Ask whose sales data your pricing tool learned from.

What the lawsuit says, and what it doesn't yet

On October 2, a McDonald's customer filed a proposed class action in federal court in Chicago. The case is Thomas v. McDonald's USA, and the complaint makes a specific claim: that the company's AI-enhanced pricing tool collected private, store-level sales and pricing data from thousands of restaurants that are supposed to compete with each other, ran it through one engine, and sent back recommended prices that restaurants then followed.

McDonald's has said franchisees set their own prices and that the tool only recommends them. Nothing has been decided. No class has been certified. A complaint is one side's account, and it'll be a long time before a court says whether any of it holds up.

So this isn't a note about whether McDonald's did anything wrong. It's about the question the case puts in front of anyone who's thinking about buying software that suggests prices, whether you run three restaurants, a hotel group, a chain of stores or an online shop. That question is simple, and most sales demos skip it: what is this thing learning from?

The useful question isn't how clever the pricing engine is. It's whose numbers it's been reading.

Why it isn't only a McDonald's story

The complaint follows a path that's been walked before. In the rental housing market, the U.S. Department of Justice sued RealPage, a company whose software recommended rents to landlords, over the same basic concern: private data from competing landlords flowing into one tool that then helped set their prices. The government's claims against RealPage ended in a settlement, and the court entered a final judgment in May 2026 that, among other things, bars the software from using competitors' nonpublic data when it generates live pricing recommendations.

The pattern regulators and plaintiffs keep pointing at isn't "software recommends prices." Plenty of perfectly ordinary tools do that. It's three things together: competitors' private numbers going into a shared pool, a common engine turning that pool into recommendations, and businesses that are supposed to compete following those recommendations.

You can't judge the legal side from a blog post, and we're not going to try. If any of this sounds like a tool you already use, that's a conversation for a lawyer. But you can ask much better buying questions than most people do, and that part is squarely our territory.

Three questions to ask before the demo ends

What goes in? Ask the vendor to name every data source the recommendations draw on. Your own sales history, your stock levels and your costs are one thing. "Market data" is where it gets vague. Is it public, like prices anyone can see on a menu or a website? Or is it private numbers from other customers of the same vendor, maybe including the business down the street? If the answer is "aggregated and anonymized," ask what that means in practice, and how many other businesses in your area are in the pool.

What comes out, and who acts on it? A tool that shows a suggested price with its reasons is a different animal from one that pushes prices to your point-of-sale or storefront automatically. Find out which you're buying. Ask whether you can set your own limits, override any suggestion, and see afterwards which prices the tool changed and which a person did.

Who else is using it near you? It feels like an odd question to ask a vendor, but it's a fair one. If most of your direct competitors run the same engine fed by the same pool, your "independent" prices may not be very independent. A good vendor will have a clear answer and won't be offended.

Write the answers down. If they change between the sales call and the contract, that's useful information too.

One more thing to watch for in the demo: features that quietly line your prices up with everyone else's. A setting that discourages you from dropping below "the market," or a dashboard that ranks you against named competitors using their private figures, tells you a lot about how the pool is being used. Ask the vendor to show you where those numbers come from, on screen, before you sign anything.

Keep the useful parts in your own hands

Here's the thing that gets lost in the excitement about pricing engines: for most businesses, the data that matters most is already yours. What sold, at what price, on which day. What it cost you. How much you had on hand. Which promotions ran and what happened afterwards. A lot of good pricing work is simply getting that information into one place and looking at it honestly.

A few habits make pricing easier to defend, whatever software you use:

• Rules you can explain. "We raise the weekend price on this item when stock is low" is a rule a manager can say out loud. "The model said so" isn't.
• A person who approves changes, at least above a threshold you set, with the reason recorded.
• A log of why each price moved. Date, old price, new price, who or what changed it, and the reason. If anyone ever asks, you'll want this.
• Your own limits. Floors, ceilings and the products you never want touched automatically.

None of this needs a pooled dataset. It needs your own records to be clean and connected.

Buy a tool, or build the part that matters

Sometimes a vendor pricing tool is the right answer. You get a team that maintains it, and you don't have to staff a data project. If you go that way, put the answers to the three questions into the contract, along with your right to export your own data and your price history if you leave. Our note on scoring build vs. buy on evidence has a useful way to compare options side by side.

Sometimes the better answer is smaller and more yours: a pricing screen inside the systems you already run, fed by your sales, stock and margins, where suggestions come with reasons and a manager approves them. That's often less software than people expect, and it keeps the data you'd least like to share exactly where it belongs.

We connect storefronts, inventory, payments and fulfillment in our ecommerce work, and we build AI applications that show their sources and leave the decision with a person. If you're weighing a pricing tool, or wondering what your current one is actually reading, let's look at it together.

Related work: Paddle8.