Review Generation Engine

Review Generation Engine

Overview

Product reviews are the most trusted form of social proof — 88% of shoppers consult reviews before purchasing, and products with 50+ reviews convert 4.6% better than those with none. The fastest path to high review volume is a systematic post-purchase request sequence triggered by delivery confirmation, not order placement. Dedicated review apps (Judge.me, Yotpo, Stamped) handle the entire request flow, photo upload UI, and verified buyer badges without custom code.

When to Use This Skill

• When a new product has fewer than 10 reviews and needs social proof to convert
• When overall review volume is low despite healthy order volume
• When wanting to add photo or video review incentives to an existing text-only system
• When migrating to a new review platform and needing to rebuild the request sequence

Core Instructions

Step 1: Choose the right review platform

Recommendation: Use Judge.me for most stores — the best feature-to-price ratio, free plan allows unlimited reviews, and $15/month unlocks photo reviews (the single highest-converting feature).

Step 2: Configure your review request sequence

---

Shopify with Judge.me

1. Install Judge.me from the Shopify App Store 2. Go to Judge.me → Settings → Review Requests 3. Configure the request timing:

• First request: 7 days after order fulfillment (not order placement — wait until the product is received and used)
• Second request (if no review): 14 days after first request

4. Go to Judge.me → Settings → Emails and customize:

• Email 1 (day 7): Simple request with star rating widget — clicking a star pre-fills the rating in the form
• Email 2 (day 21): Add a photo review incentive — "Add a photo and get 15% off your next order"

5. Enable Verified Buyer badge — displayed automatically on reviews from confirmed purchasers 6. Go to Judge.me → Settings → Moderation and set auto-publish rules:

• Auto-publish: ratings 3–5 stars
• Hold for manual review: ratings 1–2 stars

---

WooCommerce with Judge.me

1. Install the Judge.me for WooCommerce plugin 2. Configuration mirrors the Shopify setup 3. Alternatively, use Yotpo for WooCommerce if you want UGC and social sharing built in

For WooCommerce's built-in reviews + AutomateWoo: 1. Enable WooCommerce → Settings → Products → Reviews 2. Create an AutomateWoo workflow:

• Trigger: Order Delivered (or Order Status = Completed + 7 day delay)
• Action: Send Email with a link to the product review page

3. This is free but lacks photo review support and star rating widgets — use Judge.me for higher conversion rates

---

BigCommerce with Yotpo

1. Install Yotpo from the BigCommerce App Marketplace 2. Go to Yotpo → Reviews → Mail After Purchase and configure:

• Send timing: 7 days after order fulfilled
• Enable follow-up email: 14 days if no review submitted

3. Configure the smart prompts — Yotpo's review form is embedded in the email itself, allowing customers to submit a rating without leaving the email

---

Step 3: Set up photo review incentives

Photo reviews increase conversion rate 4× more than text reviews. The incentive should reward the photo specifically, not the review itself (incentivizing reviews violates FTC guidelines; incentivizing photo submission is permitted with proper disclosure).

In Judge.me: 1. Go to Judge.me → Settings → Coupons 2. Enable Photo review coupon: "Add a photo to your review and receive 15% off your next order" 3. Judge.me issues the coupon automatically after a photo review is approved

In Yotpo: 1. Go to Yotpo → Reviews → Incentives 2. Enable Media Reviews incentive with a discount code

Step 4: Display reviews on product pages

All review apps provide a widget snippet or theme block. For Shopify:

1. Go to Judge.me → Widgets → Review Widget 2. Install the widget in your theme via Shopify → Online Store → Themes → Customize → Add Section → Judge.me Review Widget 3. Place the review widget below the product description on product pages 4. Enable Review Carousel for the homepage to display your best reviews site-wide

Step 5: Measure review program health

If submission rate is below 5%, the request is being sent too early (before delivery) or the form has too many required fields. Set the request to fire 7 days after fulfillment and reduce the form to star rating + 1 text field.

Step 6: Custom / Headless

For headless stores, use a review platform API (Judge.me, Yotpo, or Stamped all have REST APIs) rather than building a custom review system. The review collection, moderation, and display widgets are all available as embeddable components.

If you must collect reviews via a custom form, generate a tokenized review link that pre-identifies the customer and product so they do not need to log in:

// Generate a signed review link — valid for 30 days
function generateReviewToken(orderId: string, productId: string, customerId: string): string {
  const payload = { orderId, productId, customerId, exp: Math.floor(Date.now() / 1000) + 30 * 86400 };
  return jwt.sign(payload, process.env.REVIEW_JWT_SECRET!);
}

// Include ?rating=4 in the link when the customer clicks a star in the email
// This pre-fills the rating on the form — single most important UX optimization for review volume

// Schedule review requests via your order webhook (Shopify/WooCommerce/BigCommerce)
// Trigger: order status transitions to "delivered" (use carrier tracking webhook)
// Step 1: 7 days after delivery
// Step 2: 14 days after delivery (if no review submitted yet) — add photo incentive
// Cancel remaining steps when a review is submitted

Best Practices

• Time the request to after confirmed delivery — asking for a review before the product arrives damages trust; use your platform's order fulfillment status, not order placed
• Make the star rating widget clickable in the email — each star links to the review form with ?rating=N pre-filled; this single feature typically doubles review submission rates
• Never pay cash for positive reviews — financial incentives for reviews violate FTC guidelines; only offer incentives for adding media (photos/video) to an existing review
• Verified buyer badge drives conversion — mark reviews from confirmed purchasers and display it prominently; shoppers trust verified reviews significantly more than anonymous ones
• Respond to negative reviews publicly — brands that respond to 1–2 star reviews with empathy and resolution offers build more trust than brands with only 5-star reviews
• Cap at 2 review requests per order — a third request causes unsubscribes; two requests (day 7 + day 21) is the maximum

Common Pitfalls

Related Skills

• @social-proof-widgets
• @ugc-campaign-management
• @email-marketing-automation
• @loyalty-program-optimization
• @customer-retention-engine

Post-Purchase Review Collection Automation

Problem/Feature Description

A mid-sized e-commerce brand sells physical goods directly to consumers and wants to grow its product review volume without hiring a dedicated customer success team. They currently have zero automated outreach — reviews only come from customers who proactively seek out the form. The engineering team has a background job queue already running (BullMQ) and a basic order management system with delivery webhook support. They want to add systematic, multi-touch review request outreach that starts after each order is confirmed delivered and stops automatically as soon as a customer submits a review.

The head of growth has specifically asked that the solution not feel spammy — customers should get a first gentle nudge shortly after delivery, a light follow-up only if they haven't engaged, and one final ask with a meaningful incentive to capture photo reviews. The business also ships several hundred orders per week, so the team needs the approach to be queue-based, idempotent, and clean.

Output Specification

Implement a TypeScript module called review-scheduler.ts that exports:

• A scheduleReviewRequests(order: Order) function that enqueues all outreach steps for a delivered order
• A onReviewSubmitted(orderId: string) function that cancels any remaining scheduled outreach for that order

Also write a short DESIGN.md documenting the timing of each step, the channel used for each step, and the rationale for when each step fires.

Use the following type definitions as a starting point (add whatever additional types you need):

interface Order {
  id: string;
  customerId: string;
  lineItems: Array<{ productId: string }>;
}

You may assume a reviewRequestQueue BullMQ queue is available as an import. Do not implement the actual email/SMS sending logic — just the scheduling.

Review Moderation and Rating Aggregation System

Problem/Feature Description

An e-commerce platform has been accumulating reviews in a pending queue and needs a backend pipeline to automatically moderate them and update product ratings. The site has been targeted by waves of fake positive reviews — several products have been flooded with suspicious 5-star entries from a small set of IP addresses, and others show unusual bursts of activity from single accounts. The team has also noticed some reviews that are too short to be useful, contain nonsense characters, or awkwardly disclose that the reviewer received a benefit. They want a defensible, auditable moderation layer rather than relying entirely on manual review.

After moderation, approved reviews should be reflected in the product's star rating immediately, and the product page served from the CDN should be refreshed to show the latest rating. For reviews that include photos, an incentive (discount code) should be sent to the reviewer after approval.

Output Specification

Implement the following in a file named moderation.ts (TypeScript):

• A moderateReview(reviewId: string) function that inspects a single review and returns a moderation decision
• A processReviewModeration() function that processes all pending reviews
• A refreshProductRatingAggregate(productId: string) function that updates the product's aggregate rating

Also write a MODERATION_RULES.md file documenting each flag your system can raise, the condition that triggers it, and the decision logic that maps flags to outcomes.

You may assume these imports are available: db (ORM), subHours, subDays (date-fns), checkProfanity, createUniqueDiscount, sendReviewThankYouEmail, refreshProductRatingAggregate, and invalidateProductCache. Do not implement those imported functions.

Customer Review Submission Endpoint

Problem/Feature Description

A growing direct-to-consumer brand has recently switched from a third-party review app to a custom-built solution. Their backend is a Node.js/Express API backed by a PostgreSQL database (accessed via a db ORM object). Review request emails are already going out with a special link, but clicking that link currently leads to a 404. The engineering team needs to build the review form rendering and submission endpoints from scratch.

A key product requirement is that customers should be able to submit reviews without creating an account or logging in — the review link should carry enough authentication context on its own. The team also wants the system to capture the star rating a customer clicked in the email, so the form feels pre-filled and reduces abandonment. Photo uploads should be encouraged, with an incentive automatically queued for later fulfillment once the review passes quality checks.

Output Specification

Implement the following in a file named review-endpoints.ts (TypeScript, Express handlers):

• A generateReviewToken(orderId, productId, customerId) function that returns a signed link token
• A GET /review/:token handler (renderReviewForm) that verifies the token and renders the form
• A POST /review/:token handler (submitReview) that validates and saves the review

You may assume the following are available as imports: jwt (jsonwebtoken), db (ORM with db.products, db.productReviews, db.pendingIncentives), processReviewPhotos, onReviewSubmitted, and an Express Request/Response types. Use process.env.REVIEW_JWT_SECRET for the JWT secret. Do not implement the actual email sending or photo storage — focus on the endpoint logic.

Also produce a short NOTES.md describing the token lifecycle (how it's generated, what it contains, when it expires) and the photo incentive flow.