Method v0.3 / evidence beta

An estimate you can challenge.

We compare documented capabilities, not perceived prestige. Feature matching does not establish equivalent sound quality, ANC performance, durability, comfort, support or ecosystem value.

1. Identify the exact model

Different generations and bundles stay separate. URLs and names resolve to a source-checked canonical product. Unrecognized products enter a review queue, never an invented analysis.

2. Compare sourced features

Noise-control capabilities carry 30%, advertised ANC battery 25%, weight 15%, connectivity 20%, and extra capabilities 10%. Each group's weight is divided equally among its fields. Boolean features match exactly. Numeric battery requirements reward meeting or exceeding the target; weight rewards meeting or being below it. This is directional requirement matching, not a universal similarity measure.

Unverified, expired or missing specifications remain unknown. Their weights stay in the denominator and reduce both evidence coverage and the documented match. More microphones and more metal are not scored as better performance.

3. Gate the result

A qualified comparable needs at least 85% feature match and 80% weighted coverage. A similar alternative needs 75% and 80% respectively. At least five qualified, priced products across three brands are needed for an estimate. The original product and duplicate offers never count as additional comparables.

4. Use comparable prices

Only new, in-stock US offers in USD, checked within 48 hours, are eligible. We compare item prices before shipping and tax. Each product contributes its lowest eligible recorded offer; this is not a claim to find the lowest price on the entire market. The top ten feature matches feed a similarity-weighted median. At an exact half-weight tie, the adjacent prices are averaged.

5. Communicate uncertainty

The evidence rubric weights comparable count (30), coverage (30), distinct source domains (20), and actual price recency (20). Domains are not necessarily independent organizations. The score is not a calibrated probability, and the beta cannot claim HIGH confidence. When a mandatory gate fails, the numerical premium is withheld.

Known limitations

Manufacturer battery claims use different test conditions. Spatial audio implementations and device compatibility differ. Weight does not capture fit. A small catalog can miss important competitors. The method remains provisional until a separately reviewed benchmark validates its selections.

Commercial independence

Retailer commissions never enter features, qualification or market value. Current links are non-affiliate. NoBranding does not sell the products or guarantee availability.