Competitor Price Monitoring: Is It Legal? A Seller's Guide

Is it a crime to open a competitor's website and look at their prices?

Obviously not. But what if software does it for you a thousand times a day? What if that software then changes your prices based on what it sees? That's where most sellers get nervous, and the answers floating around online cluster at two extremes: "it's all fine" and "scraping is illegal, don't". Both are wrong in different ways.

This guide separates the legal layers of competitor price monitoring, explains how dynamic pricing and repricers actually work, and stress-tests the glossy numbers vendors advertise against a more realistic band. For a broader view of where AI fits your business, our industry-by-industry AI map has the wide-angle version; here the focus is purely e-commerce pricing.

Is competitor price monitoring legal? A three-layer answer

The short version: watching a competitor's public prices is fine; aggressive, unauthorized scraping is a gray zone; and agreeing with a competitor about prices, even through an algorithm, is the most serious competition-law violation there is. The question has three layers, and which layer you're standing on decides everything.

The first layer is safe ground. A price shown to every visitor is public information; whether you note it down by hand or with software doesn't change its legal character. The one condition that matters: you make your pricing decisions independently.

The second layer depends on method. Courts in most jurisdictions haven't treated web scraping as automatically unlawful; outcomes turn on the specifics. If a site's terms of service explicitly forbid automated collection, if your crawler puts real load on their servers, or if you're harvesting copyrighted content or protected databases, you're accumulating contract and unfair-competition risk. Bare public price data sits at the low-risk end. On marketplaces, mind one more distinction: tools that work through official APIs and approved integrations versus tools that scrape the site from outside. Check your seller agreement before you connect anything.

The third layer is the red line. Price coordination between competitors is treated as a hardcore violation by regulators on both sides of the Atlantic, and routing it through software rescues no one. We'll come back to exactly how that happens with a real case.

One more note while we're on this layer: it cuts both ways. While you monitor competitors, competitors monitor you. Putting an explicit automated-collection clause in your own terms of service, keeping an eye on bot traffic, and not leaking campaign prices before launch are the defensive side of the same framework. Price intelligence is a two-way game, and the party with written rules argues from the stronger position.

How dynamic pricing works, and what the Buy Box changes

Dynamic pricing means updating prices automatically based on signals like competitor moves, demand, stock, and seasonality. In practice, the seller sets a floor (a minimum price or minimum margin), and the tool scans competitors periodically and adjusts within those bounds. On marketplaces, the whole exercise usually aims at one thing: the Buy Box.

When several sellers list the same product, an algorithm decides whose offer gets the "Add to Cart" button. Amazon runs this way, and so do most large marketplaces; price is the heaviest factor. That's why repricing software has grown into a sizable industry of its own.

Price is the heaviest factor, though not the only one: shipping speed, cancellation and return rates, and seller rating feed the same calculation. That fact makes it easier to give your bot boundaries. Winning every last penny of the price war guarantees nothing; a seller with clean operations can hold the button while sitting second on price.

Before any of this, ask whether your floor is actually right. A floor set without subtracting commission, shipping, returns, and advertising costs turns the tool into a machine that sells you into losses, quickly and automatically. That is the most common damage a repricer does, and the cause is rarely the algorithm; it's incomplete cost accounting.

Can an algorithm form a cartel? A real case and an uncomfortable experiment

Two distinct dangers deserve separating. The first is deliberate agreement. In 2016, the UK's competition authority fined two sellers of posters on Amazon, Trod and GB eye, for a cartel: they had agreed not to undercut each other and used repricing software to enforce the pact. The penalty attached to the agreement, not to the tool. However innocent your software, a pricing understanding with a competitor is a cartel.

The second is stranger. Academic experiments have shown that reinforcement-learning pricing algorithms, with no communication and no one designing them to collude, can learn their way into supra-competitive prices through trial and error. Regulators are watching: one allegation in the FTC's 2023 case against Amazon was an internal algorithm that raised prices to test whether rivals would follow.

For a small seller the practical lesson is modest but real: write your bot's rules yourself instead of handing it a blank "maximize profit" mandate, log its price moves, and never discuss prices with competitors. A seller-group chat message saying "let's not go below $30 on this item" is the Trod case in miniature.

What the numbers say: the 25% promise versus the realistic band

Vendor websites advertise margin gains "up to 25%" and revenue jumps of 10-25% within six months. The source for those figures is, almost always, the company selling the tool. On the independent side, McKinsey's retail work points to a much humbler band for well-built dynamic pricing pilots: roughly 2-5% sales growth and 5-10% margin improvement.

Does 5-10% margin improvement sound small? For a seller doing $30,000 a month at a 15% gross margin, a 10% margin improvement compounds to roughly $5,000 a year. If the tool costs a tenth of that, the math speaks for itself. Base the purchase decision on that kind of conservative arithmetic; a vendor's landing page can't be the justification. Sellers who set expectations low are the ones who don't have to switch the bot off during their first price war.

One worked example of a correct floor. A product costs you $80 to buy; with a 20% marketplace commission, $9 shipping, 3% returns reserve, and 5% advertising, selling at $135 leaves you roughly $15, about an 11% margin. Let the bot drop the floor to $122 and the margin effectively evaporates; $117 is loss territory. The only correct number for the "minimum price" field is the one that still protects your target profit after every one of those lines is subtracted. Doing that math per product group takes minutes; skipping it is the fastest route to a loss machine.

Dynamic pricing on your own store: a different game

On a marketplace, dynamic pricing is a Buy Box race; on your own store, it becomes a customer-trust exam. The technical setup looks similar, but the risk profile changes: frequent, visible price swings can scare a customer out of the cart, and showing different prices to different people carries its own sensitivities in both perception and data protection.

Running price automation on your own store gives you two freedoms marketplaces don't: you choose the update frequency, and you choose what to scan. In exchange, the customer's price memory attaches directly to you. Someone who saw $95 in the morning and $110 in the evening reads it on a marketplace as "different seller has the offer now"; on your store, they email you. Sensible practices follow directly: keep update frequency low, freeze prices during campaigns, and avoid sudden increases on items already sitting in carts.

Personalized pricing, showing the same product at different prices to different customers, is something we'd steer you away from. It corrodes fairness perception, the reputational cost when discovered is high, and processing customer data for that purpose raises its own data-protection questions. Keep a bright line between transparent segment offers (a clearly advertised first-order discount) and hidden personal prices.

A concrete scenario: the repricer's first week

Picture a kitchenware seller on a large marketplace. Day one, the tool detects competitor movement on 90 of 340 products and updates 60 prices. Day three, two products look strange: a competitor ran out of stock, their bot pushed prices up, and our bot followed them to 20% above market. Sales stopped. Day five, the opposite: an aggressive rival starts selling at cost, our floor holds, the tool refuses to match, and the Buy Box is lost on that item.

By week's end the picture is clear: small automatic wins on 60 products, human judgment needed on 2. That ratio is dynamic pricing's real face; the bot's job is absorbing the routine, yours is managing the exceptions. There's also an invisible win in the logs: because every price decision is recorded, month-end shows exactly which product changed price how often and why. That audit trail is the least-discussed benefit of running a repricer, for reviewing your own decisions and for having a documented answer if pricing is ever questioned.

Price is only one lever, of course. For the others, our guides to writing product descriptions with AI and recovering abandoned carts cover the two next in line.

Frequently asked questions

Do I need permission to track competitors' prices?

For public prices, no. The risk comes from method rather than from watching: aggressive scraping that violates terms of service, heavy server load, and harvesting protected content are what create problems. On marketplaces, official API integrations are always the safest route.

Will a repricing tool put my marketplace account at risk?

Tools working through official integration channels carry low risk. Tools that scrape the site from outside, ask for your account credentials, or won't explain how they get data can put you in conflict with your seller agreement. Ask any vendor, in writing, which channel they use before connecting.

When would I actually be violating competition law?

The moment you reach an explicit or tacit pricing understanding with a competitor. Software changes nothing: in the UK's Trod case, the fine attached to the sellers' agreement, and the repricer was merely its enforcement mechanism. Setting your own prices, by your own rules, on your own, remains entirely legal.

How often should prices update?

On a competitive marketplace, hourly scans are common; on your own store, once or twice a day covers most categories. The honest yardstick is competitor volatility: if your rivals move prices once a day, minute-level scanning buys you nothing but server load and noise.

So what should you do?

• Finish your cost accounting first: no repricer until you know true unit cost including commission, shipping, returns, and ads.
• Make the data channel your first vendor question: official API or scraping?
• Constrain the bot: floor price, ceiling price, and a daily change limit, with exceptions routed to a human.
• Pilot for a month on 30-50 products, measure margin and Buy Box impact on your own data, and scale based only on what you measured.
• Never discuss prices with competitors, and stay away from "common floor" proposals in seller groups.

We opened with skepticism, so let's close with it: dynamic pricing is neither a miracle nor a trap. A bot whose rules you wrote and whose limits you set is a good employee; one left unsupervised is an expensive intern. If you'd like a second opinion on how to set that balance up for your own catalog, our inbox is open.