The 2026 Subscription Friction Index: How Hard Is It to Leave?

How hard is it to leave a subscription you no longer want? For most people, the answer is a guessing game: you hunt for a hidden button, sit through retention screens, and sometimes end up on the phone. The Subscription Friction Index turns that guesswork into a single public number — a “Lighthouse Score for leaving,” but for cancellations instead of page speed. Where Lighthouse tells a developer how fast their page loads, the Friction Index tells a subscriber — and a journalist, a regulator, or a competitor — how hard a company makes it to walk away. Every score on this page is published openly by Cancellation Intelligence, so a brand’s cancellation difficulty is no longer a private secret it controls.

This report is the methodology and headline release for the 2026 Index. It covers 203 digital subscription services, scores each on six public friction axes, and prints the formula so you can reproduce any number yourself. The cancellation paths behind every score were last re-verified on 2026-08-14. If you only remember one thing, remember this: cancellation friction is not random. It is engineered, it is measurable, and now it is public.

Key takeaways

• Across 203 services, 95.6% (194) turn auto-renew on by default — leaving is the exception, not the rule.
• 80.8% (164) hide their refund terms (refund-visibility score of 6 or higher). This is the single strongest pattern in the dataset.
• 29.6% (60) make the cancel entry point hard to find, and 18.2% (37) deploy retention dark patterns (multiple confirm screens).
• 33.0% (67) take five or more steps to cancel, and 5.4% (11) force you to call a phone number.
• The overall mean Friction Score is 3.99 and the median is 4.3; the hardest single service scores 7.2.
• The methodology is independent of any law and works whether or not the FTC’s Click-to-Cancel rule is in force.

The headline numbers

These are the figures you can quote. Every one is a count out of the 203 services in the 2026 dataset, with the share in parentheses.

Friction signal	Services	Share
Auto-renew is on by default	194 / 203	95.6%
Refund terms are opaque (refund-visibility ≥ 6)	164 / 203	80.8%
Cancel entry point is hard to find (find-entry ≥ 7)	60 / 203	29.6%
Retention dark patterns (multi-confirm ≥ 6)	37 / 203	18.2%
Cancellation takes 5+ steps	67 / 203	33.0%
A phone call is required to cancel	11 / 203	5.4%
Billed through the App Store	77 / 203	37.9%
Billed through Google Play	76 / 203	37.4%

The 80.8% figure is the one to remember. Four out of five digital subscriptions make you work to understand whether — and how much — you get back if you leave mid-cycle. Refund opacity is the most widespread friction signal we measured, more common than hidden buttons and more common than phone calls. A service can let you find the cancel link in seconds and still score poorly here, because the real question is whether the money side of leaving is explained in plain language before you commit.

The 95.6% auto-renew number explains why friction matters at scale. When nearly every service renews unless you act, the cost of a confusing cancellation is not a one-time annoyance — it recurs every billing cycle, silently, for as long as the friction keeps you from leaving. Default-on renewal converts a moment of inattention into a continuing charge.

The step-axis average is 3.8 out of 10, which sounds mild until you meet the services that stack five screens, a chat bot, and a phone call on top of it. The overall mean of 3.99 hides a long tail: most services are moderately annoying, a stubborn minority are genuinely engineered to retain you, and a small set — the 11 that demand a call — sit in a category of their own.

How the Subscription Friction Index is calculated

Too many “hardest to cancel” lists publish a subjective Easy / Medium / Hard label with no method behind it. The Index does not. Every score comes from one public formula that anyone can recompute.

Each service is scored 0–10 on six axes. The axes are weighted to reflect how much each one actually slows a real person down:

Axis	Weight	What it measures
Find entry	28%	How hard it is to locate the cancellation option at all
Steps	18%	How many steps the cancellation flow actually takes
Refund visibility	15%	Whether refund and proration terms are stated clearly
Confirmation (dark patterns)	15%	How many “are you sure?”, upsell, or “save this plan” screens you hit
Contact support	12%	Whether you are pushed toward chat, email, or a support ticket
Phone	12%	Whether you are forced to make a phone call

The weighted result is clamped to 0–10 and rounded to one decimal place. The same function runs on the server and in the browser, so this page and every individual cancellation guide show the same number. Because the weights are public and the per-service inputs come from observed flows, a reader can reproduce any Friction Score from the inputs we publish.

A note on design: find-entry carries the heaviest weight (28%) because “I can’t even find the button” is the most common and most universal complaint, and it gates every other axis. You cannot evaluate steps or refund clarity until you have found the door. Phone carries 12% but is the sharpest single penalty — a service that forces a call almost always lands in the high-friction tier, because a phone call introduces wait time, business-hours constraints, and a live retention agent all at once.

Worked example. Take a service that scores find-entry 8, steps 6, refund-visibility 7, confirmation 6, contact-support 4, and phone 9. The weighted sum is 8×0.28 + 6×0.18 + 7×0.15 + 6×0.15 + 4×0.12 + 9×0.12 = 2.24 + 1.08 + 1.05 + 0.90 + 0.48 + 1.08 = 6.83, which rounds to 6.8. Raise any one axis and the total moves predictably; that is the point. The formula is linear and transparent, so a company can see exactly which axis to fix to drop its score.

How the six axes map to the OECD consumer-friction framework

The Index is built independently, but its six axes line up cleanly with the consumer-friction concerns the OECD has documented — including its 2024 Council Recommendation on dark patterns and its ongoing work on subscription traps and switching costs. The mapping below is our analytical alignment, not an official OECD score, and we are not implying OECD endorsement of any individual result.

Our axis	OECD consumer-friction concern
Find entry	Pre-contractual transparency and the right to exit; hidden exit points are a core subscription-trap signal
Steps	Cost of switching or exiting; more steps mean higher real-world exit cost
Refund visibility	Clarity and fairness of price and refund terms; opaque terms read as unfair contract terms
Confirmation (dark patterns)	Manipulative design; directly addressed by the OECD’s dark-patterns recommendation
Contact support	Barriers to exercising exit rights through customer-service friction
Phone	The most severe involuntary-retention barrier; a forced call is the highest-friction exit

The point is not to borrow the OECD’s authority but to show the Index measures the same thing regulators care about: whether a consumer can leave as easily as they joined. When a watchdog flags “negative option” subscriptions or manipulative design, it is describing the exact behaviors our axes quantify. A public score makes that description comparable across 203 services instead of anecdotal.

What’s in the dataset — and what we left out

Honesty first. The 2026 Index covers 203 digital subscription services. It deliberately excludes gyms, telecoms, and insurance. That is a narrow cut, and it is intentional.

Digital subscriptions share a comparable cancellation surface: a website or app, an account settings page, and a billing relationship you manage yourself. That homogeneity is what makes the scores meaningful — we are comparing like with like. Gyms, telecoms, and insurance cancellations run through very different machinery (in-person contracts, regulated number porting, underwriting), so mixing them in would drag the averages without improving the comparison. A streaming service and a mobile carrier are not the same cancellation problem, and pretending they are would make both numbers less useful.

The trade-off is real: no service in our 203 reaches the “extreme” tier (8.1–10). The genuinely brutal cases — the ones that require a notarized letter or a branch visit — sit outside the digital universe we measured. Inside our cohort, 32 services score 3.0 or below (easy), 161 fall in the moderate 3.1–6.0 band, and just 10 cross into the 6.1–8.0 “high” tier. The highest score is 7.2. We report the limit plainly so no one reads a 7.2 as “the worst in the world” rather than “the worst in consumer digital subscriptions.”

The 15 hardest subscriptions to cancel

Here is the top 15 from the 2026 dataset, with each service’s Friction Score. These are the services where the cancel button is hardest to find, the confirmation screens pile up, and the refund terms stay vague. Note the pattern: the top of the list is dominated by SaaS and news media, not the streaming services people complain about loudest.

#	Service	Category	Friction Score	Cancel guide
1	Flipboard	SaaS	7.2	Flipboard
2	Wrike	SaaS	7.2	Wrike
3	Bloomberg	Newsletters & Media	7.2	Bloomberg
4	HubSpot	SaaS	6.5	HubSpot
5	Datadog	SaaS	6.5	Datadog
6	Zendesk	SaaS	6.5	Zendesk
7	Intercom	SaaS	6.5	Intercom
8	Pocket	SaaS	6.5	Pocket
9	The Wall Street Journal	Newsletters & Media	6.5	WSJ
10	Politico	Newsletters & Media	6.1	Politico
11	Adobe Creative Cloud	SaaS	5.6	Adobe Creative Cloud
12	Netflix	Streaming	5.4	Netflix
13	Hulu	Streaming	5.4	Hulu
14	SiriusXM	Music	5.4	SiriusXM
15	WeightWatchers	Fitness	5.4	WeightWatchers

A note on ties: 13 services score exactly 5.4, so positions 12–15 are a photo finish. We list the best-known consumer names (SiriusXM, WeightWatchers); Stability AI, Semrush, Xero, GoDaddy, and others share the same score. If a service is not in the table, look it up directly. For the full ranked list of all 203, see the hardest subscriptions to cancel guide.

Why SaaS and news media lead is no mystery. SaaS products are sold to teams, not individuals, so the account owner who can cancel is often not the person who feels the charge. News media sits behind paywalls whose entire business model depends on involuntary renewals surviving past the introductory rate. Streaming, by contrast, competes on churn-and-return: it is usually easier to leave because the companies would rather have you back next month than trap you this month.

What a low-friction cancellation looks like

The other end of the scale is worth showing, because it proves the Index is not just a wall of bad news. The easiest services in the dataset — Tubi and Pluto TV at 1.1, followed by Vudu, Apple Arcade, and Discord Nitro at 2.1 — put the cancel control inside account settings, explain refunds in a sentence, and ask at most one confirming question. These services treat cancellation as a normal account action rather than a threat to the business. The gap between 1.1 and 7.2 is the entire story of the Index: leaving should be a settings click, and for most of the industry it still is not.

For journalists and researchers

This page is built to be quoted. When you cite the Index, the canonical reference is:

Cancellation Intelligence, 2026 Subscription Friction Index — 203 digital subscription services, six public friction axes, last verified 2026-08-14.

Suggested pull quotes:

• “80.8% of digital subscriptions hide their refund terms.”
• “95.6% turn auto-renew on by default.”
• “11 of 203 services still force you to cancel by phone.”

The methodology above is the source of truth and is reproducible: the six-axis weights are public, the formula is printed, and the per-service inputs come from observed cancellation flows. If you need the underlying per-service scores or a CSV extract for analysis, the data is published openly alongside this report. Please attribute the Index to Cancellation Intelligence and link back to this methodology page so readers can verify any number.

FAQ

What is the Subscription Friction Index? The Subscription Friction Index is a public, 0–10 score that measures how hard it is to cancel a digital subscription. Published by Cancellation Intelligence, it covers 203 services scored on six friction axes: find entry, steps, refund visibility, confirmation dark patterns, contact support, and phone. A higher score means a harder exit.

How is the friction score calculated? Each service is scored 0–10 on six axes, then weighted: find entry 28%, steps 18%, refund visibility 15%, confirmation 15%, contact support 12%, and phone 12%. The weighted result is clamped to 0–10 and rounded to one decimal. The full formula is printed on this page so anyone can recompute any score.

Is the data reproducible? Yes. The six-axis weights are public and the per-service inputs are derived from observed cancellation flows last re-verified on 2026-08-14. Because the scoring formula runs identically on the server and in the browser, any reader can reproduce a given Friction Score from the published inputs.

Does the Index reflect any law, like the FTC’s Click-to-Cancel rule? No. The Index measures observable cancellation friction independent of any regulation. On July 8, 2025, the U.S. Eighth Circuit Court of Appeals vacated the FTC’s Click-to-Cancel rule; our methodology does not assume that rule is in force and scores what a subscriber actually experiences.