How brand similarity is evaluated
Our method separates source-backed brand facts from editorial similarity judgments and does not turn uncertainty into a fake score.
1. Brand facts start with first-party evidence
Official brand, company and product pages are preferred for claims about positioning, product focus and how a brand describes its own offer. A checked date travels with each source record, and old evidence stops supporting public brand pages until it is reviewed again.
2. Similarity is a relationship, not a universal number
Two brands can be similar in one way and different in another. We therefore store price proximity, aesthetic overlap, product/category overlap, audience overlap and geographic overlap separately.
3. Unknown is an acceptable answer
If a dimension has not been established confidently, it remains “Not established.” Uncertainty stays visible instead of being converted into synthetic precision.
4. Difference-first comparisons
A useful alternative relationship must explain both shared ground and the trade-off. A standalone comparison has a higher bar: it also needs a separate, current pair-specific decision brief covering style, price position, product strengths, audience, market access and scenarios where each brand can make more sense. Not every relationship receives a dedicated comparison page.
5. Coverage has minimum evidence thresholds
A brand profile needs current verified sourcing. An alternatives guide needs several substantial reviewed relationships. Category and style collections need enough currently supported brands to be useful rather than repetitive.
6. Focused product-intent pair briefs require two-sided current evidence
A mission-specific pair brief can appear only when the alternative already passes the focused guide’s direct product-evidence gate and also has a separate current first-party source that adds decision context for that mission. If the extra source becomes stale, that brief disappears without keeping the parent route alive. The brief combines those sources with the existing reviewed relationship trade-offs; it does not create an overall winner or make SKU-level quality claims.
7. Price bands are brand-level positioning, not live SKU prices
Value, mid, upper-mid, premium and luxury are reviewed brand-level positioning bands used for discovery. They do not claim that every item from one brand is cheaper or more expensive than every item from another. Product prices overlap and change, so focused price pages state this limitation explicitly.
8. Market availability has its own provenance and freshness window
Market access is stored separately from broad geography tags. A first-party shipping page, country selector or local official online store can establish online access for a specific market. Local ecommerce and international shipping are not treated as identical, and item-level restrictions may still apply. Verified market evidence ages out after the review window; stale evidence stops supporting public market claims until it is checked again. Market-specific Brands Like pages are published only where enough reviewed alternatives also have current market evidence.
9. Popularity does not change the evidence standard
Search interest can help us choose what to research next, but it never substitutes for evidence and never changes how a brand relationship is evaluated.