CartParity

Who it's for

Compliance and e-commerce leads at grocery chains and grocery-delivery platforms operating in Seattle (NYC and other cities as bans spread); today they cope with a one-time outside-counsel memo or by trusting their pricing vendor's word.

The problem

Legal: the Seattle ordinance bans using personal data to set grocery prices, and documented Instacart-style price experiments prove the variance is real — one profile-based price difference is per-violation penalty exposure and plaintiff bait, and an internal self-check carries no credibility in discovery or before the regulator.

How to build it

Hosted monitor, zero integration (sweeps run against public digital storefronts): web dashboard of per-SKU variance, weekly PDF certificate, Slack/email alert within 24h of any detected profile-based price difference, public methodology page for evidentiary weight.

How it makes money

Grocery chains and delivery platforms pay roughly $300-800/month per monitored storefront for the independent certificate and variance alerts, because no free tool runs continuous two-profile sweeps and a self-run internal audit is worthless as evidence — the same dynamic that paid cookie-consent vendors after GDPR.

Why it doesn't exist yet

Incumbents skip it: the pricing vendors (Instacart, Eversight, Revionics) are the ones doing the personalization and will never audit themselves, while consultancies sell $100k assessments, not $400/mo continuous monitors. The indie gap: the detection technique (sterile vs data-seasoned synthetic profiles comparing identical baskets at the same minute) was published by consumer nonprofits like Consumer Reports and Groundwork but never productized — the first city ban just converted a research method into a compliance line item.

First users

The ban's backers named the practice in national press this week, so 'surveillance pricing compliance' search demand is spiking exactly now: first 10 are Seattle-market regional grocers and their outside counsel (reachable via the Northwest Grocery Association and local coverage), plus one plaintiffs' firm paying for raw access to the public variance ledger as a case-finder.

Build size

2 people x 10 weeks: Playwright-based two-profile sweep harness with synthetic-profile provisioning and hygiene, basket-diff engine, weekly scheduler covering 5 Seattle-serving storefronts, public sweep ledger, PDF certificate/complaint generator; excludes real-user credential checks, mobile apps, and non-grocery retail.

Biggest risk

A concrete killer: Washington State passes preemption or Seattle enforcement guidance blesses vendor self-certification, evaporating the independent-audit budget — or Instacart/Kroger harden bot detection enough that synthetic-profile sweeps fail at scale.

Conditions for a hit (all 3 required)

• Given a 10+ SKU basket on a supported storefront, produces a side-by-side same-minute price table from one sterile synthetic profile and one data-seasoned synthetic profile, flagging every SKU with a price delta and attaching timestamped captures from both sessions
• Runs unsupervised weekly sweeps of at least 5 grocery storefronts serving Seattle ZIP codes and publishes per-storefront variance counts and sweep logs on a public ledger
• On detecting a profile-based price difference, emits a regulator-ready evidence PDF citing the Seattle ordinance section with both sessions' timestamped screenshots, downloadable without contacting support

How it's judged (in 6 months)

Public sweep ledger showing at least 1,500 completed two-profile basket audits across at least 5 Seattle-serving grocery storefronts, plus at least 3 publicly named paying customers (logo, testimonial, or case study) by the judge date — any vendor name implementing this shape counts(judgment date 2027-03-27)

AI self-confidence 35/100 — self-reported likelihood of meeting the criterion, not a business success rate