Why Cereal Prices Can Differ at the Same Store

Every inventory system I have ever worked in stored the price the same way: as a field attached to the product. Article number, description, unit of measure, price. If two customers bought the same article on the same day, the system pulled the same number, because there was only one number to pull.

That is not how grocery delivery apps have been working.

In December 2025, Consumer Reports, Groundwork Collaborative, and More Perfect Union published the results of a months-long investigation into Instacart’s pricing. They recruited 437 shoppers across four US cities and had them load identical baskets from the same Safeway and Target locations, at the same time, and then recorded what each person was shown.

Nearly three-quarters of the items came back at more than one price. https://americankitchentales.com/shrinkflation-is-real-it-is-also-not-why-your-grocery-bill-doubled/

An 18-ounce box of Signature SELECT Corn Flakes at a Safeway in Washington, DC was offered at $2.99, at $3.49, and at $3.69. Same box. Same store. Same moment. The top price was 23% higher than the bottom one.

I have read a fair number of pricing reports and this one bothered me more than most, because there is no operational story that explains it. A shipping surcharge has a reason. A regional price gap has a reason. This is the same box on the same shelf.

What the numbers actually showed in store

Across the items where the researchers found price experiments running, the average spread between the cheapest and priciest version of an item was 13%. Twenty-three percent was the ceiling, not the norm, and I want to be fair about that, because “up to 23%” is the number that traveled and it is not the typical experience.

The basket-level effect is smaller and, to me, more useful. Total basket prices for identical carts varied by about 7% on average.

Look at what that meant in practice. A basket at a Safeway in Seattle came out at $114.34 for some shoppers, $119.85 for others, and $123.93 for others. At a Target in North Canton, Ohio, the same cart produced five different totals: $84.43, $84.81, $84.92, $87.91, and $90.47.

Eight percent of the shoppers in the test got the lowest basket total. The other 92% would have paid more for exactly the same groceries.

Extrapolated across a year, the researchers put the cost to a household of four at roughly $1,200. That figure assumes you keep ordering the same way and keep landing on the average spread, so treat it as an order of magnitude rather than a bill. It is still not a rounding error.

The software behind it was Eversight, a pricing tool Instacart owns and also sells to other retailers.

Why this is different from a sale

A sale is public. It is printed on the shelf, it applies to everyone who walks up, and you can decide whether it is worth changing your plans for. Same with a loyalty discount: you know you are in the program, you know roughly what you traded for it.

Personalized price experiments break the part of shopping that lets you compare at all. If the number on my screen is mine alone, I cannot tell whether $3.69 is high. There is no reference point. In a physical store, the person next to me is looking at the same tag, which sounds trivial until it is gone.

The researchers also found evidence of price experimentation running through Albertsons, Costco, Kroger, Safeway, Sprouts, and Target on the platform.

Instacart’s position has generally been that retailers set their own prices on the platform. That is a real part of the picture and it matters for who is responsible. It does not change what shoppers were shown.

The part that surprised me: it stopped

On December 22, 2025, about two weeks after the report came out, Instacart announced it was ending the program that produced different prices for different shoppers. Consumer Reports confirmed the change.

The company said retail and brand partners could still test promotions and discounts on the platform, which is a meaningful carve-out and worth reading twice.

Then the rest of the machinery moved. Reuters reported that the FTC sent Instacart a civil investigative demand about its algorithmic pricing software. New York’s attorney general sent a letter demanding details. The House Committee on Oversight and Government Reform wrote to the company on March 5, 2026, raising the question of whether pricing software sold to many retailers at once could coordinate prices across them.

On August 4, 2026, a Senate Judiciary subcommittee held a hearing on AI surveillance pricing. This is not a partisan file, for what it is worth: the House bill introduced in May 2026 to ban surveillance pricing at grocery stores and delivery platforms was introduced by a Democrat with a Republican co-lead.

The law is moving faster than I expected

By April 2026, more than 40 bills across at least 24 states dealt with surveillance pricing in some form. A few have landed.

Maryland’s Protection From Predatory Pricing Act was signed in April 2026 and takes effect October 1, 2026, restricting food retailers and delivery services from using personal data to set individualized prices. New Jersey’s Fair Price Protection Act was signed July 23, 2026, though it does not take effect until August 1, 2027. Connecticut has also enacted a version.

New York took a different route: rather than banning the practice, it requires a disclosure. A retailer setting prices with an algorithm based on your personal data has to tell you so on the price itself.

The New Jersey law carves out loyalty programs, bona fide discounts, and price differences that reflect genuine cost differences. That is the hard part of writing these rules. A coupon and a targeted price are both “you get a different number than someone else,” and the line between them is not obvious in a statute.

There is also a piece of the New Jersey law I keep thinking about: a moratorium on electronic shelf labels starting February 1, 2027. Digital shelf tags are the mechanism that would bring changeable pricing into physical stores. Lawmakers noticed.

What I do with this

Four things, and none of them are clever.

I check the app price against the store’s own price before ordering anything big. Not every retailer marks up on delivery platforms, and some display the shelf price directly, but you will not know which is which without looking once.

I use pickup over delivery when the trip is short, since pickup usually avoids the delivery fee layer and gets you closer to shelf pricing at many retailers.

For the basics I buy every week, I go into the store. This is the whole point of the study for me. A physical shelf tag is a public number. That property is not nostalgia, it is the thing that makes comparison possible, and it is the one part of grocery pricing that is still hard to personalize.

And I pay attention to whether my state has passed anything, because the answer determines whether a delivery app owes me a disclosure or nothing at all. That patchwork is moving month to month right now.

The corn flakes number stayed with me because of how small it is. Seventy cents. Nobody storms out of a store over seventy cents, and nobody notices it on a receipt with thirty other lines. That is exactly why it works, and it is the same reason the shrinking cereal box works. The grocery industry has gotten very good at charging you more in amounts too small to argue about.

source

https://groundworkcollaborative.org/work/instacart/ https://www.regulatoryoversight.com/2026/08/new-jersey-bans-surveillance-pricing-for-groceries-and-other-products/