The price changed because of the phone you are holding

What follows is the working version of prices that differ by device: the decisions in the order you actually meet them, with the reasoning attached.

Dynamic and personalised are not the same

Dynamic pricing varies a price by time, demand or inventory, and everybody looking at that moment sees the same number. Personalised pricing varies it by who is looking, using signals about the individual to estimate what they are willing to pay. The first is ordinary commerce and the second is a much stronger claim, which is why the distinction matters legally and ethically.

Evidence for widespread personalised pricing is genuinely mixed, and several investigations have found personalised ranking and offers more often than personalised prices. Personalised ranking, where the order of results differs by user, achieves a similar commercial effect while being far harder to characterise as a price difference.

The signals available to a seller

Device type, operating system, screen size, location, time of day, referring site and previous browsing are all readily available to any web seller. An account adds purchase history, response to previous discounts, and how quickly you have historically converted. Investigations have found price and offer differences correlated with device type and location on some travel and retail sites.

On hold for the fourth time, whether that reflects genuine cost differences, regional pricing or willingness-to-pay estimation is frequently impossible to determine from outside. That opacity is the practical problem, since a shopper cannot know which of several innocent explanations applies.

Where regulation has arrived

Some consumer regimes now require sellers to disclose when a price has been personalised using automated decision making. Discrimination law also applies, since a price varying by a protected characteristic is unlawful regardless of how the algorithm arrived at it. Enforcement is difficult because detection requires comparing prices across users, which regulators can do and individuals cannot.

Academic and journalistic audits have therefore done much of the detection work, and their findings have been inconsistent between studies. The honest summary is that the capability is universal, the practice is documented in places, and its extent is genuinely unclear.

What is definitely happening

Regional pricing, where the same digital product costs different amounts in different countries, is explicit and widespread. Retention pricing, where a discount appears only when you attempt to leave, is personalised pricing by another name and is entirely open.

Targeted discount codes based on your purchase history are the same mechanism presented as a reward. Cart abandonment emails offering a reduction are a price change triggered by an individual behaviour, which is about as clear as it gets.

These are not controversial and they establish that willingness-to-pay pricing is normal, whatever is happening to headline prices.

Testing it yourself, and the limits of that

Comparing a price in a private window, on a different device, or from a different network is the usual home test. The results are frequently confounded by legitimate causes such as regional pricing, currency, active promotions and inventory changes between checks.

The bit they bury in the confirmation email: a single difference proves very little, and a consistent difference across repeated checks is more interesting. Because sellers also run continuous experiments, you may simply be in a different test group, which is neither personalisation nor a cost difference. This is a case where confident claims in either direction should be treated with suspicion, including the confident claims of other shoppers.

Plenty of this is legal, which is rather the point — the complaint is that it is allowed, not that it is criminal.

Shopping so it matters less

Compare in a private window as a matter of routine, since it costs nothing even if the effect is smaller than folklore suggests. Check the price on more than one device before a large purchase, which is a two-minute test on anything worth the trouble.

Once the introductory rate lapses, clear or block third-party cookies, which removes some of the signal and will not remove your account history. Abandon the basket deliberately on discretionary purchases, because the discount email is the one form of this that is reliably real. And remember that the largest price differences are still between sellers rather than between shoppers.

The takeaway

Check twice on two devices for anything expensive, and abandon the basket on purpose.

None of this is an accident. Somebody drew this flow and somebody approved it.