26 Nov 2025 Search

Mark Ritson writes, in Black Friday is a global price trap and marketers keep falling for it,

Price discounting is the most stupid fucking thing you can ever do.

In theory, Black Friday is a simple exercise in price discrimination. People who can afford to avoid the crowds can stay home on Black Friday and pay a higher price on some other day. People who are willing to have a hellish shopping day can get a discount.

But, as Ritson points out,

Thanks to scholars at Wharton, we have known for decades that the single most effective way to drive corporate profits is to maintain a premium price. The reverse logic also applies. If you want to destroy your company’s profitability – ultimately the only thing that matters – just keep running ever-deeper price promotions.

Customers who would have purchased at full price feel like they don’t want to get ripped off, so they learn to wait for the next sale. The better-managed retailers can avoid the whole mess. REI has an #OptOutside event instead. Employees get the day off, the brand is protected, and the overflow crowds from the other stores can park in the REI parking lot.

AI Black Friday is 364 days a year

Black Friday is simple price discrimination, but retailers are running more complex kinds the rest of the time. And the same arguments against Black Friday discounting also apply to surveillance pricing—which works like Black Friday discounting does, but in a more complicated and costly way.

Instead of a simple rule for who gets the discount, like are you willing to fight the crowds on a particular day, modern surveillance pricing uses more complicated rules, managed by machine learning. If an ML system identifies me as a likely car owner and Costco member, who has better alternatives to the product on offer, then I’m going to get a discount code. If ML classifies me as someone with fewer options, I won’t get the code.

Put surveillance pricing on a slide in an Econ 101 lecture, and it looks like a big win for the retailer. Consumer surplus captured, total profit higher, what’s not to like? In the real world, though, Surveillance Pricing Strikes a Nerve because it sits at the intersection of two things most people revile: feeling tracked and feeling ripped off. The customer reaction to surveillance pricing is never, wow, that’s neat, how can I get an offer better personalized to me? it’s always how can I get an offer better personalized to some other person, the person you offered the best deal to?

The Honey browser extension was tremendously popular even though users don’t install extensions because it offered a credible promise of de-personalizing interactions with a retailer. Instead of getting the for me price, you could get the for bargain hunters price. Honey was so popular that it had to be bought and subverted. But a retailer who invests in price discrimination today has to be looking over their shoulder for other anti-price-discrimination tools and services. A retailer’s choice to do price discrimination is a choice to start playing a game that has its own costs. The services and employees who enable price discrimination aren’t free—and the cost of operating a price discrimination scheme is money that can’t be put into product or service improvements.

Lots of posting going on right now about the potential of personal AI shopping agents. But there’s some adversariality here. The best-performing personal agent is not going to go tell retailers about your personal needs and preferences, it’s going to make an effort to represent you as a price-sensitive consumer with a lot of alternatives. If you absolutely need some kitchen appliance to make a recipe you already told people you would make, it’s going to represent you as a Costco member who has all the appliances they need. Internet Marketing Thought Leaders tend to make the mistake that people want personalized ads and offers, when the real situation is more like people want the ads and offers that the Internet Marketing Thought Leaders get.

If AI agent shopping ends up working, it will because some kind of ML on the customer side will be able to size each possible deal up and reveal information only as needed. Although only 8% of Americans would pay extra for AI, people are motivated to avoid price discrimination. Retailers are not going to stop price discriminating any time soon, which means there’s a sustainable niche for more sophisticated shopping optimization tools on the customer side. More: A structural subsidy for product quality