roscaftccompliancecancellation flowauto-renewal

Forcing a Phone Call to Cancel Has Cost Vonage $100M and Amazon $2.5B. Neither Case Needed the Vacated FTC Rule.

Vonage, Amazon, and Chegg paid a combined $2.6B+ for cancellation friction — all under ROSCA, a law the FTC's vacated click-to-cancel rule never touched.

XY
13 November 2026 · 8 min read

Every subscription business has run the math at some point: route cancellation through a phone call, staff that line with people trained to save the account, and some fraction of subscribers who would have clicked "cancel" in three seconds will instead give up, get talked into a discount, or just stay on hold long enough to miss the billing cycle. It works, in the narrow sense that fewer cancellations complete. It has also produced some of the largest consumer-protection settlements the FTC has ever secured, and the legal theory behind every one of them predates the click-to-cancel rule everyone thought got killed in 2025.

Key stat
$2.6B+
Combined penalties and consumer redress paid by Vonage, Amazon, and Chegg over cancellation flows the FTC said were harder to use than signup — none of it collected under the now-vacated click-to-cancel rule
Source: FTC v. Vonage Holdings Corp. (Nov. 2022), FTC v. Amazon.com (Sept. 2025), In re Chegg, Inc. (Sept. 2025) — ftc.gov press releases and consent orders

ROSCA never needed the rule that got vacated

When the Eighth Circuit threw out the FTC's Negative Option Rule in July 2025, a lot of subscription businesses treated it as the end of federal cancellation scrutiny. That reading skips over what the rule actually was: an attempt to spell out, in more operational detail, requirements that already existed under a 2010 statute called the Restore Online Shoppers' Confidence Act. ROSCA requires clear disclosure of negative-option terms and a "simple mechanism" to stop recurring charges. It was never challenged in the Eighth Circuit case, was never part of the vacatur, and has been the FTC's primary enforcement tool in every major cancellation case brought since the rule died.

"Simple mechanism" sounds vague until you read how the agency has actually applied it. In each of the three cases below, the FTC's theory wasn't that cancellation was impossible — it's that cancellation took meaningfully more effort than enrollment did, usually by forcing subscribers onto a different, slower channel than the one they signed up through.

The enforcement record

CaseResolvedAmountWhat the cancellation flow actually did
FTC v. Vonage Holdings Corp.Nov. 2022$100MLet customers enroll online in minutes; routed every cancellation through one retention-staffed phone line, then cut the line's hours — producing long holds and dropped calls
FTC v. Amazon.com (Prime)Sept. 2025$2.5BInternally named "Iliad" flow required roughly 4 pages, 6 clicks, and a choice among 15 options to cancel what took 1–2 clicks to join
In re Chegg, Inc.Sept. 2025$7.5MCancel links were hard to locate in-app; flow routed subscribers through repeated "save" offers; some customers were billed again after completing cancellation
New York v. SiriusXMNov. 2024 (state court)Liability found; damages pendingCourt called the required call-in cancellation process a "burdensome endurance contest," ruling it violated ROSCA independent of any FTC rule

None of these are software companies, and that's worth sitting with rather than dismissing. The legal theory in all four cases is channel asymmetry — sign up one way, forced to cancel a harder way — and that theory doesn't care whether the product is a telecom plan, a retail membership, a homework-help subscription, or a B2B SaaS seat. If your onboarding is entirely self-serve and your cancellation requires a call, an email to a support queue, or a conversation with an account manager before it takes effect, you're running the exact structure the FTC has been fining people for since before the click-to-cancel rule existed, and will keep fining people for now that it's gone again.

Why the phone channel specifically keeps showing up

Phone-forced cancellation isn't rare, and it isn't limited to the handful of companies big enough to draw FTC attention. A 2021 American Press Institute survey of 526 US news organizations found that only 41% made online cancellation easy for subscribers who had signed up online — and 60% said their customer service reps were specifically trained in save tactics for people who called to cancel. Those two numbers together describe the exact pattern regulators keep citing: a channel mismatch by design, backed by staff trained to extend the call rather than resolve it.

What a 2021 audit of 526 news subscription businesses found
Made online cancellation easy41%
Trained support reps in save tactics for cancel calls60%

Source: American Press Institute, survey of 526 U.S. news organizations, March 2021

The mechanism is straightforward even without regulators involved. A phone queue is the only cancellation channel where a human on the other end can keep a subscriber talking past the point they'd have clicked "confirm." It converts a binary action — cancel or don't — into a negotiation, and negotiations have a dropout rate that a button doesn't. That's exactly why it's tempting as a save-rate lever, and exactly why the FTC keeps calling it out: the save isn't coming from a better offer, it's coming from attrition during the call itself.

What "same channel" actually requires from a SaaS cancellation flow

If your signup is self-serve — someone enters a card, picks a plan, and starts using the product with no sales call involved — ROSCA's theory says cancellation needs to be reachable the same way, not funneled into a different, slower process. That doesn't mean you can't have a retention conversation. It means the conversation can't be the toll booth.

  • The cancel action has to complete in the same session, on request. A subscriber can decline a retention offer and have the subscription actually end right there — not "we've flagged this for our team" or "someone will reach out."
  • A phone number or a live chat can be offered, never required. Vonage's problem wasn't offering phone support. It was making phone the only door out for subscribers who'd walked in through a web form.
  • One retention offer is a save attempt; three escalating ones before cancellation completes starts to look like the "endless loop" ROSCA cases describe. The Amazon complaint's "4 pages, 6 clicks, 15 options" language wasn't about any single screen — it was about how many of them stacked up before the exit actually worked.
  • Cancellation needs to stop billing immediately, not "at the end of this cycle, pending review." Chegg's order specifically required that charges stop the moment a customer cancels — a detail that's trivial to get right in Stripe's subscription cancellation API and expensive to get wrong in a support queue.

None of this requires giving up the mechanics that make a cancellation flow worth building in the first place. A well-run flow still recovers subscribers who are leaving for a fixable reason — we cover the actual save-rate math in our pause vs. discount vs. downgrade comparison. The difference between that and the pattern the FTC keeps fining is entirely about sequencing: show the offer, let the subscriber say no, and let "no" actually mean the subscription ends. Our guide to building a Stripe cancellation flow goes through the API calls that make same-session cancellation the default instead of something a support rep has to execute by hand later.

The compliance landscape isn't just federal anymore

ROSCA sets the federal floor, but it's no longer the only exposure. California's amended Automatic Renewal Law has required same-medium cancellation since July 2025, and two separate class actions over it settled for $7.5 million each against HelloFresh and YouTube TV — we go through the specific button-placement rule driving those suits in our piece on California's auto-renewal lawsuits. Minnesota, Colorado, and a growing list of other states have added their own versions. What's notable about the Vonage, Amazon, and Chegg cases is that none of them needed any of that state-law machinery — a 15-year-old federal statute and the plain fact of a harder exit than entrance was enough on its own.

That's the part worth internalizing if your compliance read on the vacated click-to-cancel rule was "we can wait and see." ROSCA was never waiting on anything. It's been enforceable the entire time the regulation was being drafted, litigated, vacated, and redrafted, and the agency has shown no reluctance to use it at a scale well beyond what most SaaS businesses process in a decade of revenue. A self-serve cancellation flow that actually lets the cancel action complete — the kind CancelFlow is built around — isn't a defensive move against a specific rule. It's the thing that makes the question of which rule is currently in force irrelevant to your exposure.

Frequently asked questions

Is it illegal to require a phone call to cancel a subscription that was purchased online?+

It's legally risky even without the FTC's vacated click-to-cancel rule. The Restore Online Shoppers' Confidence Act (ROSCA) has required a simple cancellation mechanism for online negative-option sales since 2010, and the FTC has read 'simple' to mean the same channel used to sign up. In FTC v. Vonage, the agency specifically alleged that letting customers enroll online but cancel only by phone — through a retention-staffed line with reduced hours — violated ROSCA on its own, independent of any click-to-cancel regulation.

What's the difference between ROSCA and the FTC's click-to-cancel rule?+

ROSCA is a 2010 federal statute that has never been vacated, suspended, or seriously challenged in court. The click-to-cancel rule was a 2024 regulation meant to spell out ROSCA's requirements in more detail — same-medium cancellation, no retention gauntlet, cancellation at least as easy as signup — before the Eighth Circuit vacated it in July 2025 on a procedural technicality. Every settlement in this piece (Vonage, Amazon, Chegg) was built on ROSCA directly, not the regulation that got thrown out.

How much have companies actually paid in fines for hard-to-cancel subscriptions?+

Amazon agreed to pay $2.5 billion in September 2025 ($1 billion in civil penalties, up to $1.5 billion in consumer redress) over its Prime enrollment and cancellation flow. Vonage paid $100 million in November 2022 for routing all cancellations through an understaffed phone line. Chegg paid $7.5 million in September 2025 for cancellation links that were hard to find and a cancellation flow that kept charging some customers after they tried to leave. None of these three companies is a subscription-software company, but the underlying legal theory applies to any SaaS selling self-serve, auto-renewing subscriptions.

If my SaaS has a retention or save team, does that violate ROSCA?+

Not by itself. The violation isn't having a save offer — it's making the save offer or a phone call the only way out. Vonage's problem wasn't that it tried to retain customers on the phone; it's that phone was the only cancellation channel it offered to people who'd signed up entirely online. A self-serve cancel button that also surfaces one retention offer, with cancellation still completing in the same session if the subscriber declines, hasn't drawn this kind of enforcement.

Try CancelFlow

Stop losing subscribers today

One script tag. One function call. A live cancellation flow in under 10 minutes.

Start free trial →
← All postsHome