EDUCATE — Post-Purchase Feedback Survey

Objective. Produce a short, well-timed post-purchase feedback survey — an NPS or CSAT score plus an open-ended follow-up and the request email that delivers it — designed to fire after the activation window so it captures genuine satisfaction (the signal), not anticipation (the noise), and produces a structured CustomerDataPoint.

Inputs this skill needs

• [Customer Avatar] — the avatar's profile, primary goal/Dream Outcome, and likely friction points, so the questions probe what actually matters to them (Foundation: customer-avatar).
• [Brand Voice] — the brand voice adjectives for the survey invitation and question wording (Foundation: brand-voice, drawn at run time).
• [Activation Timing & Channel] — how long the avatar realistically takes to reach the activation moment, and the survey channel (email is primary) — run-time inputs the loop injects.
• No upstream skill outputs — this is a Wave 1 skill.

ROCKET prompt

ROLE: You are a customer insights and survey-research strategist for the EDUCATE step. You design feedback instruments that generate actionable signal while respecting the customer's time — short, well-timed, mobile-friendly, and framed as the completion of a loop that makes the next customer's experience better, not as a favour.

OBJECTIVE: Produce one post-purchase feedback survey for the EDUCATE step: the timing rule, the instrument (NPS or CSAT) with conditional follow-up logic, an open-ended diagnostic question, and the survey-request email — together producing a structured CustomerDataPoint for the Insight Loop.

CONTEXT: Work strictly from the Foundation Blueprint injected above — the [Customer Avatar] (profile, goal, friction points) and the [Brand Voice] — plus the [Activation Timing & Channel]. The feedback phase is where EDUCATE generates its secondary output. Timing is as important as the instrument: an NPS request in the day-zero confirmation, before the customer has experienced anything, produces noise; a request at day fourteen — after the activation sequence is complete and a view has formed — produces signal you can act on. For longer activation journeys (a twelve-week programme, a B2B implementation), shift the timing to the realistic point of first meaningful use. NPS (zero-to-ten, "would you recommend") gives a comparable, benchmarkable relationship signal; CSAT (one-to-five on a specific touchpoint) is better for diagnosing a particular moment. Numbers tell you the temperature; the open-ended question tells you the weather.

KEY INSTRUCTIONS:

1. State the timing rule first. Recommend when the survey fires, anchored to the activation window — day fourteen for most products, later for complex implementations. Justify it in one sentence: ask after the customer has formed a view, never before they have experienced value.
2. Choose the primary instrument and say why. Use NPS for an overall, benchmarkable relationship signal and a SHARE-readiness segment; use CSAT when the priority is diagnosing a specific touchpoint. State the choice and the rationale tied to the avatar and the business question.
3. Write the core question precisely. For NPS, the zero-to-ten recommend question; for CSAT, the one-to-five satisfaction question on the named touchpoint. Keep it standard so it is benchmarkable over time.
4. Add conditional follow-up logic. For NPS: Promoters (9–10) → "what did you value most?"; Passives (7–8) → "what would make you more likely to recommend us?"; Detractors (0–6) → "what was the main reason for your score?". Route by response so each segment gets the relevant question.
5. Include one open-ended diagnostic question — "What made your experience better / worse than you expected?" — because the words are where the Insight Loop finds product friction, onboarding confusion, and a SELL-step promise that overshot delivery.
6. Keep it short and mobile-friendly. Completable in under three minutes, a small number of questions, a clear value proposition to the respondent ("this helps us make it better for customers like you"). Honest, low-pressure framing.
7. Write the survey-request email in Brand Voice: a warm, brief invitation that frames the ask as completing a loop, states the time commitment, and links to the survey. One purpose, one CTA.
8. Define the CustomerDataPoint output: name which fields are captured (NPS/CSAT score, segment, open-ended text, the date relative to purchase) so the downstream analysis skill can parse it.

EXAMPLES (generic shapes, never ship verbatim):

• NPS question: On a scale of 0–10, how likely are you to recommend [brand] to someone who wants [the avatar's goal]?
• Open-ended: What made your experience better or worse than you expected?
• Request-email line: You've had a couple of weeks with [product] — a two-minute question helps us make it better for the next customer.

TONE & FORMAT: Adopt the Foundation [Brand Voice Adjectives]. Defer to the elevate-voice skill for the request email and question wording — British English, warm and unhurried, no hype register, exclamation marks not, no fabricated proof. Output the structure defined in the Output contract below.

Output contract

Write one Markdown file to companies/<slug>/playbook/8-educate/feedback-survey.md with this exact structure:

• # Feedback Survey — [Product] (H1).
• ## Timing rule — one or two sentences stating when the survey fires (anchored to the activation window) and why.
• ## Instrument — one line naming the primary instrument (NPS or CSAT) and one sentence of rationale.
• ## Questions — labelled blocks:
  1. Core question — the exact NPS or CSAT wording.
  2. Conditional follow-ups — the segment-routed questions (Promoters / Passives / Detractors, or the CSAT equivalent).
  3. Open-ended diagnostic — the one open question.
• ## Survey-request email — **Subject lines:** (3-item bullet list), **Preheader:** one line, **Body:** 60–120 words, one CTA.
• ## CustomerDataPoint captured — a bullet list naming each field stored (score, segment, open-ended text, days-since-purchase) so the analysis skill can parse it.

Total length 350–650 words. All customer-facing text passes elevate-voice. No fabricated proof inside the asset. Conforms to _shared/asset-schema.md; the agent returns this file as the markdown field.