The Anonymity Tax

Based on the research ofWang, Jiang and Singh, "Impact of the Invisibles: Personalized Pricing on Platform with Anonymous Users," Information Systems Research, 2026

The Opt-Out Economy Is Already Here

For most of the last decade, "anonymous user" meant a rounding error: a handful of browser-privacy enthusiasts blocking cookies while everyone else clicked accept. That era is over. When Apple made tracking permission mandatory with iOS 14.5's App Tracking Transparency prompt, Flurry Analytics found that only 4 percent of US iPhone users opted in when asked directly, meaning 96 percent chose to stay untraceable to the apps requesting it. California's Consumer Privacy Act, amended by the CPRA in 2023, gives residents a standing right to opt out of the sale or sharing of their personal information, and lets them exercise it automatically through a single browser-level signal, the Global Privacy Control, rather than clicking through each site's own settings. The EU's GDPR defines personal data broadly and requires consent that is specific, informed, and freely given, conditions a large share of EU visitors now simply decline to meet. None of this is a leak in the targeting system. It is the system now, on purpose, at population scale.

Even the infrastructure built to phase out cross-site tracking has bent to the same reality. In July 2024, Google walked back its plan to deprecate third-party cookies in Chrome, telling developers it would "introduce a new experience in Chrome that lets people make an informed choice that applies across their web browsing" instead. Google did not reverse course because tracking was thriving. It reversed because a browser-level choice architecture, letting a user decide once and have it stick, was becoming the baseline both regulators and users expected. Sellers now have to price two populations at once inside the same product: consumers whose data flows freely, and a growing bloc who, by law, by device default, or by a single toggle, no longer show up as anyone in particular.

Fuzzy Segmentation: Pricing What You Cannot See

This is the setting Jieqiong Wang, Zhaohui Jiang, and Param Vir Singh model directly in a paper newly published in Information Systems Research. Their platform faces two kinds of shoppers: consumers who share data and can be individually targeted, and privacy-preserving consumers who cannot. The textbook prediction, going back decades in the segmentation literature, is that finer information sharpens competition: the more precisely a market can be sliced into groups, the harder rival sellers must fight for each one, because a perfectly identified consumer is a consumer who can be perfectly poached by an undercutting rival. Anonymity should break that machinery. If a seller cannot identify you, it cannot target you, and a market it cannot slice should default toward something closer to one competitive price for everyone in it.

The paper's central move is to show that platforms do not respond to anonymous users by leaving them alone. They respond by building what the authors call fuzzy segmentation: pooling privacy-preserving users together with a selectively chosen subset of the data-sharing users into one blended group, rather than pricing the untraceable at whatever the competitive rate happens to be. That pooled group is priced as a group, not as individuals, which sounds like protection. The paper's first result says otherwise: this pooling softens competition and raises seller profits, the reverse of the standard prediction that segmentation intensifies price competition. Blending together consumers a rival cannot cleanly poach turns out to be a more comfortable position for a seller than losing sight of them entirely.

The Twist: Hiding From Trackers Can Raise Your Price

The second and third results are where the paper earns its title. Common intuition holds that if a seller cannot see you, it cannot charge a price tailored to your specific willingness to pay, so privacy should function as a shield. Wang, Jiang, and Singh find the opposite can hold: privacy-preserving consumers may end up facing higher prices than they would if they were simply identified and priced like everyone else, because the fuzzy segment they land in carries less internal competitive pressure than the cleanly identified segment enjoys. And because the fuzzy segment also absorbs some data-sharing consumers, pulled in only because grouping them with the privacy-preserving cohort happens to be profitable for the seller, those data-sharers take on a negative spillover of their own. They handed over their data and still end up priced as if they hadn't, penalized by proximity to consumers who withheld theirs.

A consumer who shares everything and a consumer who shares nothing can land on the identical, elevated price.

What Staples and Orbitz Got Backwards

None of this is unprecedented in kind, only in mechanism. In December 2012, the Wall Street Journal found that Staples.com charged different customers different prices for the same stapler, $15.79 in some places and $14.29 in others, driven mostly by distance to a rival office-supply store: shoppers within 20 miles of an OfficeMax or Office Depot saw the discount, while shoppers in rural and urban areas without nearby competitors, disproportionately less wealthy, paid the higher price by default. That same June, the Journal found Orbitz was steering Mac users toward pricier hotel options than PC users, after noticing that Mac owners spent as much as 30 percent more per night on travel. Neither Staples nor Orbitz needed your name. A rough proxy, device type or estimated location, was enough to sort you into a bucket and price the bucket.

Privacy law exists in large part to stop exactly this kind of proxy sorting, and in the narrow sense it works: a seller that cannot see your ZIP code or your operating system cannot run the Staples or Orbitz playbook on you directly. But Wang, Jiang, and Singh's model shows the underlying instinct, price the bucket rather than the individual, survives the loss of the proxy. Sellers just build a coarser bucket, one defined by data-sharing status instead of device or geography, and coarser buckets carry their own membership tax. The FTC's 2025 issue-spotlight on what it calls surveillance pricing describes, in granular detail, the tools sellers still have for consumers who do share data: location, browsing or shopping history, even mouse movement and abandoned carts, all folded into "advanced algorithms, artificial intelligence and other technologies" used to "categorize individuals and set a targeted price." That contrast is the point. The gap between the two populations, the finely priced and the fuzzily pooled, is now doing the sorting work that ZIP code used to do.

The Rule for Platforms and Regulators

The unifying lesson across the paper and a decade of pricing controversies is that segmentation, not identification, is the variable that sets your price. Staples and Orbitz sorted shoppers with a proxy signal because full identification wasn't available at the granularity sellers wanted. Today's platforms sort the anonymous into a fuzzy segment because full identification is legally unavailable at any granularity. In both cases, the group a consumer is placed into, not the specific facts a seller happens to know about that consumer, decides what number appears at checkout. Anonymity does not opt a consumer out of that sorting. It changes which bucket does the sorting, and this paper's finding is that the new bucket is not obviously cheaper.

For regulators drafting the next generation of privacy rules, the implication is uncomfortable: expanding the right to withhold data, without also touching how platforms are allowed to price the group that results, can raise prices for the very consumers the rule was meant to protect, and for some of the consumers who complied with data requests in good faith besides. For platforms, the honest reading is less about compliance and more about restraint. Fuzzy segmentation is profitable precisely because almost nobody outside the pricing team can see it happening, one blended number arriving quietly in a cart rather than a visible, individualized quote anyone could screenshot. The next Staples-style investigation will not find a ZIP-code rule in the code base. It will find a segment boundary, and a question about who ended up standing just inside it.