C (2:35)

Derek yeah, so Instacart bought a small AI company a few years ago called Eversight, and as a result, they are far and away the most sophisticated tech company in retail. And so many companies are investing in AI. But, but specifically, Instacart uses AI in a way that Supercharges or turbocharges their data collection and their use of data for individual shoppers. And they're huge. They've grown to now roughly 300 million orders on pace for 2025. So we really just wanted to know how much AI is driving grocery store prices through a service like Instacart. How can grocery retailers use this to set prices that you see in app or online?

B (3:29)

Welcome, welcome, welcome. You have found your way to the Consumer Reports grocery pricing webinar.

C (3:35)

So we recruited hundreds of volunteers from across the country to hop on video calls with us.

B (3:40)

What has begun. Virginia, Denver Yo, Gene, thanks for being with us.

C (3:45)

And we all shopped live, right? We all shopped for the same exact items from the same exact grocery stores at the same exact time.

A (3:53)

Enter the word honeycrisp H O N E Y C R I S P.

C (3:59)

Trying to control for as many variables as we could to understand how Instacart is using this tech, this AI to inform their grocery prices.

B (4:08)

Sarah is getting charged $2.44 for an apple.

A (4:12)

Can anybody beat Sarah?

C (4:15)

$2.

B (4:16)

Brian. Winner, winner, chicken dinner.

C (4:18)

What we found were, you know, 75% of the products we tested had algorithmically changed prices, everything from 7 cents on the all the way up to $2.56 on the high end. And some of the products had really large price variances. Skippy Peanut butter had a 23% price variance between the low and the high. And, you know, that adds up to real money over time. When we extrapolate that out over the course of a year using Instacart's own estimates for how much Americans spend on groceries in a given year, that's the difference between $1,200 for a household of four.

B (4:56)

We'll keep on it. We'll figure out what, what we can do and how we can guide you all through it. And again, thank you all so very much. 9:01 Eastern standard. So we'll, we'll sign off now.

A (5:09)

What are you saying when you say prices were algorithmically changed, what was actually happening?

C (5:15)

You know, a lot of people liken it to like a B testing. And when I say that, I mean, you know, they're toggling one price, 5, 10, 15 cents more, and then another price, 5, 10, 15cents less. And trying to figure out that perfect M of price points that will compel you, the shopper, to buy those products. If you're more willing to pay more money to buy alcohol or sweets on a Friday, late on a Friday, and you group all those products together on a Friday for a company, they understand that purchase history, and they understand that they can charge you more based on that purchase history. And they can sell those insights, that information and data to retailers as a really valuable bit of information on you, the shopper. And of course, retailers want that. And they see, you know, the ability to make more money, especially in a business like grocery that has really low margins. Historically, this means something. This, this 1, 2, 3 percentage points means hundreds of millions of dollars in additional revenue. And that's the difference between, you know, making a profit or not.

A (6:24)

Okay, so your investigation finds out what instacart is doing. And then you call up instacart and you say, guys, we know what you're doing. What does Instacart say?

C (6:34)

So they did say, look, this is something that we're pretty open about. We've been telling our business clients for years this is happening. It's online, it's out there. And look, we still believe that it's negligible price differences, small, limited time and randomized. And really, at the end of the day, it helps grocery retailers and us know which products people care most about. And trying to, you know, in the attempt to try to make groceries more affordable for more Americans. You know, our data showed something a little bit different. Very few people in our testing got the lowest price possible. When we looked at it, 8% of our testers got the lowest available price on products, and everyone else, 92%, got something higher. And, you know, generally speaking, when we look at the marketing materials and things that are out there buy Instacart for their grocery clients, it's all in an attempt to max, maximize profits and to, again, get a little bit more sales revenue or a little bit more profit margin. And really, that means you're either buying more or you're paying more per item. Right. So it's. And usually a combination of both.

A (7:48)

Okay, so instacart is able to do this because they have this small AI company, Eversight. And Eversight AI is, is clever. And it's figuring all this out. Is it just Instacart? Are other companies doing this to us?

C (8:01)

Other retailers are doing it, yeah. And, you know, the more technologically savvy they are, the more likely they are to be using this at a scale that really affects more Americans. When we looked at a grocery retailer called Kroger, which is actually the second or third largest retailer by most metrics, they own a bunch of different banners, King Supers and Fred Meyer and QFC and all these grocery chains around the country, we found that they actually use A lot of personal and demographic data on their customers. And they drive a lot of people through their free loyalty program. And then with that data, they tailor promotions and discounts to particular people. And they use that personal data, those, those demographics, to dictate who gets what promotion or discount. So we, we find that, again, a lot of retailers are using this data and using it effectively, and really it bears out in their ear and in their quarterly reports. If they're publicly traded or even if they're privately held, we see a lot of increase in meaningful profits and revenue for some of these companies.

A (9:16)

Is this legal?

C (9:18)

It's a very good question. It's really a gray area. I think everyone we spoke to about this acknowledged that, look, there's no specific federal surveillance pricing bill. And when I say surveillance pricing, I'm talking about the idea of using your shopping history or your buyer behavior or demographics even to inform the prices you see, whether that's a set final price or a discount or a promotion. There's no federal surveillance pricing bill. Some states now have laws on the books. New York is one of them. They just enacted a law where if someone is algorithmically changing the price you see in an app, whether it's an Uber or a Lyft or a grocery retailer, they have to have a label in the app saying this price is being algorithmically changed. And here's why. It's a disclosure, basically. But that doesn't inhibit or it doesn't prevent them from doing the practice. Right. It just shows you that they're doing it. And so other states are figuring this out. We've been talking to a lot of regulators recently who are actively considering new laws. And the FTC even has an active probe into surveillance pricing. So it's an open question.

A (10:28)

Yeah, you know what, I wonder, I wonder if there is a case for this. So let's say you have two American neighborhoods. One of them is very upper income, one of them is very lower income. And the people in the upper income neighborhood, they spend a lot more because they can spend a lot more. Lower income neighborhood, they're strapped. And food is very expensive. Is there, is there a case for this that says, look, if you're rich, you should be able to pay the 350 for the peanut butter or the 550 for the peanut butter. And if you're in a poor neighborhood, okay, we charge you 250?

C (10:59)

Yeah, there is a clear case for that, except that most state and even federal laws bar discrimination, price discrimination on the basis of your zip code or your household income or wealth for good reason. That can be proxies for race, ethnicity, various things that we as a country do not believe that you should be discriminated against because of your background or who you are. We're moving into a very fast paced, quick use of this really sophisticated tech with now AI involved and you know, regulators and the American public are playing catch up. And so we're trying to figure out, you know, what this all means and how we should best react to it.

A (11:44)

You know, one thing that we hear about AI Derek is that it is inevitable. This is the future. This is where things are headed. Do you think we are headed toward a future where different people pay different prices for any number of things? Peanut butter, airline flights, concert tickets, based on what retailers know that they can and will pay.

C (12:08)

So we're already there to some degree. But you know, one thing that really sticks out, we spoke to Lina Khan who is the chair of the FTC during the Biden administration. Now she's advising the new New York City Mayor elect Zoran Mamdani. And you know, she, when we sat down with her at Columbia Law School, she was, was pretty clear. We need to have a first order conversation is what she said. We need to think about. Do we want, want our personal data? Do we want things that companies know about us to be used against us in this way? Do we want our own personal data to inform the prices we see? Because if we do, then we're there. If we don't, then we need to pass some meaningful laws and regulations that speak to that, that address that.

A (13:06)

That was Derek Kravitz of Consumer Reports. Coming up, Congress acts. Jk, jk, jk. Congress doesn't do anything. Support for today's show comes from Indeed. Indeed says right now there is someone out there in the world who could help you take your business to the next level. And if you want to find that person, you can do it with Indeed's sponsored jobs. Indeed's sponsored jobs can give your job posting the best chance to be seen by quality candidates, says Indeed. And their data shows it works.works says Indeed. Indeed says sponsored jobs posted directly on Indeed are 90% more likely to report a higher than non sponsored jobs. Why? Because apparently you reach a bigger pool of people with Indeed. You can spend more time interviewing candidates who check your boxes. Less stress, less time, more results. Now with Indeed sponsored Jobs, listeners of this show will get a $75 sponsored job credit to help get your job the premium status it deserves. @ Indeed.com today explained you can go to Indeed.com todayexplained right now and support our show by saying you heard about it on this podcast. Indeed.com todayexplained terms and conditions do apply. Guys hiring do it the right way with Indeed. Support for Today Explained comes from Vanta. Customer trust can make or break your business, according to Vanta. And the more your business grows, says Vanta, the more complex your security and compliance tools can get. Left unchecked chaos. That's where Vanta comes in, says Vanta. Think of Vanta as your always on AI powered security expert who scales with you. Vanta says Vanta automates compliance, continuously monitors controls, gives you a single source of truth for compliance and risk. Whether you're a fast growing startup like Cursor or an enterprise like Snowflake, Vanta fits easily into your existing workflows so you can keep keep growing your company. Get started at Vanta.com explained that's V-A-N-T-A.com explained Vanta.com explained. Support for today's show comes from Mint Mobile. The holidays are supposed to be a time of joy. Mint Mobile reminds you. Mint Mobile says you don't have to let overpriced phone bills take that joy away from you. And so Mint Mobile is offering the following 3, 6 or 12 months of unlimited premium wireless for $15 a month you can turn your expensive wireless present into a huge wireless savings future Ghost of Christmas whatever joke there. By switching to Mint. You can shop Mint unlimited plans@mintmobile.com explained. That's mintmobile.com explained this is a limited time offer. Let's do a little math. Upfront payment of $45 for three months, $90 for six months or $180 for 12 months. It down is $15 per month. Taxes and fees are extra, guys. And this is the initial plan term only. Greater than 35 GB may slow when network is busy. Capable device required availability, speed and coverage will vary. See Mint mobile.com. This is Today.

B (16:58)

Explained. My name's Alfred Ng. I'm a tech policy reporter for.

A (17:01)

Politico. In the first half of the show, we talked about surveillance, pricing, companies charging you more because they think or know that you will pay more. And we also heard that the ftc, which protects us from monopolies and other unfair trade practices, launched a probe into this practice. Where does that.

B (17:23)

Stand? So the probe was released in the final days of the Biden administration in January, just a couple of days before, you know, President Trump had taken.

A (17:32)

Over a new report this year from.

B (17:33)

The Federal Trade Commission confirms something known as surveillance pricing. A person's precise location or browser history can be frequently used to target individual consumers with different prices for the same goods and services. It was very much released as, you know, kind of preliminary findings from, you know, a study that they opened up in, I think, July of 2024. So it wasn't a completed study. And the premise was, you know, we hope that the next administration will continue this study that didn't end up happening. The study was actually ended under the new FTC chair, Andrew Ferguson, in.

A (18:13)

January. Hmm. Okay. So they didn't have very long to do their investigation, but they did have some time. Do we know what they.

B (18:20)

Found? Yeah, they had found from the report that, you know, a lot of these companies that were contracted with, you know, a lot of retailers were making these promises, right, of, you know, we can use AI to basically set different prices for people. They didn't exactly use the term pain points, but that is a term that's been, you know, going around a lot with relations to surveillance pricing, where it's basically what is the most someone is willing to pay for this certain product or something like that, and basically saying, we can, you know, set these prices this way so that your company can earn more.

A (18:59)

Money. Do we know how widespread this is? We know that Instacart is doing it. Do we know for sure who else is doing.

B (19:06)

It? So most of what we know about use of this usually comes from the companies themselves. Right. There's no federal requirement to say, like, you know, hey, you have to tell us if your company's using this algorithm to set individual prices for people. But, you know, Delta Airlines, for example, you know, people learned about them using dynamic pricing. And I'm saying this air quotes through, like, their earnings calls, right, where they're telling investors, here are some ways we can see ourselves making more profit here by, you know, setting different prices. Delta Airlines told investors their pilot program to tailor pricing based on an AI analysis was successful. They've set a goal to have a fifth of all of their fares set by AI by the end of the.

C (19:57)

Year. We like what we see. We like it a lot, and we're continuing to roll it.

B (20:01)

Out. And they've basically said, we don't use personalized data for people to set these prices. And I think that's why, you know, people use terms like dynamic pricing or individualized pricing, as opposed to what the FTC called it, which is surveillance.

A (20:15)

Pricing. The response to this from consumers, you know, no surprise, is a lot of outrage. I know this sounds Legitimately crazy. But I stopped signing up for loyalty programs with companies because of surveillance pricing. I mean, how crazy would it be if brands jacked up prices based on your previous shopping history, had data on you that said you were desperate, and knew that you would spend the money no matter.

B (20:41)

What? Oh, they.

A (20:43)

Do. This seems like easy pickings for Congress. Has Congress shown any interest in taking on surveillance.

B (20:51)

Pricing? You're right. This does seem like easy pickings for Congress. Right. Because it hits on these two issues that are very salient right now, which is affordability and also regulating big tech. And we've seen lawmakers, both Republicans and Democrats, talking about it being a big problem. When it comes to legislative solutions, though that really hasn't been quite the case. Senator Ruben Gallego introduced a bill last week that would ban surveillance pricing. We as policymakers want to make sure that there is an even playing field between the consumer and the corporations. But he has no co sponsors on it. He doesn't have any Republicans backing the bill or anything like that. It's one Democrat that's proposed legislation.

A (21:37)

On this that is fascinating to me because it seems like it would be a big win for anybody who starts banging the drum early and gets everybody else on.

B (21:46)

Board. It does seem like it would be a big win, you know, for, for somebody to get on this issue early. I know that Senator Josh Hawley, a Republican, has actually criticized, you know, airline CEOs for this.

C (22:00)

Practice. Okay, so you don't collect personal information before people can see a.

B (22:04)

Price. No, we don't. We don't. We don't. We don't use personal.

C (22:07)

Information. You collect it, though. We collect it, but we don't charge you differently. Well, how do I know that? Why are you collecting all of this personal information? What guarantee do I have that you're not using this to set my price? I have. I have no idea how you're setting the.

B (22:19)

Price. I just guarantee you we're not. I think it really comes down to, you know, what is in the terms of the bill. Right. Because I know that a lot of lobbyists have worked against these kinds of surveillance pricing bills to say, you know, well, if you pass this legislation the way that it's written right now, we're not going to be able to offer military discounts or we're not going to be able to offer discounts to teachers or, you know, any of these discounts that like, have existed for many years without algorithms suddenly are not able to exist if you ban setting prices based on people's personalized.

A (22:57)

Information. So no action. Very little action at the federal level. I wonder about the states though, because we talked in the first half of the show about a New York law where a company has to say whether or not they've used customers data to set prices. And that's going on in New York State. New York is a pretty progressive state, all things considered. How has that been.

B (23:19)

Received? So in New York, that bill was actually watered down from its original intent, huh? Originally, that bill was supposed to be an outright ban on surveillance pricing. A lot of lobbying groups, you know, went against it, making the arguments that I mentioned earlier, saying if you pass this, you're actually going to hurt people. On affordability where we can't use discounts, we can't offer discounts to people for like, frequent flying, we can't offer discounts to, like, potential new customers who've never shopped with us.

C (23:53)

Before. Consumers will be harmed too, as.

B (23:56)

Many businesses are likely to forego beneficial.

A (23:58)

Algorithmic pricing altogether rather than recite a.

B (24:02)

State mandated script that wrongly requires them.

A (24:05)

To disparage their own.

B (24:06)

Offer. And it seemed to have worked because surveillance pricing, as you know, is still allowed in New York State. And all the law does is say that you have to tell people that they're doing this. And even that disclosure law, like, did have some challenges. I believe there was a lawsuit from the National Retail Federation basically saying this violates our free speech rights. Right. Like they're saying that this is compelled speech. You can't make us say these things as a government.

A (24:38)

Entity. Algorithms are created by humans, not computers. And they are an extension of what retailers have done for decades, if not centuries, to use what they know.

B (24:48)

About their customers to serve them.

A (24:50)

Better. It's just done at the scale of the modern.

B (24:53)

Economy. I believe that lawsuit was thrown out or dismissed because I know that it didn't, like, get very far. But it is interesting that a lot of companies are trying to fight back against this law even after already getting it down to the level that it currently is.

C (25:11)

At.

A (25:15)

All. Right? So for the moment, it looks like we will not get meaningful regulation. We will not get a ban, we will not get something that says instacart, you can't do this. Is there anything we can do to avoid surveillance.

B (25:28)

Pricing? I think it's really difficult to try to avoid surveillance pricing as an individual person. Right. I think that's why the Consumer Reports study was so fascinating, because whenever I've thought about this, it would be the only way that you would know that you were subject to surveillance pricing is if you were comparing your prices with other people at the exact same time for the exact same product. And Consumer Reports went and did it with 400 plus different participants doing it. And I think that if you as an individual are trying to do it, it's the least you need is like two phones to figure it out. And then even then it's not like you're able to change the pricing on your own because you don't know why each device or each person is getting a different price. Right. Where even in the Consumer Reports study, I don't think at least they were able to pinpoint a reason why the prices were different, just that they were so it's hard to combat that when you don't know what the causes of it are. And in most cases you don't even know that it's.

C (26:33)

Happening.

A (26:33)

Happening. Politico's alfred ng, miles bryan and dustin desoto produced today's show. Julie meyers is our editor. Patrick boyd and david tadashore engineered and laura bullard check the facts. I'm noel king. It's today expl.