How to speed up checkout lines when your counter carries thousands of SKUs

A hardware and fasteners distributor running a walk-in counter alongside its van sales routes carries about 4,000 SKUs — bolts, hinges, adhesives, tools, a dozen variants of the same bracket in different finishes. Contractors and small retailers stop by in person to pick up what they need without waiting for a route day. The counter should be the fastest way to buy from this distributor. Most days, it's the slowest.

The line isn't a staffing problem

The owner's first instinct was to add a second register during the Saturday morning rush. It didn't help. The bottleneck wasn't the number of registers — it was what happened at each one. Every sale started the same way: cashier types a few letters of the item name into a search box, scans through a dropdown of near-identical results, picks the right SKU, repeats for the next item. A contractor buying eighteen line items meant eighteen rounds of type-scroll-select, each one a chance to grab the wrong gauge or finish.

None of this shows up as a single dramatic failure. It shows up as a customer who's been standing at the counter for six minutes buying eight things, a line of three people behind them, and a cashier who has learned to type fast instead of a system that lets them not type at all. Multiply six minutes by every counter sale in a week and the "fast in-person purchase" the counter was supposed to offer stops being true.

Search boxes don't scale with catalog size

A search bar works fine when a store carries 200 items. At 4,000, it actively fights the cashier. Typing "bracket" returns eleven results that differ only in size or finish, and the customer standing in front of the register rarely knows the exact SKU name printed on the box — they know it by look, by aisle, by "the one I bought last time." The cashier ends up guessing, then confirming with the customer, then re-typing when the guess is wrong.

The deeper issue is that a generic search box treats every item as equally likely to be bought next, when in reality a small fraction of the catalog accounts for most transactions. Fast movers — common bolt sizes, popular adhesive, the top three bracket finishes — get typed dozens of times a day, identically, by every cashier. A system built for "search anything" makes no distinction between an item sold fifty times a day and one sold once a quarter. Both cost the same three or four keystrokes and the same visual scan of a dropdown.

What actually fixed it

The distributor switched from a search-first checkout to a grid of large, tappable tiles arranged by category — fasteners, adhesives, hand tools, finishes — with the genuine best-sellers pinned to the top row regardless of category. A cashier ringing up a common item taps twice: category, then item. No typing, no dropdown, no guessing between near-duplicate names. Search still exists for the long tail of items that don't sell often enough to earn a tile, but it's no longer the default path for the 80% of line items that are the same handful of SKUs over and over.

The tile layout isn't static. It's set up per counter and can be re-arranged as the mix of what sells shifts by season — more adhesive tiles up front during a renovation season, more of a specific finish when a contractor client places recurring orders. Because it's configured once by whoever runs the counter, a new or part-time cashier reaches full speed almost immediately instead of needing weeks to memorize where things live in a search index.

The measurable result wasn't a stopwatch improvement on any single sale — it was that the Saturday line stopped forming past the third person. Average items-per-transaction didn't change; time to ring up each item did.

What to check before you blame the register hardware

• Count keystrokes, not minutes. Watch one busy hour and count how many times a cashier has to type or scroll to find an item that sold yesterday and will sell again today. That number is your real bottleneck, not checkout duration.
• Separate fast movers from the long tail. Pull a sales-by-SKU report and see how much of counter revenue comes from a small set of items. If it's a small set, they deserve one-tap access; everything else can stay behind search.
• Make the layout ownable by the counter, not IT. If changing what's on the quick-pick screen requires a support ticket, it will go stale and stop matching what actually sells.
• Keep search as a fallback, not the front door. The long tail still needs to be findable — just not at the cost of making the common case slow.
• Re-check the layout each season. What's fast-moving in spring isn't what's fast-moving in winter; a tile layout set once and never revisited slowly turns back into a search box with extra steps.

iotoms' point of sale ships with QuickPick layouts built for exactly this — a counter arranges its own tiles by category and best-seller, shares the same layout across every register, and keeps full search available underneath for anything that isn't pinned. The rest of the sale — stock deduction, pricing, and the ledger entry — happens the same way whether the item was found by tile or by search, so speeding up the counter doesn't cost you accuracy anywhere else in the business.