Personal Money · Saturday, 29 August 2026

01 Briefing what happened

Two shoppers, one shop, one minute - and two different prices for the same eggs

Personal Money 1 min 48 sources

A posted price used to be the same for everyone because the seller could not tell buyers apart. Data ends that, and the poverty premium shows the older version has been running for years.

20%

gap between two shoppers

same eggs, same store, same moment [1]

£493

extra a year in Peckham

paid by low-income households for identical goods [8]

$10,000

Maryland's fine per breach

from 1 October, and $25,000 for repeats [5]

52%

of Canadians would ban it

once the practice is explained to them [10]

The lead story — what happened

Nearly 200 volunteers checked prices on 20 grocery items on Instacart in four US cities. [1]
At one Safeway in Washington DC they picked the same eggs at the same moment. Some were offered $3.99. Some were offered $4.79. [1]
The same test found Skippy peanut butter at $2.99 or $3.59 in an Ohio Target, and Wheat Thins at $3.99 or $4.89 in a Seattle Safeway. [1]
Charging each buyer the most they will pay is called first-degree price discrimination, and it is only possible when the seller knows the buyer. [2]
It is not the same as prices that move for everyone at once. A European Parliament study defines the personal kind as different prices for identical things at the same time, based on what the seller knows about the customer. [14]
Online the seller knows a great deal about the buyer, and the buyer knows almost nothing about how the price was set. Brookings calls that a one-sided bargaining situation. [11]
The US Federal Trade Commission studied the practice in January 2025 and found retailers using precise location, shopping history, demographics and browsing history to set prices. Its then-chair Lina Khan said the tailoring went down to a person's mouse movements on a webpage. [3]
The tracking is not only online. Shops use cameras and Bluetooth signals from your phone to see how long you linger and what you look at. [3]
Maryland became the first US state to ban it in groceries. The law was signed in April, takes effect on 1 October, and carries fines of $10,000, rising to $25,000 for repeats. [4][5]
The Maryland law bans raising a price with personal data. It does not stop lowering one. A shop can raise its prices for everyone and then hand out individual discounts, says Tom McBrien of the Electronic Privacy Information Center. [4]
Colorado's bill went further and covered wages as well as prices. Governor Jared Polis vetoed it in June, saying it would punish differentially lower prices, not just higher ones. It was his second veto of a bill like it in twelve months. [6]
JetBlue is facing a proposed class action in Brooklyn federal court. It followed a public reply telling a passenger flying to a funeral to try clearing their cache and cookies. JetBlue says it does not use personal data or artificial intelligence to set fares. [7]
None of this is needed for the oldest version. In Peckham, south London, low-income households pay an average of £493 a year more than wealthier ones for identical goods, and over £600 in the worst-hit parts of the capital. [8]
The gap is built from ordinary charges. Prepayment meter users pay £129 a year more than direct-debit customers. Drivers in deprived postcodes pay £153 more for motor insurance. Food is the single largest driver, because 39% of affected families shop at convenience stores rather than supermarkets. [8]
Insurers have a term for the same idea. Price optimisation works out the highest price a customer will tolerate before they switch. [9]

What is pushing on this

What the seller knows about you High↑

browsing history, location, device and mouse movements all feed the price [3]

What you know about the seller Easing→

online there is no price tag to compare, only the price on your screen [16]

State-level pushback Building↑

Maryland has banned it in groceries, and a privacy lawyer counts bills in about 33 states [5]

Federal appetite Steady→

the FTC told retailers they must disclose personalised offers, while its chair has called the earlier study a rush job [12][6]

Who is involved

Instacart — says it has never personalised prices, and ended item price tests after the investigation Federal Trade Commission — studied the practice, then told firms they must disclose it Maryland and Colorado lawmakers — one ban signed, one wider bill vetoed Fair By Design and the University of Bristol — measured the version of this that needs no algorithm

How it unfolded

Jan 2025 the FTC publishes its study of pricing built on personal data [3]
Dec 2025 volunteers find different prices for the same items on Instacart [1]
Apr 2026 Maryland signs the first state ban; JetBlue is sued over its fares [4][7]
Jun 2026 Colorado's governor vetoes a wider bill covering prices and wages [6]
1 Oct 2026 Maryland's ban takes effect [5]

Where this points

Watch whether any large retailer is ever shown setting a different starting price by shopper. Every company asked so far says it does not. Researchers say the data to do it is already sitting there. [15]

The rest of the day

29 more stories on this beat.

Each with its own sources. None of these is a link to the story above.

02
Amazon's prices moved under school districts

Denver's school district placed two orders for bulk dry-erase markers on the same day in August 2023, and paid $114.52 for one and $149.07 for the other. Across 55,000 repeat purchases by 23 public bodies, the buyers could have saved about 17% at the platform's own lowest prices. [17]

Why it matters — A public budget is spent by people who cannot watch a price all day, which is exactly who a moving price costs most.
03
Canada's parliament votes down a ban

MPs defeated a motion on 15 April to prohibit pricing set from personal data. A poll by Abacus Data in March found 52% of Canadians would ban it once it was explained, and another 31% would regulate it more tightly. [10]

Why it matters — Public opinion and the law are moving at different speeds, and the practice is spreading in the gap.
04
In fashion, waiting can cut the price

Prices on items left in a cart at one large clothing retailer moved repeatedly over several days, and in some cases waiting produced a discount of up to 17%. [18]

Why it matters — The same tool that punishes urgency rewards patience, which is why it is not simply a price rise.
05
Readers on a rule that did not reach them

Guardian readers writing about renewals pointed to the regulator's ruling that insurers cannot charge existing home and car customers more than new ones, and said insurers carried on anyway. One reader was paying £424.40 a year for breakdown cover and bought cover elsewhere for £101.40. [19]

Why it matters — It is the same mechanism as the algorithm, running on nothing more than the assumption you will not check.
06
State Farm sends back $5 billion

The insurer began paying a record $5bn dividend to eligible car insurance customers on 31 July. It follows a J.D. Power finding that 57% of US drivers re-shopped their cover in 2025, a record. [20]

Why it matters — Prices set on how likely you are to leave move when people start leaving.
07
A 20-year swipe-fee fight settles

Visa and Mastercard settled with merchants over the fees shops pay on every card payment. The average fee is about 2.33%, and US businesses paid well over $100bn in a year. [21]

Why it matters — The fee never appears on your receipt, because it is inside the shelf price everyone pays.
08
The rule that made every card equal is broken

The settlement ends the requirement that a shop accepting one Visa credit card must accept all of them. Fees will likely be layered by card, and some businesses may refuse the most expensive ones. [22][23]

Why it matters — The average shopper carries about four cards, and they are about to stop being interchangeable. [23]
09
Where card points actually come from

Interchange fees run at roughly 1.5% to 3.5% of each purchase, and they fund the rewards. When the US capped interchange on debit cards, debit rewards programmes essentially disappeared. [24][25]

Why it matters — Interchange can feed into what a shop charges, and some shops now add a card fee of about 2.5% instead of absorbing it. [25][23]
10
Britain's bills all moved on the same day

Uswitch put the average annual increase in combined UK household bills at £214.10 from 1 April. Council tax was more than half of it, with a typical band D bill in England up £111 to £2,392, and water up £33 to £639. [26][27]

Why it matters — Nothing about April makes things cost more; it is simply the date most of the annual rises are set for.
11
The energy cap fell, and then the war moved it

The price cap dropped 7% in April, cutting a typical annual bill by £117 to £1,641. Cornwall Insight then forecast £1,929 from July, a rise of 18%. [28][27]

Why it matters — The saving came from moving some charges off bills and onto general taxation, which is a different pocket rather than a smaller cost. [28]
12
1.7 million homes sit outside the cap

About 1.7 million households in England and Wales heat with kerosene, which the energy price cap does not cover. Their bills doubled after the strikes on Iran, some orders were cancelled, and 62.5% of homes in Northern Ireland rely on it. [29]

Why it matters — A protection defined by the fuel you use leaves whoever uses a different fuel entirely exposed.
13
US electricity is up 40% since 2020

Electricity prices have risen 40% since February 2020, against 26% for the overall cost of living. Demand from new data centres, retired power plants and gas prices are all named as causes. [30]

Why it matters — It is the clearest case of a household bill rising because of something happening in an industry the household has no part in.
14
Who pays for the grid the data centres need

Data centres are putting pressure on the grid and on electricity costs. That has reached national, state and local leaders, and the people living near them. The Trump administration announced a Ratepayer Protection Pledge. [43]

Why it matters — The question is not whether the power gets built but which bill it lands on.
15
Cooling a home this summer costs 8.5% more

The average June-to-September electricity cost was forecast to rise 8.5% to $778. In the region covering Florida, Georgia and the Carolinas it was forecast at $860, up 13.5%. [45]

Why it matters — Summer cooling is not optional in those states, which is what makes the rise land unevenly.
16
Electricity costs three times more in some states

Prices in the lower 48 states ranged from 8 cents a kilowatt-hour in North Dakota to 27 cents in California in 2024. A typical home uses about 900 kilowatt-hours a month. [44]

Why it matters — The same appliance costs three times as much to run depending only on where it is plugged in.
17
One in twenty US homes near utility collections

Nearly one in 20 US households risked having utility debt sent to collections going into winter, and severely overdue utility debt rose 3.8% in six months. Average US energy bills exceed $260 a month, and top $300 in some north-eastern states. [39]

Why it matters — A bill you cannot pay stops being a bill and becomes a credit record.
18
Trump's promise and the actual prices

Average US electricity prices rose about 6.7% over the past year and natural gas about 10.8%, against a campaign promise to halve energy prices. [46]

Why it matters — Energy prices answer to weather, exports, infrastructure and demand, so a pledge to halve them was never a lever anyone held. [46]
19
A bank got a third of fraud calls wrong

The Financial Ombudsman found Monzo had wrongly rejected 34% of the 3,372 fraud and scam complaints referred to it last year, the worst rate among the most-complained-about UK banks. NatWest was second on 33% and HSBC third on 32%. [31]

Why it matters — Fraud is now an estimated 41% of all crime in England and Wales, so the appeal rate is not a small corner of banking. [31]
20
A 13-month clock on reporting a scam

UK rules from October 2024 require a refund within five working days, up to £85,000, but the bank must be told within 13 months of the last payment. One woman was told she would get £1,000 of £20,000 stolen, and was refunded in full a day after journalists asked. [32]

Why it matters — National Trading Standards wants the clock to start when a victim realises, because investment fraud can run for years before anyone knows. [32]
21
Australia would auto-refund only under $3,000

A proposed scheme would automatically reimburse verified scam losses below A$3,000. Australia had 481,523 scam reports worth A$2.18bn last year, with a median loss of $400. [33]

Why it matters — Most complaints sit under the threshold, but most of the money lost sits above it. [33]
22
The supervisor's own view of the fraud problem

The Federal Reserve's vice chair for supervision, Michelle Bowman, told a Washington symposium in May that fraud is as old as money but now differs in scale, sophistication and impact. [42]

Why it matters — It puts a bank supervisor's name to the claim that the defences have not kept up.
23
The poverty premium costs one region £241m

Low-income households across the West Midlands pay an extra £241m a year for essentials. That is the second worst-affected region in Britain after London, on research for Fair By Design by the University of Bristol. [34]

Why it matters — One Birmingham mother described buying a cheaper fridge freezer she could afford and then paying for it in electricity. [34]
24
A quarter of people in Wales skipping meals

Around a quarter of people in Wales have cut down or skipped meals to make ends meet, the Bevan Foundation reported. Disabled people and families with children are under most pressure. [35]

Why it matters — It is the point where a household stops absorbing higher prices and starts absorbing less food.
25
36% of Stoke's children in low-income homes

A report on Stoke-on-Trent found 18,989 children, or 36%, in low-income households, up from 25.7% a decade earlier, with Universal Credit claims up 158% since before the pandemic. Its lead researcher called it a humanitarian crisis. [48]

Why it matters — Every extra cost in this edition lands hardest on exactly this population.
26
Detroit's older homeowners and the fixed income

Nearly half of Detroit residents over 65 spend at least 30% of their income on housing costs, even as property values in the city fall. [36]

Why it matters — An asset that is worth less can still cost more to keep, and a fixed income cannot absorb either.
27
Poverty numbers changed when the measure did

Northern Ireland's relative poverty rate for 2023 fell from 18% to 14% when officials switched from surveys to benefits records, and pensioner poverty from 12% to 7%. UK child poverty figures are expected to be revised down for the same reason. [41][47]

Why it matters — The Joseph Rowntree Foundation said this is a more accurate measurement of incomes, not a real fall. [41]
28
The average savings account pays 0.63%

Bankrate's survey of more than 500 banks and credit unions put the US national average savings rate at 0.63% in August 2026, while the best high-yield accounts paid around 4%. [37]

Why it matters — The gap is not a market rate; it is what a bank pays customers who do not move.
29
The same data is also sold to the government

Data brokers buy phone and browser data to sell to advertisers, and sell the same information, including bulk phone location data, to police forces and federal agencies without a warrant. [38]

Why it matters — The profile that sets your price has more than one buyer.
30
Jet fuel up more than 80%, fares next

Aviation kerosene rose from $830 a tonne to more than $1,500 after strikes on Iran disrupted Gulf supply, which provides about half of Europe's imports. Wizz Air warned of a €50m hit to annual profit. [40]

Why it matters — It is the cleanest example of a price that moves for reasons nothing about the passenger explains.

02 Lesson why it matters

The haggle never ended. It moved somewhere you cannot see it.

One price for everyone was never fairness. It was what a seller settled for while they could not tell buyers apart.

The twist

Haggling never ended. It moved to a room where only the seller can see the other side of the table, and your side is made of your own browsing history.

How it works

A seller would like a different price from each buyer
But they cannot tell buyers apart, so they post one price
That single price leaves money with whoever would have paid more
Data now tells the seller who those buyers are
The price moves toward what each person will pay
And the money that used to stay with the buyer goes the other way

Where you've seen this

A job offer

the employer knows the whole pay band and you know one number

A car forecourt

the dealer knows what your trade-in is worth before you say a word

An insurance renewal

the quote is set on how likely you are to move, not on your risk alone

A landlord and a moving date

whoever knows the other's deadline sets the terms

The catch

Different prices are not automatically worse. Student fares, senior discounts and cheap off-peak seats are the same machinery, and they exist because someone can tell buyers apart. The line people actually object to is between what you can afford and what you cannot avoid.

And the whole of it

Everyone sees one price and takes it for the price. The person beside you saw a different screen, and neither of you can see the other's. The same technique reads employees as well as shoppers, which is why Colorado's vetoed bill covered wages too. Most of us are on both sides of it, seeing our own number clearly and the whole shape not at all.

03 Lab your turn

The Other Side of the Table

Set one price for six strangers, then set six prices once you can see them, and watch where the difference comes from.

04 Truth what's really going on

What is really going on

A single posted price was never a promise of fair treatment - it was the best a seller could do while they could not tell buyers apart, and the data that ends that ignorance is already collected and sold.

Why it works on us — Everybody sees one number on their own screen and reads it as the price of the thing, because a price tag in a shop window really was the same for everyone who walked past it.

Who gains

Sellers holding the most data on their customers — A price built from browsing history, location and mouse movements collects money that a single posted price left with the buyer. [3][2]
Insurers that price by tolerance rather than by risk alone — Price optimisation sets the premium at the highest level a customer will accept before switching, which charges the least mobile customers most. [9][19]
Card issuers and the two card networks — Interchange of roughly 1.5% to 3.5% on every purchase funds the rewards, and when the US capped debit interchange the debit rewards disappeared. [24][25]
Data brokers — The same phone and browser data that profiles a shopper is sold on to police forces and federal agencies without a warrant. [38]
Some low-income buyers, in the other direction — The same machinery is what makes student fares, senior discounts and targeted markdowns possible at all. [2][13]

Who pays

Anyone buying in a hurry — Urgency reads as willingness to pay. JetBlue told a passenger flying to a funeral to try clearing their cache and cookies. [7][13]
Low-income households in Britain — £493 a year more in Peckham for identical goods, £129 more on a prepayment meter, £153 more on motor insurance in a deprived postcode. [8]
Customers who stay put — One Guardian reader was paying £424.40 a year for breakdown cover and bought cover from another firm at £101.40. [19]
Customers of shops carrying the card fee — The average swipe fee is about 2.33%, interchange can feed into a merchant's pricing, and some shops now add a card charge of about 2.5% instead. [21][25][23]
Public bodies buying on a platform — Denver's schools spent $5.7m with Amazon in a year and could have saved about 17% at the platform's own lowest prices. [17]

What nobody knows yet

Open questions from across today’s stories — ours included.

01

Whether any large retailer is setting a different starting price by person today.

Companies say they are not. Instacart says it has never personalised prices and stopped its item price tests, while volunteers in the Consumer Reports and Groundwork test were offered different prices for the same item at the same store and moment. Researchers quoted by the WSJ say the data to do it is already there. [4][1][15]
02

Whether Delta uses artificial intelligence to set individual fares.

A business professor writing in The Conversation refers to Delta announcing it would use AI to set individualised ticket prices. Two dozen members of Congress asked Delta the same question in November and Delta said no. We cannot settle it from the pool. [13][7]
03

What Maryland's ban will actually stop.

It bans raising a price using personal data but not lowering one, so a shop can lift its prices and hand out individual discounts. The state attorney general is the only party who can enforce it. [4]
04

How much of the poverty premium is deliberate pricing and how much is the cost of serving small, irregular payments.

The research measures the size of the gap - £493 a year in Peckham, £241m a year across the West Midlands - not the intent behind each charge. [8][34]
05

Whether lower swipe fees will reach a shelf price.

The settlement will probably lower future fees on some cards. A small-business columnist writing in the Guardian expects layered card-by-card fees and more confusion at the till rather than lower prices. [21][23]
06

How many scam victims fall outside the UK's 13-month reporting rule.

UK Finance says only a small number of cases do. National Trading Standards wants the rule reviewed and neither side has published a count. [32]
07

Whether poverty in Northern Ireland fell or only the measurement changed.

The rate moved from 18% to 14% for the same year when officials switched from surveys to benefits records, and the Joseph Rowntree Foundation says that is a better measurement rather than a real fall. [41][47]
08

Where the UK energy price cap lands in July.

One forecast puts a typical bill at £1,929, an 18% rise. The chancellor said the government was working through different scenarios and looking at more targeted options. The Gulf disruption behind it is unresolved. [27][29][40]

05 Hope carry this

None of this was visible until somebody counted it. Nearly two hundred volunteers with the same shopping list, and researchers adding up what a poor postcode pays, turned a private suspicion into a number - and a number is something a law can be written about.