Category pages worked when online stores resembled printed brochures. "Running Shoes" might hold fifty products, "Outdoor Jackets" thirty, and the homepage eight featured items. A merchandiser could review the catalog in an afternoon and place products sensibly.
Scale changes that. With 200,000 products, 400,000, or more, a category page is not a considered view of inventory. It is a sample. Whether sorted by bestsellers, recency, or margin, most products stay off-screen and unseen by most shoppers.
The Math Behind What Shoppers See
Take a mid-size outdoor and sporting goods retailer with 180,000 active SKUs across twelve top-level categories. Each category load shows 40 to 80 products. Pagination exists, but shoppers seldom pass page two. In a typical session, about 94% of the catalog never appears on a category page.
That 94% is not merely clearance or niche stock. It includes products that fit particular shoppers. Someone searching for "lightweight trail running shoes for narrow feet" might find the right pair deep in the catalog, if it were surfaced. Category pages show the same products to everyone, regardless of intent.
Retailers described the same pattern while we discussed catalog coverage before our early-access program. In most cases, their storefront showed every visitor a small, repetitive inventory slice, largely unrelated to what that visitor came to find.
What Category Pages Were Built to Do
Category architecture assumes shoppers browse by product type. Someone seeking a sofa visits "Sofas"; someone seeking a monitor visits "Monitors". With a small catalog, that is workable because most of a category can be displayed. Merchandisers can serve most visitors without individual-level logic.
Scale breaks that assumption in two places. First, categories become too large to display fully. A "Women's Clothing" category with 40,000 SKUs cannot show everything, so the visible selection feels arbitrary. Second, shoppers often lack a clear category. Their intent crosses categories, while category pages cannot combine those results even when the need does.
Pagination may address the first issue in theory, but conversion rates consistently fall after page two. Products on page twelve may as well be invisible.
Ranking Challenge
Large catalog retailers often answer by adding merchandising work. They create subcategories, editorial collections, "shop the look" pages, and improved search. These measures help at the edges, but the core issue remains: one ranked list for everyone misses most individual purchase intent.
The issue is not which sorter powers a category page. It is whether the products shown to a shopper are the ones that shopper is most likely to buy. Bestsellers optimize for the average shopper, but no person is average. If someone views three pendant light pages in thirty seconds while browsing homewares, that is a clear intent signal. Showing the same category bestsellers afterward disregards it.
What Per-Shopper Ranking Changes
Per-shopper ranking makes each page request an opportunity to return a session-specific list. Instead of one sorted "Lighting" category, each shopper sees results shaped by signals from the current session and any available visitor history.
For a new visitor, ranking begins with contextual popularity and adjusts as the session supplies signals. Purchase history provides a prior for a returning shopper. After twenty minutes and six clicks among a particular price range and material category, the system can distinguish the next results from generic bestsellers.
The result is broader catalog coverage. Products absent from page one can appear for shoppers likely to buy them. Across our early-access retail partners, catalog surface area roughly doubled during the first 30 days versus the static category page baseline, measured by unique products appearing in at least one ranked result per day.
This Does Not Replace Merchandising
When merchandising teams first see per-shopper ranking, they often ask whether it removes control over new arrivals, collections, and business rules. It does not.
Per-shopper ranking covers the long tail, giving each product a chance to reach shoppers who may respond to it. Merchandisers can still pin a seasonal collection, exclude clearance from placements, or boost margin categories. Those controls sit above the ranked list, while the engine handles the remaining catalog.
That distinction changes the job. Rather than "curate 60 products for 100,000 visitors", the merchandiser sets constraints and lets the engine handle the rest. The task is more manageable and leaves time for editorial choices needing human judgment.
What This Change Means
Moving from category pages to per-shopper ranking is more than a technology change. Teams must accept that page-build time cannot reveal which products the next shopper will prefer. A model must make that decision in real time, which can initially feel uncomfortable.
Start with the real comparison. A category bestsellers list was already a model, just a simple one based on aggregate historical sales without individual signals. Per-shopper ranking is more precise because it uses more information. The test is whether, across thousands of sessions, it beats that simple model on average. That measurable comparison is the right way to judge the investment.
The difference is clearest for retailers above 50,000 active SKUs. Below that point, category pages can still expose a meaningful share of inventory. Above it, the math strongly favors per-shopper ranking when the goal is showing each shopper what they are most likely to buy from the available catalog.