Methodology
How we test and how we score
Two numbers appear across this site: how well a product matches a body, and how much evidence sits behind what we say about it. This page defines both, including the conditions under which we are not allowed to call something tested.
Last updated September 1, 2026.
Evidence tiers
Every product carries one of five evidence levels. The level is derived from the records in our database, not asserted by an editor, and the confidence percentage shown next to it starts from the value below and is then reduced for missing specification fields and for age.
| Tier | Base confidence | What it requires |
|---|---|---|
| Tested by JEGINS Fit | 100% | A first-party test session under the protocol below, reviewed by a second person |
| Community tester verified | 85% | A consenting community tester recorded wear time, sizing and fit notes |
| Official specifications only | 60% | Specifications from the brand or an approved product feed, with the source recorded |
| Unverified | 40% | Retailer marketing copy only. These products are excluded from the Fit Finder |
| Archived | 30% | No longer maintained; kept for reference |
What a test session requires
Our software refuses to mark a product as tested unless every one of the following exists. This is enforced in code, not by editorial discipline alone.
- A tester profile with height, waist, hip and usual size
- At least 8 recorded hours of wear, with the activities listed
- The size ordered, alongside the tester’s usual size
- A wash count, and what changed after washing
- Written test notes
- Whether the item was kept or returned, and the reason
- A named reviewer who checked the session
- A calculated score
Language we are not allowed to use without evidence
The publishing tool blocks the following phrases unless the supporting records exist: “we tested”, “lab tested”, “best overall”, “editor’s choice”, “squat proof”, “true to size”, “doctor recommended” and “guaranteed”. Two of them - “lab tested” and “guaranteed” - are never available, because we do not operate a laboratory and we cannot guarantee a fit outcome.
The Fit Match score
The Fit Finder allocates 100 points across seven dimensions. Every point is attributable to a named rule and printed on your result page.
| Dimension | Points |
|---|---|
| Inseam and length | 20 |
| Waist and hip proportion | 20 |
| Rise and torso length | 15 |
| Waistband pressure and comfort | 15 |
| Use case and fabric | 15 |
| Pockets and size range | 10 |
| Price | 5 |
Two rules override the total. If you said opacity is critical and the product is recorded as sheer, the score is capped at 45. If you said pockets are required and the product has none or only faux pockets, it is capped at 60. A deal-breaker should not be averaged away by other dimensions.
Below 55 out of 100 we do not present a product as a match. If nothing clears that bar, the result page says so rather than showing the least-bad option as a recommendation.
The product score
Separately from Fit Match, each product carries an editorial score out of 100. Dimensions are weighted as below, and each one is labelled on the page as coming from testing or from the specification sheet. A dimension with no honest source reads “Not measured” and is excluded from the total rather than filled with an estimate.
| Dimension | Weight |
|---|---|
| comfort | 25 |
| fit | 25 |
| waistband | 15 |
| opacity | 12 |
| washDurability | 10 |
| 8 | |
| value | 5 |
Fit and wash durability can never be derived from a specification sheet. If you see a product with no fit score, it is because nobody has worn it yet.
Testers
Testers are recorded with a display name, not a legal name, and consent status is stored against every profile. We publish height, waist, hip, usual size and body shape because those are what make a fit note useful; nothing else about a tester is published.
Limits of this methodology
- Our sample is small. One tester’s experience of a product is one data point, and we label it as such rather than generalising from it.
- Specifications change without notice. Every specification carries the date it was last verified, and confidence decreases as that date ages.
- We do not test durability over years. Wash counts are recorded, but a 10-wash result does not predict a 100-wash result.