AI audit vs rubric — DropCommerce
An independent LLM scored DropCommerce against the same published rubric. The deterministic rubric result is our canonical score. The LLM's result is shown here as a sanity check — never mixed into the scoring formula.
| Dimension | Rubric | LLM | Δ (LLM − Rubric) |
|---|---|---|---|
| Pricing transparency | 84 | 20 | -64 |
| Business transparency | 40 | 25 | -15 |
| Shipping clarity | 92 | 85 | -7 |
| Public reviews | 86 | 0 | -86 |
| Product range | 62 | 30 | -32 |
| Access & onboarding | 85 | 40 | -45 |
| Support track record | 82 | 0 | -82 |
| Store integrations | 72 | 40 | -32 |
| Overall | 75 | 28 | -47 |
What this means: Large disagreement — investigate. The LLM read the published signals very differently from the deterministic rules.
Median per-dimension |Δ| = 38.5.
pricingTransparency: Pricing is opaque; contact-sales. businessTransparency: Private company with about page. reviewScore: < stars or no reviews. productRange: Curated small set. access: Paid only with demo. support: Consistently poor support feedback. integration: Manual / API only.This is the LLM's own explanation, not editorial commentary from SupplierSpy. The LLM result is a sanity check on the rubric — never mixed into the scoring formula.