Say you are late for lunch. You pull out your phone, open your usual food delivery app, and before you even begin searching, the app seems to have already adjusted itself to your situation. The restaurants that can get to you faster now have discounts on the kind of food you usually order, reducing the chances of you closing the app without ordering. Therefore, the path from intent to action feels shorter, almost guided.

At the same time, another app on your phone lights up with a notification. A competing platform is offering a limited-time discount on similar cuisine. You did not open it, but it seems to know enough about your moment to step in.

To some, this feels seamless. To others, it feels like apps know a little too much. How did it decide which restaurants should get discounted for you at that moment? Why were those offers available only on the day you were late? And more importantly, would someone else opening the same app see the same prices and discounts as you?

That discomfort sits at the heart of a growing debate in the digital economy. Platforms are no longer just showing you products. They are learning how much you are willing to pay for them.

This is surveillance pricing.

What is surveillance pricing

At its simplest, surveillance pricing is the practice of using your personal data to form a digital profile that enable platforms to decide what price you should see. Not a general price. Not a sale price. But a price that is personalised for you.

It is built on a fundamental shift in how pricing works. Traditionally, pricing was tied to the product or the market. A kilogram of onions costs what it costs, with fluctuations depending on supply, demand, or competition.

Surveillance pricing breaks that assumption. It asks a different question. Not “what is this product worth?” But “what is this product worth to you?” Essentially, dynamic pricing reacts to the market, while surveillance pricing reacts to the individual.

How the system learns to read you

Before a platform can adjust a price, it has to understand the user. That understanding is built through what is often called dynamic profiling.

Every action you take on a digital platform leaves behind a signal. The obvious ones include your search history, your past purchases, and the categories you frequently browse. But the more valuable signals are often more subtle. How long you linger on a product page, how many times you revisit a listing, whether you abandon a cart and return later, even the device you are using or the time of day you are active.

These signals are rarely used in isolation. They are combined to build a working estimate of who you are as a consumer. Not just your preferences, but your urgency, your flexibility, and your likely price sensitivity.

It is easy to think of this as something that only happens inside apps. But the same systems do not stop at your screen.

If you layer other forms of data, such as your location, it blurs the line between online and offline behaviour of the user. If you are walking past a retail store and have its app installed, a geofenced trigger can push a notification offering an in-store discount, nudging you to step in. If you are in a dense restaurant cluster around dinner time, multiple dining reservation apps may compete for your attention with timed offers.

Even cross-app behaviour begins to matter. Searching for flights on one platform, comparing hotels on another, and checking maps for directions creates a pattern that signals intent across an ecosystem, not just within a single app.

By the time pricing comes into play, the system is not guessing. It is making a calculated decision based on a layered profile.

From understanding behaviour to setting prices

Once a platform has a working model of your behaviour, pricing becomes a strategic lever rather than a fixed attribute.

The objective is not to offer the lowest price available. It is to identify the highest price you are still willing to accept without dropping off.

Take the case of promotions. In a paper presented at the KDD AI Conference in 2025, DoorDash described how it uses machine learning to estimate the “true incremental effect” of promotions on individual users. The company argues that not every customer responds to discounts in the same way.

Some need a deep cut to convert. Others would have ordered anyway. If a company gives everyone the same discount, it loses money on those who did not need it.

Extend that logic slightly, and pricing itself becomes flexible. A user who appears price-sensitive might be shown a discount to ensure conversion. Another user, who seems more likely to complete the purchase regardless, might see the standard or even a slightly higher price.

In some cases, the system does not even wait for explicit signals. A DoorDash patent describes a model that evaluates user “agitation” by analysing interaction patterns such as rapid swiping or erratic navigation, and responds by surfacing targeted offers designed to influence the decision in real time.

When algorithms compete for your decision

One of the more interesting consequences of this system is how different platforms begin to compete for the same moment.

Consider a late-night ride booking. You open one app and see a fare that seems higher than expected. At the same time, another app sends a notification offering a discount on rides in your area. Both platforms are reacting to similar signals, your location, the time, your past usage patterns, and perhaps even the frequency with which you have opened the app in the last few minutes.

From the user’s perspective, this looks like competition. Underneath, parallel profiling systems are attempting to interpret and act on the same recorded user behaviour.

Early signs that this is already happening

There are already instances that suggest how this could play out in practice.

Last year, The Hindu reported that on the quick-commerce platform Zepto, the same fruits and vegetables were shown at different prices depending on the device used. iPhone users were charged more than Android users for identical items.

A kilogram of onions at Rs 43 on Android appeared as Rs 57 on iPhone. The difference was not tied to demand or supply. It was tied to the perceived profile of the user.

Now, this alone does not conclusively prove surveillance pricing. There can be multiple explanations including testing, vendor differences, or pricing errors.

But it illustrates the possibility.

Similarly, patterns observed in travel platforms. Repeated searches for flights on the same route can lead to higher prices over time, as the system interprets this behaviour as a signal of urgency and willingness to pay more.

Why this is not the same as surge pricing

It is important to separate surveillance pricing from something more familiar such as surge pricing.

Surge pricing is visible. You see it during peak hours, bad weather, or high demand. Prices rise for everyone because demand exceeds supply. The logic is market-based and transparent.

Surveillance pricing is different.

It introduces variation at the level of the individual. Two users in the same place, requesting the same service at the same time, may see different prices because the system believes they have different thresholds for spending.

One reacts to the market. The other reacts to your data.

How governments are starting to react

Once you start looking at surveillance pricing closely, the obvious next question is whether anyone is actually trying to stop it.

The answer is yes, but not in the same way everywhere.

In the US, the reaction has been the most direct. Lawmakers have started calling out surveillance pricing as a problem in itself, not just a side effect of data collection. US states like Maryland have already moved to restrict it, with others such as Connecticut and New York pushing similar laws. The concern here is fairly straightforward. If two people are looking at the same product and one ends up paying more simply because an algorithm thinks they can afford it, that starts to look a lot like digital-era price discrimination.

Europe is taking a different route. Instead of going straight after pricing, regulators there are focusing on the system that makes it possible.

Under EU’s General Data Protection Regulation (GDPR), companies already have to explain how they collect and use personal data, especially when that data feeds into automated decisions. If your behaviour is being used to influence the price you see, that starts to fall into that category. On top of that, the EU’s Omnibus rules require companies to tell users when prices have been personalised based on automated decision-making and consumer profiling.

That does not ban the practice, but it takes away one of its biggest advantages, the fact that it usually happens quietly.

China, on the other hand, has gone in a much more direct direction, but with a different framing. According to a report by Sixth Tone, regulators there have already stepped in against what they call “big data price discrimination.” In simple terms, platforms have been warned against charging loyal users more than new users just because they have more data on them.

And then there is India

In India, there is no rule yet that says platforms cannot personalise prices based on user data. But that does not mean there is no regulation that could apply. The Digital Personal Data Protection Act, for instance, puts limits on how companies can collect and use personal data. It does not mention pricing directly, but if user data is being used in ways that people do not understand or expect, it could eventually come under scrutiny.

For now though, most of the awareness is coming from what people are noticing themselves.

Take the example of device-based pricing differences on platforms like Zepto, where iPhone users were reportedly shown higher prices than Android users for the same products. Or the more common experience of prices creeping up after repeated searches. None of these, on their own, fully prove surveillance pricing. But they are enough to make people start asking questions.

India, at this point, is somewhere in between.

The systems that make surveillance pricing possible are already here. The signals are visible in small ways. What is missing is a clear regulatory response that directly addresses it.