QC Photos Are Not Enough: How We Vet Products Before Adding Them
Most spreadsheet curators copy and paste links. Here is the actual five-stage quality pipeline that every product passes through before it appears on SuperBuy Spreadsheet.
Stage 1: Link Health and Seller Verification
The first filter is binary: does the Weidian link load, and does the seller have a transaction history? We run automated headless browser checks every 48 hours. If a listing returns 404, times out, or redirects to a different product, it is flagged for removal. Sellers with fewer than 50 completed transactions are marked as high-risk and require manual review before any of their items are approved.
We also verify that the weidian_id in our database matches the ID visible on the live page. Mismatches usually indicate a seller who swaps product pages to boost ranking—a common black-hat tactic. When we detect an ID mismatch, the entire seller is temporarily blacklisted until they provide an explanation.
Stage 2: Price Plausibility Check
Every category has an accepted price corridor based on batch quality. A Jordan 1 at 120 CNY is either a budget batch or a scam. A Jordan 1 at 800 CNY is either premium or overpriced. Our system flags items that sit outside the corridor and routes them to human reviewers who check comparable listings.
Price spikes trigger a special review. If a seller raises a price by 30% overnight, we check whether the batch improved or whether they are capitalizing on a Reddit hype wave. Hype-inflated prices are rejected regardless of demand. Our mission is sustainable value, not trend-chasing.
Stage 3: Community Signal Aggregation
We monitor three signal sources: Reddit mention frequency, Discord QC channel submissions, and our own user feedback form. An item needs at least two independent positive signals from different platforms to advance. A single glowing TikTok review does not count because paid influencer content is rampant in the replica space.
Negative signals are weighted more heavily than positive ones. A single confirmed bait-and-switch report removes the item immediately, even if ten other buyers had good experiences. We would rather exclude a borderline item than include a risky one. This conservative bias keeps our average return rate under 4%, compared to 12-15% on unfiltered Weidian search.
Stage 4: Batch Code Cross-Reference
When a seller claims LJR Batch, we cross-reference the factory code against known production catalogs. Not all sellers label accurately—some slap "LJR" on mid-tier shoes to justify a higher price. Our verification uses serial number patterns, material descriptions, and side-by-side photo comparison with authenticated reference pairs.
Items that pass batch verification receive a confidence score: A for fully verified, B for consistent but unconfirmed, and C for claimed but unverifiable. We only publish A and B items. C items stay in our internal research queue until additional evidence surfaces.
Stage 5: Final Editorial Review
Before publication, a human curator reviews the compiled dossier: link health, price position, community signals, batch confidence, and photo consistency. The curator writes the seo_title you see on our category pages, ensuring it contains the correct model name, colorway, and batch reference without clickbait language.
Only then does the product enter the public database. Once live, it enters the automated monitoring loop: link checks every 48 hours, price scans weekly, and community signal refreshes daily. A product can be live for months or removed within hours if new negative signals arrive. This living curation is the difference between a spreadsheet and a static list.