Customer Review System That Generates Valid Review Schema

Act as a structured-data engineer and customer-marketing strategist. Build a customer review collection system whose output maps cleanly to Schema.org Review + AggregateRating, so my reviews are marked up accurately, can be cited confidently by AI answer engines, and can show stars in Google search where the reviewed type is eligible.

Business context:
- Name + what we do: [NAME — SENTENCE]
- What's reviewed: [Product / Service / LocalBusiness / SoftwareApplication / Course] · its name: [NAME]
- Page where reviews display: [https://YOUR-DOMAIN.COM/...]
- Rating scale: [1-5 stars / 1-10 / thumbs / NPS 0-10]
- How customers reach you: [email / SMS / in-app / POS / post-call] · review-request trigger: [DESCRIPTION]
- Brand voice: [warm-expert / no-nonsense / playful / premium / technical / friendly-everyday]
- Existing review platforms: [GBP / Trustpilot / G2 / Capterra / Yelp / NONE]

Build six artifacts:
1. Trigger logic — event + delay + qualification rules, one paragraph
2. Review-request message — email/SMS/in-app copy, ≤120 words, brand voice, one clear ask + form link
3. Structured intake form spec — table: Field · Type · Required · Maps to Schema.org field · Help text. Must capture reviewer name, optional company, rating, review body, optional photo URL, consent to publish
4. Reviewer-side micro-prompt — 1-2 sentences shown inside the form (length target, what to mention, what to skip)
5. JSON-LD generator — code (Python or Node) that takes an array of review objects and outputs (a) a Review JSON-LD per review and (b) AggregateRating recomputed from the array; skip reviews with body under 50 characters or missing rating
6. Refresh rule + pre-publish audit — when to recompute AggregateRating; checklist (AggregateRating count = visible Review count; every Review has required fields; every Review.itemReviewed @id exists on the page); Rich Results Test URL; any default you applied where I gave no context

Schema requirements: each Review needs @type Review, author {@type Person, name}, datePublished ISO, reviewRating {@type Rating, ratingValue, bestRating, worstRating}, reviewBody (≥50 chars), itemReviewed @id. AggregateRating needs @type AggregateRating, ratingValue (computed), reviewCount (published only), bestRating, worstRating, itemReviewed @id matching.

Star eligibility: Google shows review stars only for the types its review snippet documentation supports; of the types above, those are Product, SoftwareApplication and Course. It does not show them for reviews a business collects and publishes about itself under LocalBusiness or any Organization type, and Service is not a supported type. State in artifact 6 which case applies to me, and promise stars only where the type is eligible.

Collection rules, applied to all six artifacts: the request goes to every customer who reaches the trigger; no satisfaction question, score or filter decides who is asked or who reaches the public form, and qualification rules test only that the purchase or service really happened. Every genuine review is published and counted in AggregateRating whatever its rating; the generator skips a review only for the two reasons in artifact 5. Offer no incentive for a review left on Google; an incentive for a review on my own site goes to every reviewer whatever the rating and is disclosed beside the review. Google's review policy prohibits selectively soliciting positive reviews, discouraging negative ones and offering incentives, and FTC guidance says not to ask only customers expected to be positive and to disclose any incentive.

Success means the generator code runs as written on a sample array and its output validates against the requirements above.

Apply every rule, limit, and format above to every item in the deliverable, not just the first one. Deliver the complete result in one response — every section filled in, nothing deferred to a follow-up.

If a required input is missing, ask for it once before starting; otherwise proceed and label any assumption you made.

Notes: R40 refresh; tuned to vendor prompting guidance