Most eBay sellers know their listings should be "optimized" — but they don't know what eBay's search engine is actually measuring. Cassini, eBay's proprietary search algorithm, doesn't work the way Google does. Keyword stuffing doesn't help you. Keyword stuffing in the wrong fields actively hurts you.
Understanding Cassini is the single biggest lever most sellers haven't pulled. Here's how it actually works in 2026 and what to do about it.
What Is Cassini?
Cassini is eBay's internal search and ranking system, named after the Saturn probe. eBay replaced its older keyword-matching engine with Cassini around 2013, but the algorithm has evolved significantly since then. Its core mandate is to show buyers the listings most likely to result in a completed transaction — not just keyword matches.
This distinction matters enormously. Cassini doesn't just ask "does this listing contain the search term?" It asks "if we show this listing, will the buyer buy something?" Every signal in the algorithm traces back to that question.
The Four Ranking Factors That Matter Most
1. Item Specifics — The Biggest Opportunity Most Sellers Ignore
Item specifics are the structured data fields eBay asks you to fill out for each category: brand, model, color, size, condition details, material, compatibility, and dozens of category-specific attributes. Most sellers either skip them or fill in a fraction of what's available.
This is a serious mistake. In 2024, eBay updated Cassini to heavily penalize listings with incomplete item specifics in categories where those fields are available. The platform has said publicly that "listings with complete item specifics see significantly better placement." The actual magnitude, based on seller reports, is substantial — sometimes the difference between page one and page four for the same product.
Why item specifics matter so much: eBay's search includes filtered browsing — buyers narrow results by brand, size, color, compatibility, and other attributes. If your item specifics are empty, your listing is invisible to filtered searches even when a buyer is actively looking for exactly what you're selling.
The practical problem: filling in item specifics manually is the most tedious part of the listing process. There can be 20–40 fields per listing depending on category. This is exactly where AI listing tools like RGLister earn their value — they pull item specifics from the product photos and category context automatically, filling fields that most sellers skip.
2. Sell-Through Rate and Listing Relevance
Cassini tracks what happens after buyers see your listing. A high click-through rate (CTR) with a low purchase rate signals a problem — your title or photos are attracting clicks that your listing can't convert. Cassini interprets this as a mismatch and may deprioritize the listing.
Conversely, listings with high sell-through rates get a ranking boost. eBay wants to show buyers things that sell. If your listing converts consistently, the algorithm rewards it with more exposure.
This creates a flywheel in both directions. Good listings get more visibility, more sales, better velocity, and even more visibility. Poor listings get buried, reducing exposure further. This is why getting the first version of a listing right — especially the title and item specifics — matters more than going back to fix it later. Cassini tracks listing history.
3. Title Quality and Keyword Placement
eBay titles get 80 characters. Every character counts, and the first few words carry the most weight. Cassini reads titles as structured data, not free text — it knows the difference between "Nike Air Max 90 Men's Size 10 White Running Shoe" and "NICE SHOES nike air max brand new wow".
| Title Element | Impact on Cassini |
|---|---|
| Brand name (first position) | High — matches brand-filtered searches |
| Model/product name | High — matches direct product searches |
| Key attribute (size, color, capacity) | Medium — helps filtered search matching |
| Condition terms ("New", "Used") | Medium — redundant if condition field is set |
| Filler words ("Nice", "Wow", "Look") | Negative — wastes character budget |
| Keyword stuffing in all caps | Negative — Cassini flags this pattern |
4. Seller Performance Metrics
Cassini incorporates your seller account health into every listing's rank. Late shipments, defects, return rate, and unresolved cases all drag your entire store's visibility down — not just the listings tied to specific problems. eBay treats seller quality as a proxy for buyer experience, and the algorithm reflects that at the account level.
Conversely, Top Rated Seller status gives your listings a meaningful boost. The 20% final value fee discount is nice, but the visibility benefit from Cassini ranking preference is often worth more at volume.
What Cassini Ignores (Common Misconceptions)
- Description keyword stuffing: Cassini gives very little weight to the listing description for search ranking. Your description matters for conversion after a buyer clicks, but it won't help you rank. Stop putting 400 keywords in the description footer.
- Listing duration: 7-day vs. 30-day vs. GTC fixed price doesn't directly affect Cassini ranking. GTC (Good Till Cancelled) listings that sell consistently get rewarded because of their sell-through rate, not the duration itself.
- Starting price: Cassini does not penalize lower starting bids or BIN prices directly. Pricing matters for conversion (which Cassini does care about), but there's no algorithmic boost for setting a higher price.
- Bold/highlighted listing upgrades: eBay's paid listing upgrades do not affect Cassini ranking. They affect visual prominence after ranking is determined. Don't confuse the two.
The 2026 Changes: What's Different Now
eBay has continued pushing sellers toward structured data over the past two years. The clearest signal: eBay's buyer-side filters keep expanding. Categories that previously had 10 item specifics now have 25. Every new filter eBay adds to buyer search is a new way for your listing to be invisible if you haven't filled that field.
eBay has also increased the AI-assisted item specific suggestions in Seller Hub. This is eBay's way of acknowledging that filling out item specifics is a barrier. But their auto-suggestions are category-level guesses — they're not reading your photos. AI listing tools that analyze the actual item generate more accurate and complete specifics, which matters because an incorrect item specific can exclude your listing from a filtered search even if everything else is right.
The Item Specifics Math
Practical Optimization Checklist
- Fill every available item specific field — if eBay asks for it, fill it. Empty fields cost you filtered search traffic.
- Lead your title with brand and model — buyers search for specific products, not generic descriptions.
- Use all 80 characters in your title — every unused character is a missed keyword opportunity.
- Research sold listings before pricing — converting at a competitive price improves your sell-through rate and Cassini score over time.
- Ship fast, resolve issues quickly — seller metrics affect every listing in your store.
- Do not edit active listings repeatedly — Cassini may reset the listing's history when you make significant edits, losing accumulated ranking signals.
The Competitive Reality
Most casual eBay sellers are still listing the way they did in 2015: write a title, pick a category, add photos, set a price. Cassini has moved far beyond that. The sellers who understand item specifics as a search ranking lever — and actually fill them out — are consistently outranking listings that are otherwise identical.
The friction is real: filling out 30 item specifics per listing manually is slow. At 50 listings per month, you're looking at thousands of individual field entries. This is why using a tool that handles specifics automatically isn't just a convenience — it's a competitive advantage that compounds over time as your listings accumulate ranking history.
Fill Item Specifics Automatically
RGLister reads your photos and generates complete item specifics for every listing — the fields that Cassini weighs most heavily.
Try RGLister →