Claims & Proof Checker

The Claims & Proof Checker reads marketing copy one sentence at a time and flags every claim a buyer, a competitor or a regulator could ask you to back up. Paste a landing page, an ad, a launch email or a case study. Each sentence gets a verdict: needs proof, needs consent, review wording, or OK. This flags claims that need support. It is not legal advice: check the legal consequences with counsel before publishing.

What it checks

Fourteen kinds of claim: specific numbers, percentages and multipliers; result claims such as "increase revenue" or "improves reading scores"; "proven" and "studies show" language; superlatives such as "best", "#1" and "industry-leading"; comparisons with a named or implied competitor; star ratings and review counts; testimonials, endorsements and creator posts; compliance badges such as "HIPAA compliant", "SOC 2", "FERPA" and "GDPR"; security and privacy promises; health and clinical claims; AI claims such as "AI-powered", a stated accuracy figure or "replaces your team"; absolutes and guarantees; "free", "risk-free" and limited-time offers; and environmental claims.

The method

Every flag quotes the words that triggered it, says why that kind of claim needs support, names the evidence or permission to keep on file, and links the regulator's own guidance: the FTC's policy statements on advertising substantiation and comparative advertising, its Advertising FAQs for small business, the Endorsement Guides FAQ, the Consumer Reviews and Testimonials Rule Q&A, the Health Products Compliance Guidance, Start with Security, the COPPA FAQ and the "Free" guide; HHS guidance on HIPAA marketing and on Security Rule "certifications"; the U.S. Department of Education's FERPA page; and the EDPB register of GDPR certification schemes. Choose the audience - general, B2B SaaS, AI, EdTech or MedTech - and the check adds the notes that apply, such as HIPAA authorization for a patient testimonial or parental permission for a student quote.

If you already hold the evidence, add a note: the flag id from a check, or a line that quotes the claim and says where the proof lives. The flag stays on the list, marked on file, so the record of what was checked survives the next edit.

Why I built it this way

A claim you can't support is a liability, not a selling point. I would rather a founder cut a strong-sounding line than publish one they cannot show. So the checker never says a claim is fine or legal. It tells you which sentences make a claim, what would hold up, and where the regulator explains the rule in its own words. The legal call belongs to you and your counsel.

Limits

The check matches wording. It cannot see context, images, charts or the overall impression a page leaves, and a phrase can occasionally be flagged where it is not a claim. Read each flag, and read the unflagged sentences for implied claims. It covers U.S. federal guidance plus the EU register for GDPR certification; state laws, sector rules and other countries are not covered. Your copy stays in your browser.

The same engine runs as an agent skill, claims-and-proof-check. To turn a customer result into proof you may publish, use the Case Study Kit.

Run this in your agent

The free AAJ agent skill Claims & Proof Check does this job inside a coding agent, on your own data. Install it with npx skills add sarojkjha/aaj-marketing-skills --skill claims-and-proof-check.

Using Claude or ChatGPT instead? Add the AAJ engines as a custom connector, with no install: https://aaj-engines-mcp.sarojkjha10.workers.dev/mcp