The Price Should Not Know Who You Are

Why this brief was updated

Michigan lawmakers now have several competing approaches before them, but the central drafting question remains. HB 6098 and HB 6099 passed the Michigan House on June 25, 2026, by 92-15 votes and were referred to the Senate Economic and Community Development Committee on July 1. SB 991 remains pending in that committee. HB 5771 remains pending in the House Economic Competitiveness Committee. [1-6]

At the federal level, the FTC released a proposed enforcement policy on personalized pricing on August 19 and later extended the public-comment deadline to September 25. That comment period closed on September 25. The proposal remains an agency proposal rather than a final enforcement policy. The Commission states that it does not have general authority to ban personalized pricing in every circumstance, but warns that undisclosed collection or use of personal data to set individualized prices may violate laws it enforces. [7-8]

Define the practice by economic effect

Surveillance pricing should be defined as the material use of personal data - collected directly, purchased, derived, or inferred - to change a consumer's total economic offer. The covered effect can be a higher base price, a mandatory fee, a reduced or withheld generally available discount, or deliberate suppression of a lower-priced substantially similar offer under comparable market conditions.

That definition is narrower than a ban on dynamic pricing. Prices can legitimately vary because of aggregate demand, inventory, season, time, weather, competitor pricing, taxes, actual delivery cost, or geographic operating costs. The legal distinction should turn on whether personal surveillance changes the individual consumer's economic treatment.

FTC staff reported in January 2025 that the intermediaries it examined could draw on precise location, browser history, search and purchase activity, mouse movements, abandoned carts, demographics, and inferred characteristics. The firms worked with at least 250 clients. The FTC cautioned that some examples were preliminary or hypothetical to protect confidential business information. [9]

The data pipeline begins before the price

A pricing rule tied to one sensor or device will age badly. Retail apps, loyalty systems, ad-tech platforms, connected vehicles, payment systems, and data brokers can collect or receive information for navigation, fraud prevention, convenience, analytics, voice assistance, or product personalization. The economic consequence may occur later, after the data has moved to a separate vendor or decision system.

The FTC's January 2026 final order involving General Motors and OnStar illustrates the downstream-use problem. The agency alleged that precise geolocation and driving-behavior information from millions of vehicles was collected, used, and sold without adequate notice and affirmative consent. The order includes a five-year restriction on specified disclosures to consumer reporting agencies and 20-year requirements involving consent, access, deletion, and controls. [10]

Walmart's current privacy notice also shows why source-neutral language matters. The notice describes categories that can include precise location, voice interactions, sensory information, online activity, commercial information, and inferences; for the enabled "Hey, Walmart" feature, it states that the app listens for the wake phrase while open and does not retain information before recognizing it. The disclosure does not establish that Walmart uses those data for surveillance pricing. It does demonstrate the range of data categories that may exist inside a modern commercial relationship. [11]

Seven questions Michigan lawmakers should settle

Protected-class and proxy data need a clear rule

A system does not need an explicit race, disability, age, sex, income, or other sensitive field to reproduce a risk tied to those characteristics. Location, language, device characteristics, browsing patterns, household composition, and combined inferences can function as proxies. SB 991 places protected-class and device/geolocation data directly in the pricing debate; Michigan should decide in the statutory text whether direct and proxy use are covered and how intent, materiality, and evidence will be evaluated. [5]

This question should be handled carefully. A general pricing statute should not create an undefined disparate-impact test by accident, and it should not displace existing civil-rights law. A workable provision can prohibit direct or knowing proxy use of protected or sensitive information to worsen an offer, preserve other applicable protections, and require regulators to identify the data and decision path supporting a violation.

Enforcement must be capable of reconstructing the offer

The central evidentiary question is comparative: under comparable market conditions, did personal data materially worsen this consumer's offer? A regulator cannot answer that from a privacy notice alone. The record must connect the data categories, system version, reference price or comparison method, applicable market factors, vendor role, and final offer.

Michigan's July 2026 Supreme Court decision in Attorney General v. Eli Lilly also matters to exemption design. The court rejected a broad regulated-entity approach and directed attention to whether the specific transaction or conduct was authorized by law. A surveillance-pricing exception should follow that conduct-specific logic rather than immunize unrelated pricing practices because a business participates in a regulated industry. [14]

Write the rules so businesses can comply and regulators can enforce them

• Preserve bona fide market pricing based on nonpersonal factors applied consistently to similarly situated consumers.
• Protect proprietary source code and trade secrets from public release while preserving confidential access for authorized regulators and auditors.
• Use clear definitions for material factor, generally available benefit, comparable market conditions, and bona fide reference price.
• Avoid treating a buried privacy policy or nominal consent as permission for conduct the statute otherwise prohibits.
• Avoid regulating every recommendation, product ranking, or personalization choice unless it is tied to a covered economic effect.
• Use a practical implementation period for existing systems and vendor contracts, with earlier applicability for newly deployed systems if lawmakers choose.

The total-price concept is consistent with the broader consumer-protection principle that a represented price should include mandatory charges. In March 2026, for example, the FTC warned 97 auto-dealer groups that advertised prices must reflect required fees. That action is not a surveillance-pricing rule, but it demonstrates why mandatory-fee coverage matters when lawmakers define the consumer's economic offer. [15]

What Michigan should measure after enactment

1. Complaints and investigations by pricing practice, market, and data source.
2. The share of matters resolved through correction, restitution, penalties, or no violation found.
3. Use of third-party vendors and the frequency of missing or inadequate records.
4. Changes in loyalty-program terms, mandatory fees, reference-price methods, and disclosures.
5. Compliance cost and implementation problems reported by small and large businesses.
6. Evidence of evasion through fees, rankings, selective discounts, account design, or proxy variables.

A review provision after two or three years would let the Legislature test whether the definitions are catching harmful conduct without sweeping in ordinary market pricing.

Bottom line

Michigan can draw a narrow line between legitimate market-responsive pricing and individualized economic treatment driven by personal surveillance. The strongest text will focus on the total offer, modern data relationships, seller and vendor responsibility, auditable evidence, and conduct-specific exceptions. Those choices determine whether the policy works in practice.