Research Note

Online Price Evidence Research Note

NIST describes experimental design as a plan made before data collection so observations can support valid and objective conclusions. Its guidance on blocking treats time

Aug 4, 20262 min readBy Dalton Anderson

Source conversation

THE LOYALTY PENALTY: WHY YOUR FAITHFUL SHOPPING HABITS ARE COSTING YOU MORE

Keywords Instacart, Consumer Reports, dynamic pricing, AI, surveillance pricing, Eversight, customer surplus, price discrimination, grocery staples, algorithm ethics SummaryIn this episode of the Venture Step Podcast, Dalton Anderson dives into a recent Consumer Reports investigation regarding Instacart’s use of AI-enabled dynamic pricing. Dalton explores how the technology, powered by a company called Eversight, aims to reduce "customer surplus" by charging users different prices for the exact same goods based on their willingness to pay. The discussion highlights the ethical concerns of applying airline-style pricing logic to essential items like milk and eggs, the manipulation of discount perception, and the potential for discriminatory profiling against vulnerable populations. Takeaways While society generally accepts variable pricing for airline seats due to supply and demand, applying this logic to essential goods like bread and eggs raises significant ethical questions. A Consumer Reports investigation involving 437 volunteers found that 74% of items selected had different prices for different people in the same geographic areas. The study revealed massive markups, such as a 23% increase for Wheat Thins in Seattle and a 20% price swing for eggs in DC. This pricing strategy is driven by "Eversight," which uses AI to determine the maximum price a consumer will pay, effectively eliminating the savings a customer might otherwise enjoy . Additionally, the illusion of deals is created when algorithms artificially inflate the list price for certain users to make a standard discount appear larger. Brand loyalty can also act as a disadvantage; if an algorithm knows you will always buy a specific brand, it may charge you more for it. Ultimately, there are fears this technology could evolve into discriminatory profiling, similar to failed predictive policing models, targeting users based on personal data and location.

Back to the episode hub

Online Price Evidence Research Note

Research principle

NIST describes experimental design as a plan made before data collection so observations can support valid and objective conclusions. Its guidance on blocking treats time, operator, environment, and other nuisance variables as factors that should be held constant or accounted for.

Applied to an online price comparison, the comparison unit is not merely a product name. It includes the exact product identifier, size, seller, store location, fulfillment method, delivery address or zone, time, account state, membership, coupon eligibility, quantity, inventory state, and displayed fees.

Evidence levels

Level	Record	Defensible conclusion
Anecdote	A remembered difference or cropped image	A question worth checking
Documented observation	Full screenshots and context from two sessions	Two displayed outcomes differed under the recorded conditions
Controlled comparison	Simultaneous matched sessions with known nuisance factors held constant	The difference persisted after the stated controls
Replication	Repeated sessions, products, stores, or accounts	The variation was not limited to one captured event
System evidence	Assignment logs, product rules, data dictionaries, or code	A named input or rule contributed to the outcome

Screenshots are useful for recording output. They do not reveal the assignment rule by themselves.

Consumer record

The FTC advises online shoppers to record the company and website, item, date, amount, seller policy, and communications. It also advises comparing total cost, including shipping, delivery, taxes, and other fees.

For price-variation evidence, the packet should preserve unedited originals, timestamps, URLs, product identifiers, full cart totals, promotion terms, and the account conditions needed to reproduce the view. Public copies should remove names, exact addresses, order numbers, payment details, email addresses, loyalty identifiers, and other account data.

Escalation

Instacart's current help page lets a shopper report an incorrectly priced item through the item detail view. USAGov recommends contacting the seller or website first. If the issue is not resolved, its current page links to state consumer-protection offices, state attorneys general, the FTC for scams or suspicious communications, and econsumer.gov for international purchases.

The appropriate regulator depends on the conduct and jurisdiction. The guide should not promise that a complaint portal will adjudicate a pricing study or retrieve proprietary assignment logic.

Sources

Follow the evidence.

ag.ny.gov: attorney general james demands answers instacart about algorithmic pricingag.ny.gov
itl.nist.gov: pri332itl.nist.gov
instacart.com: promotionsinstacart.com
ftc.gov: ftc surveillance pricing study indicates wide range personal data used set individualized consumer pricesftc.gov
consumer.ftc.gov: online shoppingconsumer.ftc.gov
company.instacart.com: instacartpricingcompany.instacart.com
gov.uk: tackling the loyalty penaltygov.uk
company.instacart.com: instacart makes it easier for customers to save on groceries with acquisition of eversightcompany.instacart.com
instacart.com: 1586544648instacart.com
usa.gov: online purchase complaintsusa.gov
company.instacart.com: ending item price tests on instacartcompany.instacart.com
company.instacart.com: the truth about pricing tests on instacartcompany.instacart.com
itl.nist.gov: pri11itl.nist.gov
fca.org.uk: fca confirms measures protect customers loyalty penalty home motor insurance marketsfca.org.uk
consumerreports.org: instacart ai pricing experiment inflating grocery bills a1142182490consumerreports.org
investors.instacart.com: 9e9aff2c 95db 4f75 bdf1 0f4025e1468cinvestors.instacart.com