Cancel Atlas

The Cancellation Friction Index

Documented-Policy tier only. Scores below rate how clearly each company's public cancellation page is published (cited, dated) — opinion grounded in disclosed facts, not a finding about the real cancel experience. The behavioural Verified-Flow grade is pending for every company until a human/crowd audit exists.

Open · evidence-cited · per jurisdiction

Signing up is one click.
Leaving should be too.

Cancel Atlas gives every company two scores — 📄 what they publish and ✅ what actually happens when you cancel — from 0 to 100, per jurisdiction, with the evidence beneath each. The gap between the two is the tell. Free and open, built for humans and AI agents.

¹ Mills et al., Dark Patterns & Sludge Audits, Behavioural Public Policy, 2026.

Median audited service is harder to leave than to join
57%of services studied¹
Companies scored
—documented-policy tier · cited
Two scores per company
📄 vs ✅policy vs tested flow

The index

The Index

Sort to see where leaving is hardest — or where it's effortless.

    Methodology · CES-1.1

    How the two scores work

    Each company gets two separated 0–100 scores (higher = easier), computed per service × jurisdiction. We never average them: a glossy policy page must not be allowed to mask a nightmare flow. Every score is a calculation you can reproduce from the published weights — not an opinion we ask you to trust.

    Verified-Flow score — what actually happens

    The behavioural score (the real CES): we run the live cancel flow end-to-end. Seven observable dimensions.

    Documented-Policy score — what they publish

    The transparency score from public pages only — cheap, scalable, covers the long tail. Five dimensions.

    The gap is the signal

    • 📄 B · ✅ F A wide policy-minus-flow gap means the published policy reads far easier than the cancel actually is — a dark-pattern tell, and the most newsworthy card.
    • F Flow: no online cancellation path exists at all (phone / in-person / mail only).
    • A Flow: online + ≤3 steps + no mandatory retention step + symmetric channel (an A with a mandatory retention step is demoted to B).
    • B / C / D fall between, by score. Companies we haven't flow-audited yet show the Documented-Policy score alone, clearly marked. The raw 0–100 is always shown so the letter is never the whole story.

    Campaign

    The Click-to-Cancel Challenge

    The standard — prove all five

      How an entry moves

      Pledged the company committed and submitted a live cancel link — no judgment yet.

      Verified our triple-blind human test confirmed it; carries a public last-verified date and is re-tested at random.

      Did not pass the pledge didn't hold up under test. You can't pay to remove it; re-pledge after 90 days.

      Free and unbuyable. The leaderboard sorts by recency — never by who pays.

      Leaderboard

      Verified one-click cancellers, newest first.

        Take the challenge › — companies pledge and submit a live cancel link; we verify it.

        For machines

        Built for AI agents & developers

        One canonical record, many representations — so the human page and the machine data can never disagree. Cancellation bots and assistants should be able to cite the same score a person sees.

        Structured data

        schema.org Dataset + per-company ratings in JSON-LD, parity-checked against the HTML.

        Markdown for agents

        Content-negotiated text/markdown on the same URLs (≈80% fewer tokens). Planned.

        MCP server

        A thin read-only Model Context Protocol endpoint so agents can query scores directly. Planned.

        llms.txt is published as a courtesy index (/llms.txt) with no promised SEO/citation effect.

        Neutral by construction

        Cancel Atlas is designed to be a non-profit-style public utility. We cannot be paid to change a score — grades are computed only from cited evidence and the published weights. Any sponsored slot is clearly labeled, kept separate from the rankings, and never affects a company's grade; no tracking, and no affiliate links to the companies we rank. The work is funded by labeled sponsorship and by licensing the open dataset to the AI agents and compliance teams that need it — never by selling a better grade.

        Fair, and accountable

        Every score is opinion grounded in disclosed, dated facts — the cancellation steps, channel and source are shown on each card. We anchor to currently-valid law (ROSCA §8403, state Automatic Renewal Laws, FTC Act §5), not the FTC "Click-to-Cancel" rule that was vacated in July 2025. Companies get a right of reply and a narrow factual-error flag — with identical handling whether or not they pay us anything.

        Participate

        Claim a listing or report your experience

        Companies can submit a direct cancel link to improve their Documented-Policy score; anyone can report a real cancel experience to help seed the Verified-Flow audits. Everything is queued for human verification against your source — nothing auto-publishes, and identical handling applies whether or not you pay us.

        Is this your company? Request a correction or submit a right of reply — email [email protected] (or see Contact) or use the form below. A human reviews every request and substantiated corrections are made promptly; sponsorship never affects a score.

        Submissions go to an isolated, off-origin endpoint — never to this static site. A human confirms every entry against its source before any score changes. See our methodology & right-of-reply above.

        Support & sponsor

        Back a neutral public utility

        Cancel Atlas is free, open, and built so no one can pay to change a grade — scores come only from cited evidence and the published weights. Here's how it's funded, all of it score-independent.

        Advertise

        Featured supporter

        By enquiryfounding rate

        Premium labelled slot pinned above the rankings. Available site-wide, or pinned to the top of any single vertical — one advertiser per slot.

        Enquire →

        Advertise

        Category sponsor

        By enquiryper vertical

        Labelled slot shown only within one industry (streaming, gyms…). Contextual, never beside your own grade.

        Enquire →

        Licensing

        Data & API

        By enquiryIndie → Enterprise

        Structured access to 1,100+ cited, dated cancellation records for compliance teams, comparison sites & AI agents.

        For agents →

        For companies

        Monitoring & re-audit

        By enquiryper brand

        Faster re-verification cadence + change alerts. Buys cadence, never a better grade; includes a Click-to-Cancel Challenge audit.

        Enquire →

        Backers

        A backer chip is a public thank-you, not a placement — like sponsorship and licensing, it never touches a score.

        Become a backer

        Monthly or yearly — support keeps the index free, neutral and ad-light. Cancel in one click anytime via the billing portal. Backing never touches a score; contributions aren't tax-deductible.

        Individual $5/mo Our thanks — keeps the index free Back monthly or $50/year
        Org $15/mo Logo on the backers wall Back monthly or $150/year
        Patron $25/mo Logo on the wall + top billing Back monthly or $250/year

        Or just chip in

        If the index saved you from an unwanted renewal, drop a tip — no account, pay what you want.

        Cancel Atlas is run by a for-profit company; contributions support the work and aren't tax-deductible.

        Sponsorship, licensing and monitoring fund the work; none of them can move a score. Sponsored slots are labelled and never a rated company beside its own grade. Slots are limited and vetted — to enquire, email [email protected] (we'll confirm fit, agree terms, then invoice — there's no self-serve ad checkout).