24,394 identical products, one store's price sitting 15.6 percent above the cheapest listing, and a half-price Shark Stratos vacuum that snapped back to full price within three days: read together through November 20, 2026, these numbers say promotional claims on e-commerce product pages cannot be trusted without a frozen basket and a matched-item check. The research behind them tracked a random set of goods across major US retailers before the sales began and then watched what each price actually did. For brand teams managing price order, the study turns a vague worry into something measurable, and it puts the burden of proof on the listing rather than on the shopper.
A Basket Frozen Before the Sales
The design is the interesting part. An AI shopping agent locked a random basket of 972 products on November 5, before any Black Friday price was public, then checked every store at the same hour each day through Cyber MondayTechBullion. A separate study of 24,394 identical products across 16 major US retailers found that a randomly chosen store averaged 15.6 percent above the lowest listed price, while comparing three stores cut that gap to 3.7 percentPress Release Feed. Freezing the list first stopped anyone from swapping in products that would make a better headline later.
Matching Identical Items
A price comparison is only as honest as the identity of the items behind it. The researchers ran a second model over each store's listing title with prices hidden, keeping only genuinely identical products: same model, size, capacity, pack count and generation. In an earlier pass, just 1,886 of 4,990 store listings survived that test. For a brand, this is a data-hygiene problem disguised as a pricing problem; if a product page omits capacity or generation, an agent cannot tell the current model from an older one.
The Three-Day Reversal
The volatility was as striking as the level. A Shark Stratos vacuum was listed at 264.99 dollars at Target, Best Buy and Kohl's three weeks before the event, then returned to 529.99 dollars within three daysPress Release Feed. A shopper who noticed the deal and waited missed it entirely. Interviews with the team behind the study describe the method as a deliberate control group, built so that late marketing pivots cannot rewrite the resultB2BDaily.
What Price Governance Can Learn
The headline number is about consumer convenience, but the operational lesson sits one layer down. About one in three products at Walmart and Amazon, and half of those at Target, were listed for less at another large retailer at the same moment. That is a price gap inside the channel, and it is visible to any agent comparing listings. A brand watching only its own product page will see a stable price and conclude that the channel is healthy, while the same item is discounted elsewhere.
Price monitoring built on titles rather than identifiers will misread this. An alert fires on a different generation, the team checks and finds nothing wrong, and soon the alerts are ignored. The fix is unglamorous: make model number, capacity and pack count structured fields on every product page, and treat the title as a description rather than the identity. Only then does a gap mean the same item is priced differently, which is the signal worth acting on.
From Anecdote to Monitoring
Three checks turn the study into a working rhythm for an e-commerce team. None needs new infrastructure, but all need a stricter definition of what counts as the same item and the same price. The table below sets out the definitions and the conditions that should trigger a closer look.
| Check | Definition | Trigger |
|---|---|---|
| Identical-item match | Same model, size, generation | Title-only match rejected |
| Price gap | Random store vs lowest listing | Gap above 5 percent |
| History check | Price vs prior 30 and 60 days | Pre-event price rise |
Read History Before the Banner
A one-year chart makes a promotional claim easy to test. Before a campaign goes live, check whether the event price is the lowest of the last thirty and sixty days and whether the price rose shortly beforehand. If it did, the campaign has a credibility problem no amount of traffic will fix. The frozen basket did not create the 15.6 percent gap; it simply made a familiar one impossible to ignore, and it shifted the burden of proof onto the seller.
FAQ
Does a 15.6 percent gap mean one retailer is overcharging?
Not necessarily. It is an average across 24,394 identical products and 16 large retailers, describing the cost of checking one random store rather than any single seller's intent. Some products were cheaper elsewhere at the same moment.
What should a brand fix first on its product page?
Identifiers. If model number, capacity and generation are missing or inconsistent, a comparison agent cannot match the item correctly and price data will be misread. Fix the structured fields before changing promotional copy.
When should a price anomaly alert be trusted?
Only after the item match is confirmed. More than half of store listings that looked similar were rejected as non-identical, so an alert built on title similarity fires on the wrong model and quickly loses credibility.
