; the extra F02 payload is an exact replay.

2. Who initially receives a routine check-in? F01 and F02 only, under the supplied purpose, permission, completion and prior-contact rules.

3. Why does F05 stay out initially? A receipt contact route is not the specified check-in permission. Do not expand its purpose by assumption.

4. Why route F06 separately? Concern C06 needs Mira's human review under H06, not routine reassurance or an invented warranty decision.

5. What changes at 10:05? W02 removes F02 before dispatch. F01 is the only remaining routine recipient.

6. How many messages and deliveries do D01-D02 show? One message and one delivered outcome across two unique events; the replay is excluded.

7. Does R01 create a booking? No. It is one reply and enquiry; resource checks, exact proposal and customer acceptance are absent.

8. What is the separate cohort result? Two completed paid repeat visits among six customers, or 33.3%. A future booking and question do not count.

9. Who qualifies in the independent challenge? F02 and F05. F01 moves to human concern handling; this branch starts before W02.

10. May the assistant promise a free repair or a standard interval? No. Actual coverage, manufacturer guidance and qualified assessment must support any such communication; this packet provides none.

Sources and limits

• FTC: Auto Repair Basics supports checking manufacturer schedules and actual written coverage rather than assuming universal intervals or warranties.

• Mailchimp: The importance of permission addresses marketing permission. This workbook's service-contact rules are supplied fiction, not a claim of legal compliance.

• Kimi: official site identifies an optional drafting assistant; no privacy, plan or performance guarantee is made.

Sources reviewed September 18, 2026. Editorial reviewer role: CreatorHQ service-communication editor. Edition 2 removes universal maintenance advice, free-remedy promises, selective review requests and guaranteed loyalty. It replaces them with a complete free exercise in permission, human escalation, state reconciliation and limited outcome reporting. Real safety concerns require appropriate human assistance, not delayed AI reassurance.