Methodology · Hard rejects
Dark patterns AUG rejects — and what to ship instead
Dark patterns spike D1–D7 retention by 5–15% for the first 2–3 weeks, then accelerate churn through months 2–6 as users notice the manipulation. AUG v3 multiplies these archetypes by -1.0 in the Retention Oracle — they're filtered out before any score is computed.
What makes a pattern “dark”
Dark patterns are UI/UX choices that exploit user cognitive biases to extract behavior the user would not freely choose. The defining trait: the operator benefits while the user loses trust. The trade is asymmetric and short-term.
Healthy products operate the opposite way — they leave money on the table in the short term to compound trust over the long term. The math is brutal: trust takes months to build, seconds to lose, and compounds geometrically with both retention and advocacy.
The 12 dark patterns AUG hard-rejects
1. Fake scarcity
“Only 2 left at this price!” — on a digital product with infinite supply. “3 other people are looking at this right now” — when the number is random. Users with any sophistication recognize this immediately.
Ship instead: real scarcity (limited-time-offer with a verifiable deadline) or no scarcity claim at all.
2. Confirmshaming
“No thanks, I don't want to save money” on a newsletter modal. “Skip — I love wasting time on bad processes” on an upgrade flow. Insulting the user as a cost of declining.
Ship instead: a neutral “No thanks” button. Users who decline politely now will say yes politely later.
3. Sneak-into-basket / pre-checked add-ons
Pre-ticked checkboxes for premium support, insurance, newsletters, donations. User has to actively decline. Often-illegal in EU/CCPA jurisdictions; always corrosive to trust.
Ship instead: unchecked boxes by default. Users who opt in willingly have higher LTV than users who didn't notice.
4. Roach-motel cancellation
Easy to subscribe (one click) but hard to cancel (call this number Monday-Friday 9-5 Eastern Time). FTC actively prosecuting this in 2026. Beyond legal risk, it produces viral negative word-of-mouth.
Ship instead: cancel flow as easy as signup. If a user has to cancel, let them. They might come back. They will definitely tell others about the experience.
5. Hidden costs / drip pricing
Quoted price doesn't match charged price. Shipping/tax/fees revealed only at checkout. Users feel tricked, even if the math is technically disclosed somewhere.
Ship instead: total price (including all fees) on the pricing page. Lower conversion in the short term; massively higher trust in the long term.
6. Trial gotchas
Credit card required for “free” trial. Auto-charge without clear advance notice. 30-day trial that auto-converts to annual subscription. Each of these is a legal-risk + reputation-risk combo.
Ship instead: no-card-required free trial when the product can support it. Email reminder 3-7 days before trial converts. Clear cancel link in every reminder.
7. Forced consent / dark-pattern cookie banners
“Accept all” button highlighted; “Customize” buried in fine print. Pre-checked tracking consent. Forced login walls that block content until consent. Most jurisdictions now flag these.
Ship instead: “Accept” and “Decline” equally prominent. Default to NO tracking; let users opt in. Run analytics through privacy-first providers (Plausible, Fathom, Cloudflare Web Analytics).
8. Manipulative notifications
Push notifications styled as direct messages from a person (“Hey, are you still there?”). Badge counts that don't correspond to real content. Email subject lines starting with “Re:” when there's no prior thread.
Ship instead: notifications tied to genuinely useful events. If the user can't predict what each notification will contain, you're abusing the channel.
9. Manufactured FOMO
Countdown timers that reset when you reload. “Sale ends in 4 hours!” that runs daily for years. Limited-edition releases that aren't actually limited.
Ship instead: if there's a real deadline, state it precisely with a calendar date. If there's no deadline, don't create one.
10. Misleading social proof
Testimonials with no verifiable source. “Used by teams at Google, Microsoft, Amazon” — implying enterprise contracts when the actual users are individual employees on free tier. Star ratings averaged in a way that hides bad reviews.
Ship instead: testimonials with full name + role + linkable profile. Logos only where there's an actual business relationship.
11. Misdirection in pricing pages
The middle pricing tier highlighted as “most popular” with no data backing the claim. Decoy tiers designed solely to make the target tier look like a deal. Most users now recognize the “3-tier decoy” layout and discount it.
Ship instead: honest pricing tiers with real differentiation. If tier 2 is most-bought, the data should drive the recommendation badge — and if you can't show it, don't fake it.
12. Bait-and-switch features
Free tier launched, users adopt, then feature moved to paid tier. Pricing increased on existing subscribers without 30-day notice. Features removed under the guise of “simplification” that turn out to be paid in the next tier.
Ship instead: if you need to change tiers, grandfather existing users. 30-day notice on price increases. Removed-feature value returned via refund or credit if controversial.
Why dark patterns kill the AUG composite
Each dark pattern spikes one factor (usually Activation or Monetization) at the cost of multiple others (Retention, Advocacy). Because AUG composite is multiplicative (geometric mean × 10), a small lift in one factor cannot compensate for a large drop in another. The math:
- +1 Activation, -3 Retention, -2 Advocacy = composite drops ~30% even though Activation looks better in isolation.
- +2 Monetization (from sneak-in upsells), -2 Retention (from trust erosion), -3 Advocacy (from negative WoM) = composite drops ~50%.
The framework refuses to credit dark patterns even when they produce a short-term revenue line. The retention cliff downstream is mathematically larger than the activation spike upstream.
The harder truth
Dark patterns work in the very short term because cognitive biases are real. Users do click the highlighted button. Users do miss the pre-checked box. Users do feel pressured by the fake countdown.
But every user who feels manipulated leaves a trace: a negative review, a churn event, a refund request, a tweet, a Reddit thread. The cost compounds. Six months after a successful dark-pattern campaign, the product's organic search traffic includes terms like “[product] scam” and “how to cancel [product].” That traffic outranks your homepage. The Acquisition damage compounds.
Healthy products leave revenue on the table in the short term to compound trust for years. The math is brutal but the direction is clear.
Diagnose your dark-pattern exposure
Run the audit. If your AUG score is high in Monetization but low in Retention or Advocacy, you may be running dark patterns without realizing it. The framework will surface the weakest factor and point at the fix.