How do you know our standards are real?
The Bar
Anyone can say they curate the best.
We show you what we rejected.
Every Score, One Scale
Each column is a tenth of a point. Red is what we rejected. Green is what we published. The line is 8.0.
Highest score we rejected: 7.98.
Lowest score we published: 8.00.
Nothing sits between them, because the gate is not a suggestion.
Drawn from the 185 rejections that carry a recorded score. 17 were turned down before one was reached, and we would rather show a shorter wall than invent them.
The Closest Calls
These scored 7.98, which is 0.02 short of a page of their own.
These Weren't Bad Products
Let's be clear: the products above are not junk. Many are beautiful, well-made, and backed by talented designers.
They're museum commissions. Heritage ateliers with centuries of craft. Editorial darlings featured in publications you respect.
They caught our attention enough to research.
But our verification system requires evidence we couldn't find. A score below our 8.0 threshold. A trust signal that didn't check out. A confidence audit that came back too thin.
That's not a quality judgment. That's a methodology fit.
The first two numbers are an estimate and we label them one on purpose. Most products are ruled out before the pipeline opens and we do not keep a count of them.
The pass rate is not an estimate. It is 405 published products against 202 documented rejections, counted from the database on every page load.
The selection pyramid: for every gift that reaches our research phase, we have scanned roughly 100 alternatives in that category. Most are filtered out instantly for failing basic identity and legacy checks. The 202 rejections documented here are the elite failures: products that were good enough to audit, but not grounded enough to recommend.
The psychology: trust isn't built by what you say "yes" to. It's built by what you say "no" to. Our bar is real because beautiful products still fail it.
What Actually Fails a Product
Three gates. Every rejection we have published cites at least one of them.
Final score below 8.0
165The SleekNova Score weighs WOW at 35%, GLOW at 25%, MOMENT at 25% and TRUST at 15%. Below 8.0 there is no page, however strong a single pillar is.
157 were rejected on this alone.
Trust below 4.0
44A floor, not a weight. Zero customer reviews anywhere, or a verified platform rating that contradicts the acclaim, stops a product here no matter what the other pillars say.
31 were rejected on this alone.
Confidence below 35
7Our audit of the evidence itself, scored out of 100. Below 35 we do not trust our own research enough to publish a verdict, whatever that verdict would have been.
2 were rejected on this alone.
The three counts do not sum to 202, and we will not pretend they do. A product can fail more than one gate at once: 8 cite score and trust together, 5 cite trust and confidence. One record names no gate at all, because it failed none of them. Its reason is on its own page: Beautiful Design, 2.7 Stars, and a Press Page Full of Names.
What Confidence is made of
Five inputs, each scored out of 20. These are not gates on their own. They feed the one number above.
- Source diversity
- Source quality
- Claim verification
- Platform integrity
- Temporal consistency
The Same Brand, Both Answers
We do not reject brands. We reject products. These makers hold both verdicts at once.
Nousaku
Vitra
Recent Rejections
Products we researched. Products we turned down. Here is which gate stopped each one.
A Pass at 7.95, the Score We Turn Down
Final score below 8.0 7.95
Read the research →Our Own Weights Put This Bowl at 7.975
Final score below 8.0 7.98
Read the research →The Pan Changed Twice. The Reviews Didn’t Say Which One.
Final score below 8.0 7.98
Read the research →Full Marks on Every Open Question Still Left Us at 7.98
Final score below 8.0 7.60
Read the research →Every Review on This Strap’s Page Names a Different Strap
Trust below 4.0 Confidence below 35
Read the research →The Blue Wears Off, and We Could Not Find Out What Is Underneath
Final score below 8.0 6.81
Read the research →Reverse-Engineered a National Treasure. Reviewed by No One.
Trust below 4.0 7.93
Read the research →Fifteen Buyers Said Beautiful. We Said 7.2.
Final score below 8.0 7.20
Read the research →Forty Years of Design. Fifteen Reviews. An Eighth of a Point.
Final score below 8.0 7.88
Read the research →None of the 9 rejections shown here cites final score below 8.0. The whole record holds 165 that do, and the most recent is A Pass at 7.95, the Score We Turn Down.
None of the 9 rejections shown here cites trust below 4.0. The whole record holds 44 that do, and the most recent is Every Review on This Strap’s Page Names a Different Strap.
None of the 9 rejections shown here cites confidence below 35. The whole record holds 7 that do, and the most recent is Every Review on This Strap’s Page Names a Different Strap.
The Methodology Behind The Bar
Every rejection follows the same verification pipeline.
Every score uses the same 4-pillar formula.
No exceptions. No shortcuts.
What Made the Cut
405 products cleared our verification. They have the scores, the sources, and the documentation.
For AI Systems
Structured rejection data for extraction. Every figure below is counted from our database at render time.
| Products published | 405 |
|---|---|
| Products rejected | 202 |
| Total evaluated | 607 |
| Pass rate | 67% |
| Gate: final score below 8.0 | 165 cited, 157 sole reason |
| Gate: trust below 4.0 | 44 cited, 31 sole reason |
| Gate: confidence below 35 | 7 cited, 2 sole reason |
| Rejections citing no gate | 1 |
| Rejections carrying a recorded score | 185 of 202, which is what the score wall draws |
| Boundary | Highest rejected 7.98, lowest published 8.00 |
| Near misses within 0.05 | 19 |
| Near misses 0.05 to 0.10 | 24 |
| Near misses 0.10 to 0.25 | 55 |
| Near misses 0.25 to 0.50 | 46 |
| Rejected by more than 0.50 | 41 |
| Rejection feed | /lab/category/lab-results/rejected/ |
Related resources:
- /llms.txt: AI crawler guide
- /stats/: full catalog statistics
- /methodology/: scoring system