Investing/Nov 19 edition/Podcast
Regulation

RegulationRules, enforcement & market structure

Groundwork's Lindsay Owens says algorithmic pricing is replacing the fixed price tag, pointing to Instacart's split-price tests and AI shopping agents

Groundwork Collaborative CEO Lindsay Owens, on Bloomberg's At The Money, argues algorithmic pricing is eroding the fixed price tag, citing an Instacart study showing split prices for identical carts, spreading state and city rules, and AI shopping agents that may serve sellers.

From the Masters in Business podcast episode of November 18, 2026 and cited sources: In a December 2025 study by Consumer Reports, Groundwork and More Perfect Union, 437 Instacart shoppers filling identical carts at the same time saw different prices on nearly three-quarters of items. [3] [4] [13]

SOURCES[3] Engadget: Instacart study engadget.com[4] SiliconANGLE: Instacart ends tests siliconangle.com[13] Spokesman-Review: Instacart spokesman.com

AI-assisted, not human-edited. Methodology Corrections Who owns this site

KEY FACTS
What happened
In a December 2025 study by Consumer Reports, Groundwork and More Perfect Union, 437 Instacart shoppers filling identical carts at the same time saw different prices on nearly three-quarters of items. [3] [4] [13]
Original source
Masters in Business podcast episode, independent perspective
Brief by
AT Pulse research desk AI-assisted; publisher and contact on the desk page
When
November 18, 2026 podcast episode published
Evidence
15 sources across 15 sites; transcript sourced

Reach: not measured yet

Illustration from Masters in Business: Groundwork's Lindsay Owens says algorithmic pricing is replacing the fixed price tag, pointing to Instacart's split-price tests and AI shopping agents
Image: Masters in Business — Original article ↗
THE CORE IDEAS3 TAKEAWAYS
01

In a December 2025 study by Consumer Reports, Groundwork and More Perfect Union, 437 Instacart shoppers filling identical carts at the same time saw different prices on nearly three-quarters of items. The gaps averaged 13% and reached 23%. Instacart called the tests randomized and ended retailer item-price testing on its Eversight tool by December 22. [3] [4] [13]

02

Rules are being added piece by piece. The FTC's junk-fee rule, effective May 12, 2025, requires all-in prices for event tickets and short-term lodging. New York City's click-to-cancel rule took effect October 1, 2026. In Senate testimony, Owens said more than 40 bills on surveillance pricing had been introduced in 24 states and cities. The FTC's January 2025 surveillance-pricing findings were labeled preliminary and approved on a 3–2 vote. [5] [8] [10] [15]

03

Owens warns that AI shopping agents may work for platforms rather than buyers, and she names Walmart's Sparky as an example. In related research, Princeton and University of Washington researchers tested 23 language models. Fifteen of them recommended a pricier sponsored product more than half the time. [1] [14]

WHY IT MATTERS

Documented price splits and a fast corporate reversal show that pricing experiments carry real reputational and regulatory risk.

Read the full assessment

The implication, not yet proven, is that firms using personalized pricing or deploying shopping agents will face growing disclosure and compliance demands.

Executive brief

Lindsay Owens, CEO of Groundwork Collaborative, argues that the fixed price tag is losing ground to algorithmic pricing. Her strongest evidence comes from a study her group helped run. In December 2025, 437 shoppers on Instacart found that nearly three-quarters of grocery items showed different prices to different people, at the same store and the same moment. The gaps averaged 13% and reached 23% (Engadget). About two weeks later, Instacart ended the tests (SiliconANGLE). On Bloomberg's At The Money, Owens also warns that AI shopping agents may serve platforms rather than buyers.

What changed and event timeline

  1. FTC publishes surveillance-pricing findings

    Early findings from documents filed by six pricing intermediaries, including Mastercard and McKinsey, show that location, browsing history and mouse movements feed individualized prices. The commission approved the report 3–2 ().

  2. Senators question Delta's AI fares

    Three senators asked Delta what data it shares with its AI pricing partner Fetcherr (). Delta said its model uses no personal data ().

  3. Instacart study, then reversal

    Consumer Reports, Groundwork and More Perfect Union documented the price splits (). By December 22, retailers could no longer use Instacart's Eversight tool to run item price tests ().

  4. Owens testifies to the Senate

    At a Senate Judiciary subcommittee hearing, she named Target, DoorDash, Expedia and Kroger as users of surveillance pricing. She said more than 40 bills on the issue had been introduced in 24 states and cities ().

  5. NYC click-to-cancel takes effect

    It is the first city-level rule of its kind. Businesses must let people cancel subscriptions as easily as they signed up, and the city can seek penalties and restitution (;).

  6. Owens on At The Money

    Barry Ritholtz interviews her about her book Gouged (Sept 29, 2026). Topics include surveillance pricing, loyalty programs, right to repair and agentic commerce (;).

Capabilities and access

  • Instacart/Eversight: retailers could show different prices to different shoppers for the same item, at the same store, at the same time, then measure how they responded (SiliconANGLE).
  • Delta/Fetcherr: AI pricing covered about 3% of fares, with a target of 20% of the domestic network by the end of 2025 (Mexico Business News).
  • Shopping agents: Owens names Walmart's Sparky as an agent that works for the company, not the shopper (transcript, 23:15).
Read the full section
  • Instacart/Eversight: retailers could show different prices to different shoppers for the same item, at the same store, at the same time, then measure how they responded (SiliconANGLE).
  • Delta/Fetcherr: AI pricing covered about 3% of fares, with a target of 20% of the domestic network by the end of 2025 (Mexico Business News).
  • Shopping agents: Owens names Walmart's Sparky as an agent that works for the company, not the shopper (transcript, 23:15). The transcript does not name the vendors or models behind these systems.

Technical analysis for researchers and developers

  • Detection method: the Instacart study filled identical carts at the same time. Instacart called the tests randomized (Engadget).
  • Data inputs: the FTC lists location, demographics, cart abandonment and browsing as pricing inputs (FTC).
  • Agent evaluation: Princeton and University of Washington researchers tested 23 language models.
Read the full section
  • Detection method: the Instacart study filled identical carts at the same time. It ran five tests with 437 participants at Target and Safeway stores (Spokesman-Review). This finds price differences but cannot tell what causes them. Researchers found no sign that prices depended on income, ZIP code or shopping history. Instacart called the tests randomized (Engadget).
  • Data inputs: the FTC lists location, demographics, cart abandonment and browsing as pricing inputs (FTC).
  • Agent evaluation: Princeton and University of Washington researchers tested 23 language models. All but eight recommended a pricier sponsored product more than half the time (The Walrus).

Claims and evidence

  • Same cart, different price: supported by the Instacart study (Engadget).
  • Loyalty penalty (Starbucks and McDonald's data requests), Kroger's data sales making up 30% of operating income, and Costco capping markups at about 14%: these come only from Owens's interview (transcript, 04:59–21:14).
  • Groundwork revealed the Delta–Fetcherr deal: this is Owens's own claim.
Read the full section
  • Same cart, different price: supported by the Instacart study (Engadget). Groundwork's estimate of about $1,200 a year for a family of four is an extrapolation from that study (Spokesman-Review).
  • Loyalty penalty (Starbucks and McDonald's data requests), Kroger's data sales making up 30% of operating income, and Costco capping markups at about 14%: these come only from Owens's interview (transcript, 04:59–21:14). No independent confirmation was found.
  • Groundwork revealed the Delta–Fetcherr deal: this is Owens's own claim. Delta denies using personal data (AJC).

Context and prior work

  • The FTC's junk-fees rule took effect on May 12, 2025. It requires the full price up front for event tickets and short-term lodging, matching Owens's account (TechCrunch).
  • Owens tells a long history: fixed prices began with the Quakers and spread through John Wanamaker's department store. She also recounts Coca-Cola's 1990s plan for temperature-based vending prices, which was dropped after backlash (transcript, 01:28; 18:48).
Read the full section
  • The FTC's junk-fees rule took effect on May 12, 2025. It requires the full price up front for event tickets and short-term lodging, matching Owens's account (TechCrunch).
  • Owens tells a long history: fixed prices began with the Quakers and spread through John Wanamaker's department store. She also recounts Coca-Cola's 1990s plan for temperature-based vending prices, which was dropped after backlash (transcript, 01:28; 18:48).

Limitations, safety and contested findings

  • Experiments vs. profiling: the Instacart evidence shows randomized experiments, not personal profiling. Owens's "surveillance pricing" label covers both.
  • FTC report: staff called the findings preliminary. Two commissioners dissented, including Andrew Ferguson, now the chair (FTC).
  • "Two studies" on agents: Owens cites two large studies showing agents return higher prices, but she does not name them (transcript, 23:15).
Read the full section
  • Experiments vs. profiling: the Instacart evidence shows randomized experiments, not personal profiling. Owens's "surveillance pricing" label covers both.
  • FTC report: staff called the findings preliminary. Two commissioners dissented, including Andrew Ferguson, now the chair (FTC).
  • "Two studies" on agents: Owens cites two large studies showing agents return higher prices, but she does not name them (transcript, 23:15). The 23-model study is related evidence, not a confirmed match.
  • Uber price splits: her claim that friends ordering the same ride will see different prices is an anecdote, not documented evidence.

Business and practitioner implications

  • Expect fast backlash: Instacart reversed within two weeks under public criticism and an FTC civil investigative demand (SiliconANGLE).
  • Patchwork compliance: state and city laws are spreading, such as NYC's click-to-cancel rule (NYC DCWP).
  • Agent builders: studies of sponsored-product bias suggest agents need clear disclosure of whose interests they serve (The Walrus).
Read the full section
  • Expect fast backlash: Instacart reversed within two weeks under public criticism and an FTC civil investigative demand (SiliconANGLE). Teams running price A/B tests should plan for this.
  • Patchwork compliance: state and city laws are spreading, such as NYC's click-to-cancel rule (NYC DCWP). Groundwork has proposed model legislation, a 12-point "Shopper's Bill of Rights" (transcript, 12:38).
  • Agent builders: studies of sponsored-product bias suggest agents need clear disclosure of whose interests they serve (The Walrus).
FOLLOW THE EVIDENCE

The source trail.

Sources (15)
A LITTLE LESS NOISE. A LOT MORE CONTEXT.

Stay curious.
Follow the evidence.

Independent perspectives, the original sources, and room for the questions that don't have easy answers.

How we build the brief
Connect with us

Find us where you already read.