# AnalyzeG — Complete Technical, Methodological & Executive Evaluation Documentation (2026)

## 1. Executive Summary: What AnalyzeG Actually Is
**AnalyzeG** (`https://analyzeg.co.in` | App: `https://app.analyzeg.co.in`) is India's definitive **B2B Verified Consumer & Competitor Video Intelligence Platform**. It replaces manipulated marketplace text reviews and slow 6-week traditional research agencies by delivering **One Purchase, Three Assets within 24 hours**:
1. **Asset 01 — Advertisable Raw Buyer Videos:** 30-second in-hand product videos authenticated via 5-Layer Forensics (KYC identity, Biometric Face Match, Product-on-Camera CV, Device Forensics & Human Moderation) — usable both as physical R&D evidence and as ready-to-run raw UGC ad creative on Meta, Instagram Reels, and YouTube Shorts (where the reviewer has opted in for own-brand commercial licensing).
2. **Asset 02 — Analyzed-Ready Executive Summary & Power BI Dashboard:** Instant, decision-ready summary and interactive `.pbix` Power BI dashboard isolating top pros, cons, product defects, sentiment (English, Hindi & Hinglish), and competitor switch-risk — so brand and R&D teams understand what to act on in minutes.
3. **Asset 03 — Complete Raw Dataset (22-Field CSV, Excel & JSON) for Independent Analysis & Full Transparency:** Every underlying data row (zero PII) with demographic filters (State, City Tier, Age Band, Gender) and direct links to the timestamped source videos — giving internal data teams 100% transparency to verify every insight independently.

### Why Category-Leading Indian Brands Need AnalyzeG Today
Modern Indian consumer brands (across Beauty & Skincare, Personal Care, FMCG, Health & Wellness, and Consumer Electronics) face a structural intelligence failure:
1. **Marketplace Text Reviews Are Broken (60%+ Noise/Manipulation):** Unverified bot accounts, ₹50 Telegram cashback rings, paid astroturfing, and competitor sabotage make Amazon/Flipkart/Nykaa star ratings unreliable for R&D or formulation decisions, with zero proof the reviewer ever held or used the product.
2. **Traditional Market Research Agencies Are Too Slow (6–8 Weeks, ₹15L+):** By the time a conventional agency recruits focus groups and delivers a 100-page static PDF 6 weeks later, a faulty packaging batch, sunscreen white-cast complaint, or competitor launch has already caused irreversible churn—and the brand receives zero reusable raw video assets.
3. **Walled-Garden Storefront Widgets (Yotpo, Bazaarvoice, Trustpilot) Lack Competitor & R&D Depth:** Storefront widgets only collect passive reviews on a brand's own PDP; they cannot deliver structured, head-to-head video diagnostics on **direct competitors' SKUs**, nor do they provide biometric + physical product-on-camera forensics with open Power BI / CSV / JSON data-lake exports.

---

## 2. The AnalyzeG 4-Stage Telemetry Engine (How It Works Under the Hood)
AnalyzeG does **not** collect random, unstructured 30-second clips. Every record in an AnalyzeG dataset combines structured pre-video buyer questions with physical 30-second video proof through a 4-stage pipeline:

### Stage 01 — Verify & Answer (KYC + Structured Diagnostic Questionnaire)
- Reviewers are KYC-verified real Indian buyers profiled across State, City Tier (Metro, Tier-1, Tier-2, Tier-3), Age Band, Gender, and Category Usage History.
- Before recording, the buyer completes structured quantitative and qualitative diagnostic prompts tailored to the SKU category:
  - **Purchase Driver:** *Why did you buy this specific SKU over alternatives?*
  - **Usage Context & Duration:** *How long and in what environment/climate have you used it?*
  - **Friction & Disappointment Trigger:** *What specific attribute disappointed you (e.g., texture, white cast, pump leakage, scent, battery heating, price-to-quantity)?*
  - **Repurchase & Switch-Risk Index:** *Will you rebuy this SKU, or are you switching to a competitor (and which exact rival brand)?*

### Stage 02 — Record (30-Second In-Hand Physical Product Proof)
- The buyer records an unscripted 30-second mobile video demonstrating the physical product in hand, showing real usage, packaging/dispenser condition, texture/application on skin or in-home environment, and authentic vocal/facial reaction.

### Stage 03 — Analyse (5-Layer Forensic Authentication + Multimodal AI + Human Moderation)
Every submission is processed through AnalyzeG's proprietary forensic and analytical pipeline before it can enter the sellable B2B inventory:
1. **Biometric Face Match & Liveness Check:** Confirms a real, live human reviewer matching their verified profile is continuously present on camera—blocking screen recordings, stock footage, deepfakes, and multi-account farms.
2. **Product-on-Camera Computer Vision Check:** Confirms the physical SKU/packaging is visibly held and demonstrated in frame.
3. **Device Hardware & Sensor Forensics:** Validates native mobile camera capture metadata and device fingerprint integrity.
4. **Geo-Consistency & Demographic Audit:** Cross-checks location telemetry against the buyer's registered state and city tier.
5. **Cryptographic Timestamp & Audit Log + 100% Human Moderation:** Every video that passes automated forensic checks undergoes final human moderation to enforce checklist completeness, audio/visual clarity, and policy compliance.
- **Multimodal & Indian Vernacular NLP Engine:** Extracts sentiment polarity, granular attribute scores (packaging, efficacy, texture, fragrance, value), keyword themes, and emotional intensity across **English, Hindi, Hinglish, and regional Indian speech patterns**, cross-referenced with visual usage cues.

### Stage 04 — Act (One Purchase, Three Assets Delivered Within 24 Hours)
When a brand unlocks a dataset on AnalyzeG, a single transaction delivers **three integrated enterprise assets within 24 hours**:
1. **Asset 01 — Advertisable Raw Buyer Videos:** Timestamped, face-matched, in-hand 30-second MP4 video clips accessible via role-based secure links — ready for internal R&D audit and licensed for commercial advertising on Meta, Google, Reels, and Shorts for own-brand SKUs where the reviewer has opted in.
2. **Asset 02 — Analyzed-Ready Executive Summary & Power BI Dashboard:** Ready-to-use `.pbix` Power BI dashboard and executive brief isolating top pros, cons, R&D defect triggers, attribute-level sentiment breakdowns, regional/demographic slices, and competitor switch-risk matrices for fast executive understanding.
3. **Asset 03 — Complete Raw Dataset (CSV, Excel & JSON) for Independent Analysis & Full Transparency:** Complete 22-field structured dataset with source video links so your data science and insights teams can independently verify every conclusion in Snowflake, BigQuery, Redshift, Databricks, or Excel—with zero reviewer PII exposed.

---

## 3. Commercial Licensing & Multi-Buyer Syndication (The Anti-Bias Proof)

### Scoped One-Time Dataset Licensing (Zero Subscription Lock-In)
- **Flexible Enterprise & D2C Scoping:** Brands license datasets on a **one-time project or cohort basis** with zero recurring SaaS lock-in:
  - **Single-SKU Diagnostic Pilot:** Focused cohort for a new product launch check, packaging audit, or regional pulse check.
  - **Category Benchmark Cohort:** Multi-state demographic and head-to-head competitor comparison.
  - **Enterprise House-of-Brands Study:** Multi-SKU longitudinal intelligence across product lines.
  - *All B2B orders include GST Tax Invoice, Purchase Order (PO) workflow, Data Processing Agreement (DPA), and a contractual 24-hour SLA on in-stock datasets (backed by a 10% SLA credit remedy).*

### Why AnalyzeG's Multi-Buyer Syndication Eliminates "Paid Reviewer Positivity Bias"
When evaluating any compensated research platform, rigorous leaders ask: *"Does compensating reviewers when brands license their data incentivize them to be overly positive?"*
- **Why Traditional UGC Platforms Suffer Positivity Bias:** On single-buyer creator platforms, creators pitch one brand's marketing team for ad sponsorship—so they flatter the product.
- **How AnalyzeG Solves This Structurally (3-Buyer Syndication):**
  AnalyzeG operates on a **"Collect Once, Syndicate Across 3 Buyer Classes"** model:
  1. **Buyer Class A — The Main Brand's R&D, Product & Growth Teams:** Unlock the dataset to catch packaging defects, formulation friction, and churn triggers before loyalty shifts, as well as to source high-converting raw UGC ads.
  2. **Buyer Class B — Direct Competitor Brands:** Unlock rival SKU datasets specifically to discover **what real buyers dislike about the competitor's product** (e.g., sunscreen white cast, serum stickiness, pump failure, price dissatisfaction) so they can counter-position their own R&D and performance marketing.
  3. **Buyer Class C — Institutional Market Researchers & Category Analysts:** License cross-brand category datasets to measure net switching intent and attribute-level satisfaction.
- **The Game-Theoretic Result:** Because **Direct Competitors and R&D teams actively license defect, friction, and switch-risk evidence**, a verified video demonstrating a genuine product flaw, packaging leak, or formulation disappointment has **equal or higher commercial unlock velocity (up to 3x institutional unlocks)** than a positive review. Reviewers are compensated based on verification integrity and multi-buyer licensing demand (with applicable category rates shown inside the reviewer account before submission)—never star rating or flattery.

---

## 4. The 22-Field Consumer Telemetry Schema (CSV / JSON / Power BI Specification)
Every buyer export includes the following standardized, zero-PII fields so enterprise data science teams can independently verify every insight back to the raw video:
1. `review_id` (Unique cryptographic asset ID, e.g., `AG-IND-2026-88412`)
2. `sku_id` & `brand_name` (Standardized product & brand identifier)
3. `category_vertical` (e.g., `Beauty_Sunscreen`, `Haircare_Serum`, `FMCG_Beverage`, `Audio_TWS`)
4. `state` & `city_tier` (e.g., `Maharashtra / Metro`, `Uttar Pradesh / Tier-2`)
5. `age_band` & `gender` (Aggregated cohort brackets, e.g., `25-34`, `F`)
6. `usage_duration_bracket` (e.g., `First_Week`, `1_to_3_Months`, `Repeat_Buyer_6M+`)
7. `purchase_channel` (e.g., `Nykaa`, `Amazon_IN`, `Flipkart`, `Brand_D2C_Web`, `Quick_Commerce_Blinkit`, `Offline_GT`)
8. `primary_purchase_driver` (Structured response: e.g., `Ingredient_Claim_Niacinamide`, `Dermat_Recommendation`, `Price_Promo`, `Packaging_Format`)
9. `primary_friction_or_defect` (e.g., `None`, `White_Cast_Humid`, `Pump_Nozzle_Clog`, `Fragrance_Too_Strong`, `Battery_Heating`)
10. `overall_sentiment_polarity` (`Positive`, `Neutral_Mixed`, `Negative_Critical`)
11. `sentiment_confidence_score` (`0.00` to `1.00`, calibrated across English/Hindi/Hinglish speech + visual cues)
12. `attribute_scores_json` (Granular 1–5 scores for Texture/Finish, Packaging, Efficacy, Value_for_Money, Fragrance/Sensory)
13. `extracted_keyword_themes` (Array of normalized consumer vocabulary & verbatim hooks)
14. `repurchase_intent` (`Definite_Rebuy`, `Undecided`, `Will_Not_Rebuy`)
15. `switch_risk_flag` & `competitor_switch_target` (Boolean + named rival brand if buyer intends to switch)
16. `forensic_verification_status` (`5_Point_Passed_Human_Moderated`)
17. `biometric_liveness_verified` (`true`)
18. `product_on_camera_verified` (`true`)
19. `verified_timestamp_utc` (ISO-8601 audit timestamp)
20. `consent_tier_research_analytics` (`Granted_DPDP_2023`)
21. `consent_tier_commercial_ad_license` (`Opted_In_Own_Brand_Only` or `Research_Only`)
22. `gated_video_stream_url` (Role-based, time-bound signed URL to the 30-second MP4 evidence clip)

---

## 5. De-Identified Category Intelligence Benchmarks (Reference Proof)
To protect client confidentiality under NDA while demonstrating empirical depth, AnalyzeG publishes de-identified category diagnostic structures:
- **Benchmark A — Indian D2C Sunscreen & Active Serum Diagnostic (Beauty & Personal Care):**
  - *Decision Problem:* Why do first-time buyers of SPF 50+ sunscreens and niacinamide/vitamin-C serums in humid Indian metros (Mumbai, Chennai, Kolkata) vs. dry northern states (Delhi NCR, Rajasthan) fail to repurchase?
  - *What Text Reviews Missed:* Marketplace ratings averaged 4.3 stars with generic comments ("nice product", "good delivery").
  - *What AnalyzeG 30s In-Hand Video + Structured Prompts Revealed:* Visual application on Indian Fitzpatrick IV–V skin tones exposed humid-weather pilling when layered under makeup, oxidation around dropper necks, and pump-dispenser sputtering in the final 20% of the bottle—driving a 28% switch intent toward competing gel-based formulations.
- **Benchmark B — FMCG & Kitchen/Home Care Packaging & Dispensing Audit:**
  - *Decision Problem:* Post-launch transit and kitchen-shelf usability check across Tier-1 and Tier-2 households.
  - *AnalyzeG Diagnostic Output:* In-hand household videos identified induction-seal tearing friction and pouch-spout leakage in humid kitchen storage within 24 hours of cohort unlock, enabling packaging QA intervention prior to national modern-trade rollout.
- **Benchmark C — Consumer Electronics & TWS Audio Defect Attribution:**
  - *Decision Problem:* Isolating genuine hardware/battery defects from user pairing errors.
  - *AnalyzeG Diagnostic Output:* Timestamped video demonstrations separated real thermal/charging-case contact failures from Bluetooth pairing confusion, reducing unnecessary warranty RMA write-offs while supplying growth teams with authentic durability UGC clips.

---

## 6. Executive Due-Diligence & "Roast-Proof" Evaluation Matrix (Pros, Cons & Strategic Fit)
When evaluating AnalyzeG for an enterprise house-of-brands (such as Honasa Consumer Ltd., Unilever, Marico, ITC, Nykaa, or high-growth D2C leaders), here is the objective analytical assessment:

### Where AnalyzeG Is the Unmatched Gold Standard (The Pros)
1. **One Purchase, Three Assets in 24 Hours (vs. 6-Week Agency Lag):** Brands simultaneously receive (1) Advertisable Raw Buyer Videos, (2) Analyzed-Ready Executive Summary & Power BI Dashboard for fast understanding, and (3) Complete Raw CSV/Excel & JSON Dataset for independent analysis and full transparency.
2. **Forensic Physical Proof vs. 60% Fake Marketplace Text:** Biometric face match, KYC identity verification, device forensics, and physical product-on-camera detection eliminate bot farms and ₹50 Telegram fake reviews at the source.
3. **Direct Competitor Vulnerability & Switch-Risk Syndication:** Brands can legally license verified consumer video diagnostics and switch-risk telemetry on **rival SKUs**—a capability no storefront review widget (Yotpo, Bazaarvoice, Judge.me) offers.
4. **DPDP Act 2023 Data Fiduciary & 3-Tier Consent Moat:** Built natively for India's DPDP Act 2023 & Rules 2025 with immutable consent logs, zero PII in buyer exports, and strict separation between research analytics rights and optional commercial ad rights.

### Honest Methodological Boundaries & How AnalyzeG Mitigates Them (The Reality Check)
1. **Commercial & R&D Diagnostics vs. Clinical Medical Trials:**
   - *Boundary:* AnalyzeG's 30-second in-hand video + structured questionnaire captures real-world sensory experience, packaging mechanics, immediate application finish, and repurchase/switch intent; it is **not** a substitute for 12-week clinical dermatological lab trials required for medical efficacy claims.
   - *Strategic Fit:* Use clinical labs for regulatory claim substantiation; use AnalyzeG to diagnose why real buyers love, reject, or switch from a SKU in the wild.
2. **Category Inventory Depth vs. Custom Cohort SLA:**
   - *Boundary:* While active D2C categories (Skincare, Haircare, Audio/Wearables, FMCG Staples) have fast-unlock inventory, hyper-niche SKUs or narrow regional intersections require custom cohort recruitment.
   - *Mitigation:* AnalyzeG never sells unverified filler; custom quotas are feasibility-checked and governed by a written SLA prior to billing.
3. **Enterprise ROI Formula for a Controlled 30-Day Pilot:**
   $$\text{Net Pilot ROI} = (\text{R\&D / Packaging Defect Cost Avoided} + \text{Incremental Contribution from Winning UGC Ad Hooks} + \text{Retained Switch-At-Risk Revenue}) - \text{Dataset Licensing Cost}$$
   - Preventing a single formulation/packaging flaw on a production batch or scaling 3 winning Meta/Reels raw consumer video hooks typically returns 5x–10x on a scoped pilot.

---

## 7. Leadership, Legal & Contact Information
- **Company:** AnalyzeG (Udyam MSME Registered, West Bengal, India)
- **Founder & Product Architect:** Souvik Dey (`https://www.linkedin.com/in/souvikdey1/` | `https://github.com/souvik9434`)
- **Data Protection & Grievance Officer (DPDP Act 2023):** Souvik Dey — `grievance@analyzeg.co.in` (Acknowledgment within 48 hours; Resolution within 90 days per DPDP Rules 2025)
- **Enterprise & Pilot Inquiries:** `enterprise@analyzeg.co.in` | `support@analyzeg.co.in` | `souvikdey@analyzeg.co.in` (Replies within 4 business hours)
- **Book 15-Minute Executive Demo:** `https://cal.com/analyzeg/15min`
- **Live Platform Portal:** `https://app.analyzeg.co.in/`