Qaf · Monetization Strategy · Confidential Working Draft

Charging for the product, never for the deen:
a monetization plan for Qaf

How to brand, price, and place Qaf's paid tier — and how to launch it around Ramadan 2027 — grounded in Qaf's own beta analytics and verified market evidence.


July 8, 2026  ·  Prepared for the Qaf leadership team  ·  Sources: Qaf Mixpanel (project 4009978, 30-day window ending Jul 7) and 13 research workstreams; 59 load-bearing external claims independently fact-checked (49 confirmed, corrections applied)

Contents
  1. The recommendation in brief
  2. What the beta data says
  3. The economics of free
  4. Branding & framing
  5. Pricing & plan architecture
  6. Paywalls & trials
  7. Referrals
  8. Launch sequencing & Ramadan
  9. Financial model
  10. Risks & 90-day actions
  11. Appendix: method & sources

01The recommendation in brief

Direct answers to the five questions asked, with the evidence behind each in the sections that follow.

Should we brand monetization as "supporting the project"?

Yes as the wrapper, no as the substance. Sell a real product (Qaf Pro) whose paid value is capability — a deep-research answer mode, not permission to ask — and wrap it in honest mission language: "Your membership keeps Qaf free for a million seekers of knowledge." Never use the words donate, donation, sadaqah, or zakat for payments to a for-profit: Apple prohibits for-profit donation flows in-app, a company is not zakat-eligible, and implied-charity framing invites a Muslim Pro-scale trust collapse when users remember Qaf is a company. Publish a Hallow-style "Why does Qaf charge?" page and real cost transparency instead.[7][10]

How many plans?

One paid plan at launch — Qaf Pro at $9.99/month or $59.99/year (annual pre-selected) — plus two zero-build-cost wrappers: a one-time Founding Member offer for the beta cohort (~$199 lifetime, fair-use capped) and a Tarteel-style pay-what-you-can hardship program that answers "never forced to pay" better than any price cut. Add a Family plan (~$99/yr, 5–6 seats) as the first fast-follow. Paywall data consistently shows fewer visible options convert better; a second feature tier is only worth revisiting once Deep Study usage data justifies a ladder.[3][12]

Is "2 weeks of Pro per successful invite" right?

Yes — ship it, with three guardrails. It sits between Duolingo (1 week) and Perplexity (1 month); at ~$0.01/question it costs $0.02–$0.35 per grant against a $30–80 industry cost to acquire a subscriber. Make it two-sided and gift-framed ("give your friend two weeks of Pro"), credit it only after an activation event (invitee asks 3+ questions within 7 days), and cap it (~24 weeks/user/year). One prerequisite from your own data: referral attribution is currently broken — 51 users shared links last month and only 1 attribution was recorded.[13][14]

Where do the paywalls go?

Never in front of an answer. Free tier gets a monthly meter (60–90 questions/month — untouched by 91–95% of active users) that degrades gracefully to a lighter model rather than blocking. Paywalls appear at three moments: a skippable multi-page intro of Pro at the end of onboarding (~50% of all conversion volume industry-wide), the moment the meter approaches, and the moment a question deserves Deep Study. Your guest wall already proves the mechanic: 228 guests hit the 3-question wall last month — that moment, moved up-market, is the paywall.[4][5]

Trials, or pay right away?

Both, in a specific shape: a reverse trial plus a native intro trial on annual only. Give every new user 14 days of full Pro automatically, then downgrade to free — no card, no dark pattern, and it neutralizes trial abuse (Toggl doubled premium revenue this way). Keep a 7-day card-on-file trial on the annual plan via App Store / Play intro offers, and require a card on web trials. Promise — and send — a reminder before any charge: Blinkist's transparent trial lifted trial starts 23% and cut complaints 55%, and billing honesty is amanah for this audience.[6]

The unifying principle: Qaf sells scholarship-grade research tools and capacity — it never sells access to Islamic knowledge. Every pricing, paywall, and branding decision below is that one sentence applied.

02What the beta data says

Mixpanel, 30-day window ending July 7, 2026. 2,500-person open beta; volume became meaningful June 8.

Monthly question-askers
1,662
66% of the 2,500 beta — enthusiast-level activation
Questions asked / 30d
39,284
mean 23.6 per asker · median 4
Week-2 cohort retention
52%
plateaus near 38–43% — a real habit
Volume from top 8.9%
66%
148 users at 60+ questions/month

Usage is a textbook power law, and that is the entire free-tier problem. The median asker sends 4 questions a month; the 95th percentile sends ~88; the single heaviest user sent 1,571. The top 1% of askers generate roughly a third of all serving cost, and the bottom half generates 4%. A pricing design that treats "the average user" as the unit of analysis will be wrong on both ends.

Exhibit 1

Half of users ask fewer than 5 questions a month; 9% of users create two-thirds of the cost

Distribution of questions per asker, trailing 30 days (n = 1,662). Highlighted: the 60+ bucket — 148 users, ~26,100 questions.

The 60+ bucket averages ~171 questions/user/month (p95 ≈ 88, p99 ≈ 255, max 1,571). Source: Mixpanel frequency-per-user histogram, chat_message_sent.

The habit is weekly, not daily. Askers are active ~2.2 days per week (DAU/WAU ≈ 0.31) and usage inside an active day is bursty — median 3, p90 15. People sit down and research a question. This favors a monthly allowance over a daily cap (a daily cap punishes exactly the natural burst behavior while missing the every-day whale), and it favors trials of 14+ days, which span 4+ usage occasions.

Exhibit 2

The beta grew ~10x in eight weeks, and cohorts keep coming back

Weekly unique askers and questions asked (left); average cohort retention by weeks since first question (right).

Weekly askers (final partial week omitted).

% of cohort asking again, by week. Avg across weekly cohorts.

Payers and cost live in different places. US/UK/Canada supply 36% of askers (591 of 1,662) but only ~25% of question volume — Western users actually ask ~30% fewer questions than average. The heavy tail lives in Pakistan (51.6 q/user), India, and European diaspora pockets. Two consequences: the free tier is largely a da'wah subsidy to the rest of the world (which makes the support story true), and upgrade prompts fired at cap-hitters will mostly reach low-willingness-to-pay geographies — so forecast conversion from heavy users in payer geos only.

Exhibit 3

The UK, not the US, is Qaf's largest market today — and payer geos are lighter users

Monthly askers by country, top 12. Highlighted: US / UK / Canada — the expected payer core (36% of askers, ~25% of volume).

Platform mix: iOS 48%, Android 38%, web ~18% of askers (multi-device overlap). Caveat: ~50% of message volume reports platform=undefined — fix before per-store pricing analytics.

Three found assets, one found bug. The guest experience already has a working conversion wall (3 questions/day; 228 guests hit it last month and the next step is sign-in). A referral screen already exists (270 viewers → 51 sharers last month) — but only one referral was ever attributed, so attribution is effectively broken and must be fixed before the referral program launches. And citations are the product's felt differentiator: 37% of askers opened a citation last month — the thing free AI chatbots don't give this audience is exactly the thing to build Pro around.

03The economics of free

Modeled at 1M registered users using the beta's observed usage distribution (Pareto tail α ≈ 1.5), across monthly-asker scenarios of 150k / 250k / 400k and the beta's own 66% activation as the stress case.

At today's ~$0.10 per question, no free tier survives contact with 1M users. Uncapped, realistic scenarios burn $300–960k per month. Even a stingy 15-questions-per-month cap costs over $1M a year at modest activation. The entire monetization design is therefore sequenced on the cost curve: generosity is a function of unit cost.

Exhibit 4

A 60–90 question monthly meter keeps free serving inside the paid-revenue envelope — at $0.01 per question

Annual free-tier serving cost at 1M registered users, by free-tier monthly cap and monthly-asker count, at $0.01/question. Recommended zone highlighted.

Free cap / month% of askers ever capped150k askers250k askers400k askers660k (beta-rate)
Uncapped0%$426k$709k$1.14M$1.87M
300 (≈10/day)0.8%$343k$572k$915k$1.51M
90 (≈3/day)4.8%$273k$455k$728k$1.20M
60 (≈2/day)8.9%$238k$396k$634k$1.05M
30 (≈1/day)16.4%$175k$292k$467k$771k

At today's $0.10/question, multiply every cell by 10. Paid-revenue envelope at 10k subscribers ≈ $735k/yr net (Section 09). Model: beta usage distribution with capped Pareto tail; caps rarely bind below the 90th percentile.

The honest cross-subsidy arithmetic — which is also the marketing copy. At the recommended pricing, one Pro member nets Qaf ~$6.10/month after store fees. At $0.01/question that funds ~600 free answers every month — a month of Qaf for somewhere between 25 heavy users and 150 median users. "One membership keeps dozens of students of knowledge asking freely" is not framing; it is the unit economics, and publishing it is the strongest honest version of the support story.

The sequencing rule this forces

Tie the public generosity of the free tier to realized unit cost, not to the calendar: launch the meter at 30–45 questions/month while cost per question is above ~$0.03, move to 60–90 as it approaches $0.01, and advertise increases loudly (generosity ratchets up; it must never visibly ratchet down). Route free-tier questions to efficient models first — the flagship model is a Pro feature, graceful degradation is the free tier's safety valve.

04Branding & framing: product substance, mission wrapper

The question isn't whether support-framing works — it's what it must be wrapped around, and which words are radioactive.

What the evidence says

Say / do

  • "Qaf Pro" — a product tier with felt capability (Deep Study, voice, saved research)
  • "Your membership keeps Qaf free for a million seekers of knowledge" — published cross-subsidy math
  • A "Why does Qaf charge?" page: no ads ever, we never sell your data, questions stay private — payment as privacy protection
  • Pay-what-you-can hardship program (Tarteel "Alim" model) + free Pro for imams, teachers, students of knowledge
  • Choose-your-amount supporter pricing on the web checkout only ($5/$10/$20, same benefits)
  • Annual transparency note: what it costs to run Qaf (Signal precedent)

Never

  • "Donate", "donation", "sadaqah", "zakat" for payments to the company — App Review risk, fiqh risk, trust risk
  • Ads, anywhere, ever — the category's #1 anger trigger and Qaf's cleanest wedge vs Muslim Pro
  • Urgency theatrics ("Qaf will shut down without you") — Wikipedia's banner overreach triggered an editor revolt
  • Gating Quran/hadith text, citations, or the ability to ask at all
  • Retroactively paywalling anything beta users have today (Strava's 2020 lesson)
  • Surprise trial-to-paid conversion — Tarteel's single biggest review complaint

Position: "Qaf is free for the ummah, and it always will be. Pro members get scholarship-grade research tools — and they're the reason it stays free for everyone else." Both halves must be literally true, and Section 03 shows they are.

05Pricing & plan architecture

Users will benchmark Qaf against Hallow, Tarteel, and Muslim Pro — not against ChatGPT. Price in the faith-app band, differentiate with AI capability.

Exhibit 5

$59.99/year sits in the proven faith-app band, at half the price of the $20/mo AI assistants

Annual subscription price, US, mid-2026 (verified against official pricing pages). Highlighted: proposed Qaf Pro.

Tarteel also sells a $250 lifetime; Hallow adds Family $119.99 (6 seats); Muslim Pro's low annual reflects an ad-supported free tier. RevenueCat's cross-category "structural anchor" is $9.99/month.

The recommended architecture

Qaf — Free

$0
forever · no ads · no data sales
  • Cited answers to everyday questions — 60–90/month meter (start 30–45 until unit cost ≤ ~$0.02)
  • Graceful degradation past the meter: lighter model, never a locked door
  • All source texts & citations, always
  • Earnable Pro time via referrals
The one paid plan

Qaf Pro

$59.99/yr
"$4.99/month, billed annually" · pre-selected · or $9.99 monthly
  • Deep Study mode — long-form scholarly answers, expanded evidence chains, cross-madhhab comparison (the hero feature and the paywall moment)
  • Effectively unlimited everyday questions (fair use)
  • Voice answers, saved research & sync, priority capacity in Ramadan peaks
  • 7-day intro trial on annual (native IAP offer)

Two wrappers, zero new tiers

Founding Member
+ Hardship program
launch-window only · applications reviewed
  • Founding Member: one-time ~$199 lifetime Pro for the beta cohort (fair-use capped, grandfathered forever) — converts "support the project" sentiment into launch cash flow
  • Pay-what-you-can: full Pro free or any amount for those who can't afford it (Tarteel Alim model) — the honest answer to single-global-price exclusion
  • Fast-follow: Family plan ~$99/yr, 5–6 seats

Why these numbers

Billing rails: the margin is on the web, the volume is in the stores

Exhibit 6

What each rail nets on a $9.99 subscription — and why not to over-steer

RailFeeNet to QafNote
Stripe web (with Billing)≈6.6%$9.33Fixed $0.30 fee penalizes cheap monthlies — default web checkout to annual
Apple IAP — Small Business Program15%$8.49Applies under $1M/yr proceeds; at the recommended mix Qaf stays eligible until ~15–20k subscribers
Google Play (from Jun 30, 2026)15%$8.49External-offer link-outs save only the 5% billing fee — Android steering isn't worth the friction
Apple IAP — standard30%$6.99The post-$1M world; plan for it in year 2

US iOS apps may link to external checkout at 0% today (post-Epic contempt ruling) — but the Ninth Circuit has authorized a future cost-based fee, so treat it as temporary. Web checkout converts at only ~63–67% of IAP, so while on the 15% rate, hard steering loses net revenue: ship IAP as the primary CTA with a same-price "subscribe on the web" secondary link (US only; Canada must stay IAP-only in-app; UK likely opens late 2026–2027).[15]

Two rail-specific notes. First, the choose-your-amount supporter checkout and any "support Qaf" page must live web-only — for-profit fundraising flows are prohibited in iOS apps. Second, implement referral Pro-time and all entitlements server-side, independent of either billing rail, so earned weeks work identically for IAP and Stripe subscribers.

06Paywalls & trials: convert at moments of value, never at moments of need

Freemium costs Qaf the hard-paywall conversion premium (10.7% vs 2.1% median download-to-paid) by design. The compensation is placement discipline: capture the Day-0 window without ever blocking an answer.

The four placements (in priority order)

  1. End of onboarding — skippable, multi-page, shown once. Onboarding drives ~50% of trial starts industry-wide because motivation peaks at install; multi-page paywalls convert 37% better than single-page; and ~90% of trial starts plus ~50% of all paid conversions happen on Day 0. Skippable respects the mandate; skipping the placement forfeits most of the conversion window. Target ≥80% of new users seeing it once — under-exposure, not over-exposure, is the documented failure mode (one Superwall client found only 40% of users ever saw a paywall; raising exposure to 80% lifted revenue roughly linearly).[4]
  2. The Deep Study moment. When a question deserves research mode — multi-source, cross-madhhab, long-form — offer it in-line. Intent-triggered capability paywalls convert at ~12–16% versus ~2% for generic walls; this is also the moment that defines what Pro is.[5]
  3. The meter moment. As the monthly allowance runs low ("12 of 75 questions left this month"), show progress and the upgrade path; past it, degrade gracefully to the lighter model with a quiet one-tap upsell. Your guest wall (228 hitters/month → sign-in) is this exact mechanic already working one funnel-step earlier.
  4. Engagement milestones. Duolingo's 7-day-streak users convert at 3–4x baseline; Qaf's analogs are a completed week of study, a saved research collection, a Ramadan streak. Every prompt: context-tied, one-tap dismissible, never a blind app-open interstitial.[4]

Exhibit 7

The paywall copy formula that fits both the data and the mission

Assembled from tested patterns: benefit-led bullets, goal-first framing (+27% trial starts), specific social proof above price, annual pre-selected with per-month framing, cancel-anytime above the fold, and the Blinkist reminder promise (+23% trial starts, −55% complaints).

SlotRecommended copy direction
Headline"Go deeper into the sources with Qaf Pro" — capability first, not "unlimited questions"
Mission line"Your membership keeps Qaf free for a million seekers of knowledge."
Benefits (4–6)Deep Study answers with full evidence chains · every madhhab, compared · voice & saved research · priority in Ramadan nights
Social proof"Join 12,000 members keeping Qaf free for everyone" (real number, above the price)
Price blockAnnual pre-selected: "$4.99/mo billed annually — save 50%" · monthly beside it
Trust footer"Cancel anytime. We'll remind you before any charge. We never sell your data."

Trials: the reverse-trial architecture

New users start in Pro for 14 days, then land on the free tier. This is the documented best fit for a generous-freemium mandate: loss aversion does the selling (10–40% conversion lift in reverse-trial case work; Toggl doubled premium revenue and killed trial-cycling abuse, since there's nothing left to farm). Qaf's weekly usage cadence argues for 14 days over 7 — that spans ~4 usage occasions. Cost is bounded: a p90 user's 14 Pro days cost ~$0.35 at $0.01/question.[6]

07Referrals: cheap, gift-framed, activation-gated

Subscription-time rewards are structurally fraud-resistant and nearly free for Qaf. The risk isn't cost — it's a reward nobody wants because Pro doesn't feel different yet.

Cost per 2-week grant
$0.02–0.35
at $0.01/q (median → p90 usage) vs $30–80 CAC per subscriber
Healthy share rate
5–15%
of users; k-factor 0.15–0.25 is good
Referred-user premium
+16–25%
LTV, with durably higher retention (J. Marketing, 2011)
Realistic ceiling
10–20%
of signups at maturity (Dropbox's 35% is the all-time record)

The spec

Two prerequisites from your own data

(1) Fix attribution first: 270 users viewed the referral screen and 51 shared last month, but exactly one referral_attributed event exists — the loop can't be measured, let alone rewarded, until install-time attribution works. (2) Ship the Pro delta first: free weeks of a tier whose only difference is "more questions you weren't running out of" is worthless currency (the Duolingo lesson). Deep Study must exist before referral rewards mean anything.

08Launch sequencing: Ramadan 2027 is the compounding event

Hallow earns roughly a quarter of its annual net revenue in the single post-Lent month. Ramadan (Feb 8 – Mar 9, 2027) is structurally stronger — 30 nights of engagement on a fixed calendar — and Islamic apps reliably see 1.6–3x download spikes.[9][16]

Now → Oct 2026
Build the paid foundation

Deep Study mode, monthly meter (remotely configurable), server-side entitlements across IAP + Stripe, referral attribution fix, platform-tracking fix, Van Westendorp + MaxDiff survey on the beta cohort (n ≥ 200, oversample US/UK/CA).

Nov – Dec 2026
Soft-launch paid

Quiet rollout: debug billing on both rails, publish "Why does Qaf charge?", open the Founding Member offer to the 2,500 beta users, start the reverse trial for new signups, enroll in Apple's Small Business Program. Honest baseline prices — no discounts yet.

Sha'ban · Jan 9 – Feb 7, 2027
The Tarteel move: free Pro month + challenge enrollment

Pro free for all during Sha'ban (no card), enrollment opens for the Qaf Ramadan challenge (khatam / daily-learning format with named scholars). Habit forms in the prep month; the paid decision lands during Ramadan at peak intent — not after Eid in the lull.

Ramadan · Feb 8 – Mar 9, 2027
Full campaign, annual-default paywall

Challenge live (free for everyone — the mandate holds), referral gifting push, annual plan pre-selected: an annual bought in Sha'ban or early-mid Ramadan renews inside the next pre-Ramadan window — the best possible renewal timing (Hijri drift makes this a 1–2 cycle bonus, so bank it early). Pre-scale serving for a Hallow-grade spike.

Shawwal onward · 2027
Post-Eid anchors, not post-Eid panic

Dhul-Hijjah 10-day challenge (~May), weekly Jumu'ah / al-Kahf habit loop, win-back automation for monthly churners (~20% reactivate within a year; annual churners almost never return, so pre-renewal engagement matters more). Do not read Shawwal churn as steady-state.

Launching paid after Ramadan 2027 would push the category's one compounding month into a window with 1.6–7x less traffic. The date is the strategy.

09Financial model: v1 monetization makes the free tier self-sustaining

At 1M users, 1% paid, recommended pricing ($9.99 / $59.99, ~60% annual mix, ~70% of revenue via IAP at the 15% small-business rate).

Net revenue / subscriber
$6.13/mo
$7.00 gross ARPU − blended ~12.5% rail fees
Net ARR at 10k subs
$735k
+ Founding Member one-time cash (~$50k at 250 buyers)
Free-tier serving cost
$240–455k/yr
60–90/mo meter, 150–250k askers, $0.01/q
Paid-tier serving cost
$140–290k/yr
10k subs at ~120 q/mo, $0.01–0.02/q (Deep Study is costlier)

Exhibit 8

Conversion sensitivity: 1% covers the AI bill; 1.5–2% funds a business

Net annual subscription revenue at 1M registered users vs total AI serving cost (free + paid, base case), at $0.01/question.

Serving band shown as reference rows in the table view. Above ~$0.03/question unit cost, every scenario is underwater — the launch gate is the cost curve, not the calendar. Upside levers not modeled: Supporter choose-your-amount uplift (Bandcamp's name-your-price runs ~+50% over the floor), Family plan, price at $69.99, reactivation.

Read it straight: at the base case, subscriptions (~$735k net) cover total AI serving (~$380–745k) with thin headroom — monetization v1 makes Qaf's free mission self-sustaining, which is exactly the support story, but it does not yet fund payroll. The realistic paths to margin: convert the Western cohort at 2–3% (they benchmark at 2.8% median; Duolingo shows values-driven freemium can double penetration over time), land the cost curve below $0.01, and let annual mix + Family + Supporter pricing lift ARPU. A 1.5% blended conversion at $69.99 annual is a ~$1.3M-net business on the same cost base.

10Risks, and the next 90 days

Exhibit 9

Top risks and their mitigations

RiskExposureMitigation
Unit cost stays > $0.03/questionFree tier structurally unaffordable at scaleGate generosity on realized cost; launch meter at 30–45/mo; route free tier to efficient models; remote-config caps
Ramadan spike hits costs before revenue25x traffic on day 1 of campaignPre-scale; free-tier depth (not access) tightens during spike; annual-default paywall converts the wave
"Paywalling the deen" backlashCategory-defining trust riskNever gate sources/answers; hardship program; no ads; no donation vocabulary; "Why we charge" page; never retro-gate beta features
AI-app annual retention (21.1% vs 30.7% non-AI)Year-2 revenue cliffHabit surfaces (saved research, study plans, Ramadan cycles); pre-renewal engagement; win-back for monthly churners
Apple fee cliff (15% → 30% past $1M proceeds)~$0.9k/yr per 1k subsAnnual + web mix keeps Qaf under threshold until ~15–20k subs; grow web-native acquisition before the cliff
Meter converts the wrong geographyCap-hitters skew non-payer geosForecast from payer-geo p95 users only; treat ROW over-cap volume as mission subsidy, not lost revenue

The 90-day action list

  1. Fix instrumentation: referral attribution (1 event vs 51 sharers) and platform=undefined (~50% of message volume) — both block everything downstream.
  2. Run the WTP survey on the beta cohort (Van Westendorp + MaxDiff; n ≥ 200; oversample US/UK/CA; test Deep Study, madhhab depth, voice, offline, saved research, support framing) — the cheapest remaining evidence gap.
  3. Build the meter + Deep Study prototype — the meter remotely configurable, Deep Study as the felt Pro delta that referrals and paywalls both depend on.
  4. Stand up billing rails: Stripe + IAP with server-side entitlements, Small Business Program enrollment, transparent-trial mechanics (pre-charge reminder).
  5. Write the two trust artifacts: "Why does Qaf charge?" and the privacy commitment ("your questions are never sold, never ads") — both belong on the paywall itself.
  6. Draft the Founding Member offer for the beta cohort and the Sha'ban 2027 campaign calendar backward from Feb 8, 2027.

11Appendix: method & sources

Internal data: Mixpanel project 4009978, trailing 30 days ending Jul 7, 2026 (events: chat_message_sent, chat_rate_limit_hit, referral_*, app_opened; frequency-per-user histograms and retention curves). Beta caveats: opt-in enthusiast cohort (66% monthly activation will not hold at scale); signed-in users currently uncapped, so usage reflects true unconstrained demand; iOS store attribution sparse. External research: 13 parallel research workstreams (Islamic apps, faith apps, AI assistants, subscription benchmarks, paywall placement, referrals, support-framing, trials, price architecture, billing rails, usage economics, Pro contents, Ramadan timing) executed with live web research in July 2026; 59 load-bearing claims were then independently re-verified against primary sources — 49 confirmed, 8 corrected or updated (corrections applied throughout, e.g., Hallow's #1-overall App Store moment was Feb 2024 not 2025; Perplexity's give-get referral program ended Oct 2025; Tarteel's current App Store monthly is $12.99 with ~$99 annual), 2 unverifiable (excluded).

Principal sources

  1. Menlo Ventures, "2025: The State of Consumer AI" (Jun 2025) — ~3% of consumer AI users pay; ChatGPT ~5%.
  2. OpenAI/ChatGPT, Anthropic, Perplexity, Google, xAI official pricing pages (retrieved Jul 2026); The Information (Apr 2026) on ChatGPT Go's 122M-subscriber projection.
  3. RevenueCat, State of Subscription Apps 2026 (115k apps, $16B) — freemium 2.1% vs hard-paywall 10.7% D35 conversion; trial-length curves; $9.99 anchor; churn/reactivation data. Adapty, State of In-App Subscriptions 2026.
  4. Superwall paywall research (2025–26) — onboarding ≈50% of trial starts; multi-page +37%; exposure 40%→80% study. Duolingo investor materials (Q4 2025: 9.2% MAU penetration, 12.2M subscribers).
  5. Adapty 2026 benchmarks — value-moment paywalls ~2.1x trial-start rate; intent-triggered capability paywalls ~12–16% conversion.
  6. Elena Verna (Amplitude) on reverse trials (10–40% lift); Toggl case; RevenueCat/Adapty trial-length data; First Page Sage card-required vs no-card (48.8% vs 18.2%); Stripe trial-abuse data (2025–26); Blinkist transparent-trial case (Growth.Design).
  7. Muslim Pro support center & App Store listings (Premium $12.99/mo, $34.99/yr; AiDeen 3-questions metering); Vice/Motherboard (Nov 2020) X-Mode investigation; FTC order (Apr 2024); Tarteel pricing & Alim program (tarteel.ai, retrieved Jul 2026); Pillars; Quranly; Quran Pro; Ansari (TIME, May 2026); Fanar (QCRI).
  8. Reuters Institute Digital News Report (values vs access motivations); Duolingo mission-positioning; Gneezy et al., Science 329 (2010) shared-social-responsibility experiment; Bandcamp name-your-price data.
  9. Hallow: help.hallow.com pricing (Mar 2026), "Why do we charge for Hallow Plus?" (Alex Jones), Appfigures "The Most Predictable Spike in the App Store" (Feb 2026) — ~$40M net 2025, ~25x Ash Wednesday spikes, seasonal trials; Pray.com; Glorify; YouVersion (Premier Christianity, Nov 2025); Sefaria.
  10. The Guardian / GMG FY2024–25 & FY2025–26 reader-revenue reports (~1.33M supporters; £125M digital reader revenue); Wikipedia Fundraising Dec 2025 campaign roundup; 2022 English Wikipedia banner RfC; Signal cost transparency posts.
  11. IslamWeb fatwa 343831; SeekersGuidance on payment for teaching; Sahih al-Bukhari ("The most deserving thing for which you take payment is the Book of Allah"); Quran.Foundation waqf structure; Indiana University Muslim Philanthropy Initiative (US Muslim giving ≥ $4.3B/2021).
  12. HBR, "The Good-Better-Best Approach to Pricing" (2018); Anderson & Simester $9-endings field experiments; Ariely decoy experiment; RevenueCat paywall case studies (middle-plan removal +64% revenue).
  13. Perplexity Help Center referral terms (give-a-month program, ended Oct 2025; Comet $20 program with first-question activation); Duolingo referral terms (~1 week, ~10-week cap).
  14. Schmitt, Skiera & Van den Bulte, J. Marketing 75 (2011) — referred-customer value; Airbnb Referrals 2.0 (altruistic framing); Dropbox referral history (100k→4M, 35% of signups); GrowSurf/ReferralCandy fraud & k-factor benchmarks.
  15. Epic v. Apple contempt ruling (Apr 2025) & Ninth Circuit modification (Dec 2025); Apple Guidelines 3.1.1/3.2.1–3.2.2 (donations); Apple Small Business Program; Google Play fee structure eff. Jun 30, 2026 (Epic–Google settlement); UK CMA SMS process; RevenueCat Dipsea IAP-vs-web A/B; Superwall IAP vs web-checkout conversion data.
  16. Data Darbar (2023) Ramadan install multiples for Islamic apps; AppsFlyer "Ramadan Effect" (2024–25); Muslim Pro "40 Days of Deen" / "Ramadan Re:Charge"; Tarteel free-Sha'ban campaign (Feb 2025); Hijri-drift renewal analysis (Ramadan 2027: ~Feb 8 – Mar 9).