The Aggregate-vs-Individual Line: Auditing Casper's Pricing Rule Set
Objective: Given a synthetic list of 10 active pricing rules from a mattress ecommerce pricing engine, classify each as aggregate-market signal or individual-profile signal, per the lesson's regulatory line, and apply the disclosure test.
Casper Sleep's growth team is reviewing its AI pricing engine's active rule set ahead of a compliance review, after the Delta and Instacart headlines put legal on edge.
Classify all 10 rules as aggregate or individual-profile signals, flag which would fail the lesson's disclosure test, and recommend which to keep, disclose, or kill.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free, tabular, sufficient for a 10-row legal classification exercise
The process
1 step
Step 01 of 01
The lesson draws one legal line: adjusting price on market-wide signals (inventory, competitor price, time of day) is standard and unrestricted; adjusting price on who the specific person is (device, browsing history, inferred income) is what the FTC calls surveillance pricing.
Rule 4 raises price 8% when a shopper's device is flagged high-income by a third-party data broker. Rule 7 raises price 8% when warehouse inventory for that mattress model drops below 50 units. Same 8%, same trigger size, different legal risk. Why?
Procedure
- List all 10 active pricing rules with their plain-language trigger description.
- For each rule, ask: does the trigger depend on market conditions (inventory, competitor price, season) or on data about a specific shopper (device, location, browsing, inferred income)?
- Classify each rule Aggregate or Individual-Profile based on that single test.
- For every Individual-Profile rule, apply the disclosure test: would a visible 'this price was personalized' label change the offer's economics?
- Recommend Keep for Aggregate rules, Disclose-or-Kill for any Individual-Profile rule that fails the disclosure test.
Casper pricing rule audit (excerpt) Rule 4: +8% when device flagged high-income by data broker Classification: Individual-Profile Disclosure test: FAILS, conversion drops if shopper sees why price is higher Recommendation: KILL Rule 7: +8% when warehouse inventory drops below 50 units Classification: Aggregate Disclosure test: N/A, no personal data used Recommendation: KEEP
Healthy
Two rules with an identical 8% price move classified differently, correctly, because the trigger data source differs.
Unhealthy
Classifying by price-change size instead of by what data triggers the change.
What this means
The legal line is about the input signal, not the output price or its size.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| A rule using device or location data is left active because 'the increase is small' | Reclassify by trigger type, not by the size of the price change, then apply the disclosure test | 30 min |
Final deliverable
A pricing rule audit memo classifying all 10 rules as Aggregate or Individual-Profile, with a Keep/Disclose/Kill recommendation for each.
See a reference example
Robinhood, Gold subscription pricing rule audit (excerpt) Rule 2: 20% discount shown after 3rd cart abandonment on Gold signup Classification: Individual-Profile (based on this user's own behavior history) Disclosure test: PASSES, a visible 'loyalty discount' label doesn't hurt conversion Recommendation: KEEP, discounting down from a public price is the safe direction Rule 9: Price shown 6% higher for users on iOS vs Android Classification: Individual-Profile (device-based) Disclosure test: FAILS Recommendation: KILL
Success criteria
You're done when you can:
- Classifies all 10 rules correctly as Aggregate or Individual-Profile based on trigger data, not price size
- Applies the disclosure test to every Individual-Profile rule
- Reaches a Keep/Disclose/Kill call for each