Your abandoned-cart discount is now a pricing disclosure question

Almost every e-commerce site shows some shoppers a different price from others. A first-visit popup, an abandoned-cart email with 10% off, a loyalty tier, a returning-visitor offer fired by an app nobody has opened in a year. None of that felt like a pricing policy when it was set up. As of 2026 it is one, and the FTC has now published a proposed framework for when it becomes a deception claim.

What the FTC actually proposed

On 19 August 2026 the FTC sought public comment on a proposed enforcement policy statement covering "personalized pricing", which it describes as using personal data to set prices according to what a company believes an individual consumer is willing to spend. The comment period was extended by seven days, closing 25 September 2026. It is a proposed statement of enforcement intent, not a rule, and it follows the Commission's 2024 Section 6(b) surveillance pricing study and its April 2026 advance notices of proposed rulemaking.

It sets out four ways personalized pricing could give rise to liability under Section 5 of the FTC Act:

1. Representing or implying that a price is static or widely offered when it is in fact personalized.
2. Failing to disclose that a price has been personalized.
3. Misleading consumers about the reasons a price was personalized.
4. Collecting, using or sharing personal data for personalized pricing without adequate notice or consent.

Where a disclosure is needed, the proposal says it should be clear and conspicuous and should cover the fact that the price is personalized, the basis for the personalization and the types of data used. The statement gives an example in the other direction too: a clear and accurate disclosure that a price is based on a shopper's previous purchases from the same retailer under the same login would likely be enough to dispel any reasonable expectation that the posted price was not personalized.

Why this reaches further than "surveillance pricing"

The phrase most coverage uses is surveillance pricing, which conjures airlines charging more to people on expensive phones. That framing is why most marketing teams have concluded this is not about them.

Read the proposal against a typical conversion stack, though, and the gap is obvious. The Commission does not specifically address coupons, strike-through pricing, loyalty discounts or similar incentives, and there are no explicit carve-outs for them. Its definitions and theories are not limited to price increases either. That combination is what has practitioners flagging targeted discounts, abandoned-cart offers, loyalty pricing and geofenced promotions as unresolved questions rather than obvious safe harbours.

One boundary is drawn clearly: sorting consumers into broad groups that retailers have long targeted for promotions, such as students or seniors, is generally not treated as personalization. That is a useful anchor, and the distinction underneath it is the one worth internalising. A discount defined by a category anyone can join is different from a price computed from what a model thinks you personally will tolerate.

Where your own tactics sit

This table is our reading of how common conversion tactics line up against the proposal and the state laws already in force. It is a starting point for a conversation with your own counsel, not legal advice, and the middle column is genuinely unsettled.

The three-question test

When a new tactic comes up, run it through this before it ships:

1. Do two shoppers see different prices for the same item at the same moment? If no, stop; this is not the issue. If yes, continue.
2. Is the difference driven by data about the individual, rather than by a condition anyone can meet, such as holding a code, buying more units or joining a published tier? If it is an open condition, you are probably in promotion territory. If it is data about the person, continue.
3. Would a reasonable shopper assume the displayed price is the same one everyone sees? If yes, that expectation is the thing the proposal is concerned with, and a clear disclosure of the fact, the basis and the data used is the remedy it describes.

Question two is the one that does the work. It is also the one that separates a tactic you can defend from one you cannot explain, which turns out to matter commercially as well as legally.

What is already law, and where

The FTC statement is proposed. Two state-level developments are not.

Do not let the California entry confuse the issue, as a lot of vendor marketing has. AB 325 is an antitrust law about competitors coordinating through shared algorithms. It is not a personalized-pricing disclosure law, and a business can be entirely clear of one and exposed on the other.

Does disclosing a personalized price kill conversion?

It depends entirely on what the price is based on, and this is where the compliance question and the conversion question converge on the same answer.

A discount a shopper can explain to themselves strengthens the offer. "You earned 10% at silver tier", "15% off your first order", "save 20% when you buy three" all disclose the basis, and none of them prompt the dangerous question. A discount whose basis cannot be stated comfortably prompts it immediately: if this price was set from my data, what would someone else have paid? Once a shopper asks that at checkout, you have not lost a conversion, you have lost the trust that makes the next one cheaper.

Which suggests a practical reading. If a tactic's basis is one you would be happy to print next to the price, you are in good shape on both counts. If it is not, the tactic is probably not one you want, independently of what the FTC finalises. That is the same conclusion the dark patterns enforcement wave pushed retailers toward, and the same one behind the all-in pricing laws: the practices that survive scrutiny tend to be the ones that were converting on substance rather than surprise.

A six-point audit for this quarter

1. Inventory every price-varying mechanism across your site, email, ads and apps, and name the data each uses. Include anything installed and forgotten.
2. Classify each one using the table above: open condition, broad category, or individual data.
3. Check the New York exposure specifically if you ship there. Disclosure is already required for algorithmic pricing driven by personal data, and enforcement begins with a cure period, which only helps if someone is watching for the letter.
4. Reconcile your privacy policy with reality. The fourth liability theory concerns collecting and using personal data for pricing without adequate notice. If your policy does not mention pricing as a use of data and your stack uses it that way, that gap is the finding.
5. Rewrite grey-area offers around open conditions where you can. An abandoned-cart discount can often become a stated first-order or minimum-basket offer with the same commercial effect and none of the ambiguity, and it usually tests at least as well. Our cart abandonment recovery stack covers the alternatives that do not rely on a personal discount at all.
6. Put the decision in writing and review it when the FTC finalises. This is moving, and a documented position beats reconstructing your reasoning a year later.

What to expect next

Comments close on 25 September 2026. A final enforcement policy statement would follow, on the Commission's own timetable, and it would still be a statement of enforcement intent rather than a rule with its own penalties. The immediate exposure for most retailers is state law and the FTC's existing Section 5 authority, which has never required a new rule to reach a deceptive omission.

The trend behind all of it is stable enough to plan around: pricing that a shopper cannot see the logic of is being pushed toward disclosure, and personalization that a business cannot explain is becoming a liability rather than an edge. Teams whose conversion programme rests on offers with stated conditions have very little to change here. Teams relying on invisible variation have a quarter of work and a strategy question.

This article describes a regulatory development for marketing planning purposes. It is not legal advice, and pricing compliance decisions should be taken with your own counsel.

Frequently asked questions

What did the FTC propose about personalized pricing in 2026?

On 19 August 2026 the FTC sought public comment on a proposed enforcement policy statement on personalized pricing, meaning the use of personal data to set prices according to what a company believes an individual consumer is willing to spend. The comment period was extended to 25 September 2026 under docket FTC-2026-1057. The statement identifies four routes to liability under Section 5 of the FTC Act: implying a price is static or widely offered when it is personalized, failing to disclose personalization, misleading consumers about why a price was personalized, and collecting or using personal data for pricing without adequate notice or consent.

Is an abandoned-cart discount considered personalized pricing?

It is unsettled. The FTC's proposed statement does not specifically address coupons, strike-through pricing, loyalty discounts or similar incentives, and it contains no explicit carve-out for them, which is why practitioners have flagged abandoned-cart offers, targeted discounts, loyalty pricing and geofenced promotions as open questions. A discount available on an open condition anyone can meet, such as a first order or a minimum basket, sits on safer ground than one computed from an individual's behaviour.

Does any law already require disclosing personalized prices?

Yes. New York's Algorithmic Pricing Disclosure Act took effect on 10 November 2025 and requires a clear, contemporaneous disclosure reading "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA" where algorithmic pricing uses personal data. Civil penalties run up to $1,000 per violation, and the state Attorney General must issue a cease-and-desist with a cure period before enforcing. The law requires disclosure rather than banning the practice.

Is California's AB 325 a personalized pricing law?

No, and the two are frequently confused. AB 325, effective 1 January 2026, is an antitrust measure targeting "common pricing algorithms" that use competitor data in anticompetitive agreements, and it lowers the pleading threshold for Cartwright Act claims. Personalized pricing in California has been pursued separately, including an investigative sweep into surveillance pricing opened by the California Attorney General under the CCPA in January 2026.

Are student, senior or loyalty discounts affected?

Sorting consumers into broad groups that retailers have historically targeted for promotions, such as students or seniors, is generally not treated as personalization. Published loyalty tiers with stated thresholds sit on similar ground, because the discount is earned through disclosed behaviour rather than inferred willingness to pay. The risk rises as the basis for the price moves from an open condition toward a model's estimate of what a specific person will pay.

Does disclosing a personalized price hurt conversion?

It depends on the basis for the price. Disclosures a shopper can make sense of, such as a loyalty tier, a first-order offer or a volume discount, tend to strengthen the offer because they explain themselves. A price whose basis cannot be stated comfortably prompts the shopper to wonder what someone else would have paid, which damages trust at the worst possible moment. A practical test is whether you would be happy to print the basis for the price next to the price itself.

The takeaway

Nothing on your site has to change this week, because the FTC statement is proposed rather than final. What does need to happen this quarter is the inventory: every mechanism that can change what a given shopper pays, and the data behind each one. Most teams cannot produce that list today, which is the actual finding. Then move the grey-area offers onto open conditions anyone can meet, check New York exposure if you ship there, and make sure your privacy policy admits that data informs pricing if it does. The tactics that survive this are mostly the ones that were converting on a stated reason anyway.