The 2026 State of Shopify Competitor Pricing

Every merchant has a theory about how competitors price: they cut when they need volume, they hold firm on flagship products, they run sales around the same weekly rhythm you do. Most of these theories are wrong, or at least wrong at the scale that matters. This report pulls together every original finding Beaconmon has published from its own monitoring data into one place, refreshed against our full current dataset.

The underlying findings are published individually across our research index. This report is the narrative version: what the numbers mean together, not just what each one says on its own. See our methodology for how the data is collected.

Report at a glance

Every stat below links to the full post with complete methodology and breakdown.

Finding 1: Most price activity is noise, not decisions

90.76% of price change eventsin our dataset are sub-1% moves: a product shifting from $49.99 to $49.50, a variant moving from $24.00 to $23.97. These are not pricing decisions. They are artifacts of dynamic pricing apps, currency rounding, and bulk sale rules running automatically in the background of a competitor's store. Only 1.88% of all price changes are 5% or larger, the threshold above which a move is likely to represent an actual decision. A monitoring setup with no threshold filter treats all of this the same way, which means the 1.88% that matters gets buried inside the 98.12% that doesn't. Full breakdown in our noise-filtering post.

Finding 2: Prices go up more than they come down

55.1% of price changes are increases, 44.9% are decreases, the opposite of what most merchants assume competitors are doing. At sub-1% magnitudes this direction split is mostly automated noise too, but it means monitoring configured to watch only for cuts misses the majority of pricing activity. The moves worth acting on, in either direction, live in the 5%+ bucket: a meaningful increase on a hero SKU signals margin testing the same way a meaningful decrease signals a threat. See the full price-direction breakdown.

Every merchant we talk to has a mental model of their competitors built on a handful of moves they happened to notice. The data says something different: pricing behavior is mostly automated, heavily clustered midweek, and split by category in ways that don't match intuition at all.

Finding 3: Price changes cluster hard on Wednesday and Thursday

65.67% of the 472,712 price changesin our dataset happened on a Wednesday or Thursday. Combined, Saturday and Sunday account for just 1.34%. If you check competitor pricing manually once a week, checking on a Monday or Friday means you are looking at a catalog that hasn't moved much yet, or one where the week's activity has already settled. The practical implication: a monitoring cadence that only samples occasionally will miss most of the action entirely, since it's concentrated into roughly two days out of seven. Details in the full timing post.

Finding 4: Restocks are fast, except when they aren't

We matched 1,556 out-of-stock to back-in-stock pairsand found a median restock time of 22 hours against an average of 45.34 hours. That gap is the finding: most competitor restocks happen within a day, but a long tail of slow restocks, some taking several days, pulls the average well above the typical case. If you're trying to time a stockout opportunity, the median is the more useful number to plan around, not the average. Full analysis in the restock speed post.

Finding 5: Discounting depth depends heavily on category

Overall, 23.42% of the 180,065 products we track carry a compare-at-price right now. That average hides a sharp category split: apparel competitors keep 39.01% of their catalog on sale at an average discount depth of 53.00%, while skincare competitors keep just 9.04% on sale at a shallower average depth of 24.01%. Deep, frequent discounting is the norm in apparel and the exception in skincare. A monitoring setup that treats every vertical the same way will flag apparel discounting as constant noise and miss the rare, meaningful discount events in skincare.

See the category breakdowns in the fashion pricing post and the skincare pricing post.

Finding 6: Homepage changes are almost always copy, not redesigns

Of 2,666 classified homepage changes, 47.79% are copy edits, 23.93% are feature changes, and only 7.99% are layout changes, six times fewer than copy edits. Merchants who assume a homepage watch is mostly for catching redesigns are watching for the rare event. Most of what a homepage monitor actually catches is a headline swap, a new promo banner, or updated hero copy, the exact signals that matter for tracking a competitor's current messaging and offers. Full breakdown in the homepage change anatomy post.

Finding 7: Only 1 in 19 events is worth an immediate reaction

Beaconmon's AI significance classifier scored all 504,500 events in this dataset. 94.68% came back medium-significance, 5.32% high-significance, and a negligible remainder low. That 5.32% figure is the practical case for significance scoring over raw alerting: if roughly 1 in 19 events is worth immediate attention, an alert channel built on raw events will surface 18 low-value notifications for every one that matters. See the full significance breakdown.

What this means for how you monitor

Set a threshold before you turn on alerts

A 5% floor on price-change alerts eliminates the 98.12% of events that are noise without missing the moves that matter. Without a threshold, any monitoring setup degrades into background noise within days.

Weight your monitoring cadence toward Wednesday and Thursday

If you review competitor activity on a fixed schedule rather than real-time alerts, align that review with when activity actually happens. A Friday morning review catches nearly two-thirds of the week's price changes; a Monday review catches almost none of it yet.

Use category norms, not a single discounting threshold

A skincare competitor discounting 15% of their catalog is a meaningful shift from their 9.04% baseline. An apparel competitor doing the same is below their category's 39.01% norm. The same raw number means opposite things depending on the vertical.

Let significance scoring do the triage

With only 5.32% of events landing as high-significance, routing raw events straight to an alert channel guarantees alert fatigue. Significance scoring, or a manual 5%+ threshold if you're not using AI classification, is what keeps a monitoring setup usable past week one.