How Search Ranking Works
Wesker operates a transparent, merit-based ranking system designed to match hardware creators and engineers with the most reliable local manufacturing workshops. This document outlines the exact parameters, weights, and policies governing search visibility.
1Two-Stage Discovery Architecture
To ensure speed and fairness, our discovery pipeline operates in two distinct stages:
Stage 1: Candidate Retrieval
Narrows down the entire database to workshops that can technically fulfill the buyer's request:
- Machine Category & Material capability match
- Geographic proximity radius (closer workshops prioritised)
- Verified KYC status (unverified machines cannot enter the pool)
- Machine availability (active and unpaused listings only)
Stage 2: Quality Composite Scoring
Orders the retrieved eligible candidate pool using an algorithmic Quality Score (0.0 to 100.0) recalculated daily across six core performance metrics.
2Quality Score Parameters & Factor Weights
The composite quality score is computed deterministically using the following weighted parameters:
1. Verified Buyer Ratings (`avg_rating`)
Scale: 1.0 to 5.0 stars from approved reviews
Measures past buyer satisfaction across verified deliveries. New workshops without reviews start with a neutral prior score of 4.0/5.0 (80%), preventing zero-review penalties.
2. On-Time Delivery Rate (`on_time_delivery_rate`)
Ratio of orders fulfilled within promised machine lead times
Rewards makers who reliably dispatch and deliver fabricated parts on or ahead of schedule. New makers baseline at 100%.
3. Dispute Rate (`dispute_rate`)
Ratio of bookings resulting in escrow payment disputes
Workshops with minimal disputes receive full credit. Elevated dispute rates heavily penalise visibility to protect buyers.
4. Seller Cancellation Rate (`cancellation_rate`)
Ratio of accepted orders cancelled due to seller fault
Penalises workshops that accept manufacturing jobs and subsequently abandon or cancel them.
5. Capacity Headroom (`capacity_headroom`)
Available workshop throughput vs active orders in progress
Computed live at query time. Workshops operating near maximum concurrent capacity receive reduced placement for new bookings to prevent lead time blowouts and fabrication bottlenecks.
6. Portfolio Showcase Engagement (`post_engagement`)
Approved showcase posts and verified buyer likes
Provides a small discovery signal for workshops that actively demonstrate recent fabrication tolerances through portfolio reels. When the social feature flag is inactive, this 5% weight is dynamically reallocated to Rating (27.5%) and On-Time Delivery (22.5%).
390-Day Cold-Start Visibility Policy
To prevent established high-volume makers from permanently monopolising top search results, Wesker provides a transparent Cold-Start Boost for newly registered and verified workshops:
- Duration: Active for the first 90 calendar days from workshop registration.
- Magnitude: Up to +5.0 points added to the composite quality score.
- Smooth Graduation: Decays smoothly across the first 10 completed orders, allowing new makers to build an authentic reputation.
4Anti-Gaming & Policy Exclusion Rules
To maintain marketplace safety and escrow integrity, listings are excluded from organic rankings under the following conditions:
- Repeated Chat Violations: Accounts with $\ge 3$ violations for off-platform payment or contact solicitation are set to
excluded_from_ranking = true(Score: 0.0). - Banned Accounts: Any seller undergoing suspension or regulatory sanctions is completely removed from search candidates.
- Paused Listings: Listings explicitly paused by the seller are suppressed to avoid unfillable inquiries.
5Sponsored Listings & Paid Placements (Rule 5(3)(f) Compliance)
In accordance with Rule 5(3)(f) disclosures of India's Consumer Protection (E-Commerce) Rules, 2020 and ASCI digital media transparency standards:
- Strict Structural Separation: Organic quality scores and ranking positions are never influenced by advertising spend, sponsorship fees, or commission rates.
- Mandatory Visual Badging: Any listing promoted via pay-per-click advertising campaigns is prominently and unmistakably labeled with a "⚡ Sponsored" badge that links directly to this transparency disclosure.
- Generalized Second-Price (GSP) Auction: Sponsored winners are determined by AdRank:AdRank = MaxBid (Paise) × QualityScore [0.10, 1.00]The winning maker is charged the second-price clearing cost:ClearingPrice = min(MaxBid, ceil(AdRank_next / QualityScore_winner) + 1 paisa)
- Measurement Integrity & Quality CTR Window (T-ADS-011): Quality scores incorporate historical click-through rate (CTR) measured strictly over windows of $\ge 1,000$ verified human impressions. New campaigns (<1,000 impressions) enter cold start utilizing the platform category median CTR ($2.5\%$), ensuring zero cold-start disadvantage. Automated crawler, bot, and click-farm impressions are cryptographically filtered and excluded.
- Quality Precondition & Floor: Sponsored placements require verified maker KYC, active machines, and a minimum Quality Score floor ($0.35$). If no ad candidate meets the quality floor, pure organic listings are rendered with zero empty ad boxes.
6Algorithmic Weight Changelog
| Version | Release Date | Key Modifications | Compliance Status |
|---|---|---|---|
| v2.4.0 | Aug 2026 | Integrated 30% live capacity headroom weighting to prevent fabrication bottlenecks | Active |
| v2.3.0 | May 2026 | Added 90-day cold-start boost (+5.0 pts decaying over 10 orders) for new makers | Superseded |
| v2.1.0 | Jan 2026 | Introduced anti-gaming exclusion penalty for off-platform solicitation violations | Superseded |
Last algorithmic weight deployment: August 15, 2026