Skip to content
TheChatArchiveThe Chat Archive
o3-2025-04-16· OpenAIHugging Face LMArena Human Preference 140K

Write a PRD for a review platform

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-5-sonnet-20241022’s answer.
  1. User

    Write a PRD for a review platform

  2. Assistant

    Product Requirements Document

    Product: OmniReviews – Universal Review Platform
    Author: [Your Name / Team]
    Last Updated: YYYY‑MM‑DD


    1. Purpose & Vision

    Create a trustworthy, user‑friendly platform where anyone can discover, write, and manage reviews for products, services, and experiences. OmniReviews aims to become the “go‑to” destination for high‑quality, well‑moderated reviews that help users make confident decisions.


    2. Goals & Objectives

    #GoalKPI / Success Metric
    G1Grow an engaged community100k MAU by Month 6
    G2Collect high‑quality contentAvg. review rating ≥ 4.2/5, 20% returning reviewers
    G3Ensure trust & safety≤ 1% flagged reviews, 95% disputes resolved in 48 h
    G4Monetize responsibly≥ $2 ARPU by Month 12 via ads & affiliate links

    3. Stakeholders

    • Product Management (Owner)
    • Engineering (Frontend, Backend, DevOps)
    • Design (UX/UI)
    • Data & Analytics
    • Trust & Safety / Moderation
    • Marketing & Growth
    • Legal & Compliance


    4. Target Users / Personas

    1. Researching Rachel (25‑45): reads reviews before any purchase.
    2. Expert Ethan (30‑55): enjoys sharing in‑depth, knowledgeable reviews.
    3. Seller Sam (SMB owner): monitors & responds to feedback, boosts reputation.

    5. User Stories (🔑 = MVP)

    IDAs a…I want to…So that…MVP?
    U1VisitorSearch for a product/servicedecide if it fits my needs🔑
    U2VisitorView ratings distribution & top reviewsgauge overall sentiment🔑
    U3Registered UserWrite a text + star‑rating reviewshare my experience🔑
    U4Registered UserUpload photos/videosadd proof & context🔑
    U5Registered UserEdit / delete my reviewcorrect or remove content🔑
    U6Registered UserUp‑vote “helpful” reviewssurface best insights🔑
    U7Business OwnerClaim my business profilemanage info & reply
    U8ModeratorFlag / remove abusive contentkeep platform safe🔑
    U9Data AnalystAccess aggregated analyticsidentify trends

    6. Functional Requirements

    6.1 Core Review Flow

    FR‑1 Users can rate 1–5 stars (half‑stars allowed?)
    FR‑2 Text review (min 50, max 3 000 chars, profanity filter)
    FR‑3 Optional media attach (≤ 5 images, ≤ 30 sec video)
    FR‑4 Realtime validation + autosave draft
    FR‑5 Confirmation toast + undo (15 s)

    6.2 Discovery

    FR‑6 Global search bar (products, businesses, categories)
    FR‑7 Filters: rating, recency, price range, location, “only with photos”
    FR‑8 Sort: relevance, newest, highest rated, most helpful

    6.3 Accounts & Identity

    FR‑9 OAuth (Google, Apple, email/password)
    FR‑10 Unique display name, optional avatar
    FR‑11 Email verification & 2FA (TOTP/SMS)

    6.4 Trust & Moderation

    FR‑12 ML spam detection (threshold > 0.8 auto‑quarantine)
    FR‑13 Community reporting (4 reasons + optional comment)
    FR‑14 Moderator dashboard (queue, bulk actions, audit log)
    FR‑15 Fraud signals: IP clustering, review velocity, sentiment anomalies

    6.5 Business Tools (Phase 2)

    FR‑16 Claim profile via domain‑based email
    FR‑17 Respond publicly or privately to reviews
    FR‑18 Analytics: rating trend, keyword cloud, comparisons

    6.6 Notifications

    FR‑19 In‑app & email: review published, replies, helpful votes
    FR‑20 Digest options: immediate, daily, weekly


    7. Non‑Functional Requirements

    • Performance: <300 ms p95 search latency, <2 s FCP web
    • Availability: 99.9% uptime SLA
    • Scalability: 10× traffic over 24 months
    • Security: ISO 27001 controls, GDPR & CCPA compliance
    • Accessibility: WCAG 2.1 AA
    • Localization: i18n framework; launch in EN, ES, FR


    8. MVP Scope (90‑day build)

    ✅ Search & browse entities
    ✅ Create / edit / delete reviews (text + star + media)
    ✅ Basic user auth + profiles
    ✅ Helpfulness voting
    ✅ Flagging + manual moderation queue
    ✅ Core analytics events (page view, search, review submit)


    9. Out of Scope (v1)

    ✗ Business subscription tiers
    ✗ In‑house ad network
    ✗ Augmented Reality “try‑on” reviews
    ✗ Blockchain‑based review provenance


    10. Success Metrics Dashboard

    MetricTarget @ M+3Target @ M+6
    MAU25 k100 k
    Reviews submitted50 k250 k
    Avg. session length3 min4.5 min
    % reviews with media20%30%
    Flag rate (per review)≤ 2%≤ 1%
    MTTM (mean time to moderate)≤ 12 h≤ 4 h

    11. Technical Architecture (High‑level)

    graph TD
    A[Web & Mobile Apps] -->|REST/GraphQL| B(API Gateway)
    B --> C[(Auth Service)]
    B --> D[(Review Service)]
    B --> E[(Search & Index)]
    B --> F[(Media Storage S3 + CDN)]
    D --> G[(RDBMS for transactional data)]
    E --> H[(Elasticsearch / OpenSearch)]
    D --> I[(Event Bus - Kafka)]
    I --> J[(Moderation ML Pipeline)]
    

    12. Data Model Snapshot (Reviews)

    FieldTypeNotes
    review_idUUIDPK
    entity_idUUIDFK to Product / Biz
    user_idUUIDFK
    ratingFLOAT1.0 – 5.0
    bodyTEXTmarkdown supported
    media_urlsARRAY(URL)S3 paths
    helpful_countINTdefault 0
    statusENUM(draft,published,flagged,removed)
    created_at/updated_atTIMESTAMP

    13. Milestones & Timeline

    WeekMilestone
    0–2Discovery, wireframes, tech spikes
    3–6Auth + Profiles; DB schema
    7–9Review CRUD, media upload
    10–12Search & filters
    13–14Moderation queue + reporting
    15Closed beta (50 users)
    16Public MVP Launch 🎉

    14. Risks & Mitigations

    RiskProbabilityImpactMitigation
    Review fraud / spamHighHighML + manual moderation, verified purchases (v2)
    Legal takedown requestsMedHighIn‑house counsel, clear TOS
    Scalability spikesLowMedAuto‑scaling, CDN, cache layer

    15. Open Questions

    1. Do we require purchase proof at MVP?
    2. Half‑star ratings or full‑star only?
    3. Mobile‑web first vs. native apps in parallel?
    4. Monetization timeline—ads in MVP?

    16. Appendices

    16.1 Analytics Events

    • search_performed
    • entity_viewed
    • review_submitted
    • review_flagged
    • helpful_clicked

    16.2 Compliance Checklist

    ☑ Privacy Policy
    ☑ Terms of Service
    ☑ Cookie Consent Banner
    ☑ Data Deletion Flow


    17. Future Enhancements (Post‑MVP)

    • Verified purchase badge via receipts/integrations
    • AI summaries of review sentiment
    • Gamified contributor levels & badges
    • Business subscription with enhanced analytics
    • Voice review input + transcription


    End of Document
    Comments & approvals: Please add inline notes or schedule a review meeting by Friday EOD.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.