"Best Match" and "Cassini" are two different things that most eBay sellers conflate. Best Match is the visible ranking mechanism — the default sort order buyers see. Cassini is the machine-learning system that scores listings behind it. Best Match is what changed in 2013 when Cassini launched; Cassini is what actually decides the scores that Best Match then sorts.
A 30-second history
Pre-2008: eBay's default sort was recency (newest listings first). Sellers exploited this via constant re-listing.
2008-2013: "Best Match" launched as the default sort, powered by an internal system called Voyager. Ranking was heavily keyword-based with some behavioral signals layered on.
2013-present: Voyager was replaced by Cassini — a machine-learning ranking model that scores listings on 12+ dimensions. Best Match remained the visible sort label; Cassini became the engine.
The two-layer architecture
Think of it like Amazon's search: A9 ranks (visible sort), whereas machine-learning subsystems score. eBay is the same pattern: Best Match ranks, Cassini scores.
The distinction matters because the two layers respond to different levers:
Cassini score inputs: item quality, seller performance, price competitiveness, engagement, freshness, item specifics coverage, category depth, photo quality. These are your seller-side optimizations.
Best Match ranking rules: query-relevance boosts, personalization overrides, category-specific modifiers, promoted-listing insertions, buyer-specific history. These are eBay's platform-level layer that consumes Cassini scores and produces the visible ranking.
What Best Match adds on top of Cassini
Query-relevance boost: Even a high Cassini score can be overridden if your listing's title/aspects don't match the specific query. A high-scored "Toyota Camry alternator" listing loses to a mid-scored "2013 Toyota Camry LE 2.5L alternator" listing when the query specifies year and trim.
Personalization: A buyer who previously purchased from you gets your listings boosted in their results. A buyer who returned an item from you gets your listings demoted. Cassini scores are the same; Best Match personalizes per viewer.
Category modifiers: Motors categories weight freshness and item specifics more heavily. Fashion categories weight photo count and photo quality more. Best Match applies category-specific weight adjustments over Cassini's base score.
Category-specific ranking modifiers
Same Cassini score, different Best Match rank depending on category. Example: a listing with a Cassini score of 82/100:
- In Motors P&A: Best Match rank ~15 (freshness weight boosts stale listings less)
- In Consumer Electronics: Best Match rank ~8 (photo weight helps if you have 6+ photos)
- In Fashion: Best Match rank ~4 (category weights heavily favor engagement signals)
Personalization layer
Two buyers, same query, same instant, see different result orderings. The base Cassini scores are identical, but Best Match's personalization layer applies buyer-specific adjustments: recent purchase history, watch-list overlap with the current query, saved searches, and even geographic location (some sellers rank higher for buyers in their region due to shipping-speed personalization).
Practical implication: your own view of your listing's search rank is unreliable. You always see your own listing boosted (personalization) or demoted (if you've ever ended one of your listings — eBay detects this as buyer disinterest). Test rank from an incognito browser or ask a friend.
Why "sort by price" and "sort by best match" produce different pools
Buyers who explicitly sort by "price low to high" bypass Cassini scoring entirely — they see the pool ordered by BIN alone. This produces a completely different set of results from Best Match, and often surfaces listings that would never rank in Best Match (because they have low Cassini scores).
Consequence: sellers with strong Cassini scores lose visibility to price-sort buyers unless they're also competitively priced. Strong seller reputation doesn't beat a lower BIN in price-sort. This is one reason "premium seller" strategies fail against commodity competition.
Practical implications for sellers
Optimize for Cassini first (the base score). Understand that Best Match adds category modifiers and personalization on top, so results you see will vary. Measure rank using bulk tools that check multiple queries from clean sessions, not your own browser view.
Profitio surfaces per-listing Cassini score so you know which layer to work on. Try free for 14 days.