An AI flashcard study app: upload course slides, AI generates exam questions with answers, and an adaptive spaced-repetition algorithm schedules your reviews.
Retain is an AI flashcard app that started in the German-speaking world: upload PDF slides to get cards automatically, with an adaptive spaced-repetition algorithm scheduling reviews, pitched as "20% more efficient than Anki" and aimed at medical students and exam candidates who must memorize a lot (the homepage shows the crests of TUM and LMU in Munich). The domain was registered in April 2024 and an English-German version launched in May 2024. Its growth engine is the "Creator Program": it recruits several hundred amateur creators, Retain sets up new TikTok/Instagram/YouTube study accounts for them and pays a fixed fee per post, with bonuses of up to €1,000 for hits. It says these accounts get over 10 million views a week and over 500 million in total. French was added in 2025-10, and France quickly became tied with Germany as the largest market. This "pay-per-post account matrix" is in essence buying traffic, just with the money paid to amateur content; that is why its traffic is almost all direct visits and brand search (people watch a video and then search retain cards).One-line growth story, compiled by us from public sources; speculation is marked separately below.
The main engine is a paid creator matrix: it pays hundreds of ordinary people per post to run study accounts on TikTok, Instagram and YouTube, and people who watch come to search retain cards. Organic search is the largest channel, and the top terms are all branded keywords.
The lubricant is the review plan: a spaced repetition algorithm schedules reviews around exam dates, students have to come back to go through cards on schedule, and they view an average of 9.8 pages and stay 3 minutes 34 seconds per visit.
By estimate, about 40% of visits are existing users coming back to review on schedule; about 7.8 clicks to the checkout page per 100 visits.
Model: Lenny's racecar growth framework. Engine = a paid creator matrix (paying ordinary people per post to make videos on study accounts, which is essentially buying traffic with money paid for creator content); lubricant = the spaced repetition review plan built around exam dates, which brings students back; deck sharing among classmates is small-scale viral spread.
Real data: 165,700 visits, +47.89% month over month, the largest channel organic search at 45.55%, top three channel ranking (organic search, direct, referrals), organic 93.68% and paid 6.32% within search, 207 search terms in total, top search terms (retain cards, retain, retrain, retain cards gratuit, metodo retain), countries (Germany 25.51%, France 24.27%, Switzerland 13.83%, Austria 7.18%, the Netherlands 6.77%), bounce rate 32.34%, 9.78 pages per visit and average time on site 3 minutes 34 seconds come from the Similarweb public page, last month; the share of the second and third channels is not given on the public page, and referring sites and social sources are not listed either. 12,894 visits (about 12,900) to the Stripe checkout page come from quietmoney checkout page data.
Supporting evidence: The student count the site reports rose from 100,000+ to 300,000+ to 500,000+ to 1 million+ (Wayback snapshots); the creator program page says hundreds of creators, 10 million+ views a week and 500 million+ views in total; the official TikTok calls itself the main account (934 followers), which means there are many sub-accounts; the homepage shows the crests of TUM and LMU in Munich; after French was added France quickly tied with Germany as the largest market, and the branded keywords include the French word gratuit and the Italian word metodo.
Inference: The two channel circles are real, and how each path is split is an inference: organic search 75,500 = searching the name after seeing study account videos 45,000 + existing users searching the brand name 25,000 + searching the name after receiving a deck link 5,500; other channels 90,200 = existing users opening directly 45,000 + deck links from classmates 25,200 + links on video profile pages 20,000. The 1% visit after seeing a video and the 40% who open a deck link they received are guesses. Semrush desktop data is only used to judge that direct visits are the bulk of the second circle, and is not treated as a sitewide share.
Checkout: checkout.stripe.com
“Easily share your decks and flashcards with just a link. Anyone who has this link will be able to clone your deck.”Quote from the site
New-user source: Viral (Built-in sharing). Traffic is all direct visits (71.3%) and brand search (28.7%, "retain cards", "retain cards gratuit", "metodo retain"); new users are classified as viral: users can share decks by link and others clone them in one click; the homepage also has a Creator Program. There is no referral or social source data, and ChatGPT brings a small 0.06% of traffic. Returning users come back directly to review on schedule.
Semrush source breakdown (desktop)
| Attributable new users, last month | Share |
|---|---|
| 100% |
Direct + brand search 100%. Direct and brand search count as viral only when the site shows sharing mechanics; the largest share wins, under 5% means Other.
| Country | Share of visits |
|---|---|
| Germany | 25.51% |
| France | 24.27% |
| Switzerland | 13.83% |
| Austria | 7.18% |
| Netherlands | 6.77% |
Countries (Semrush): Germany 46.5%, France 16.2%, Austria 6.6%, Switzerland 4.6%, Argentina 4.4%
Multilingual: mainly English, plus German, French, Polish, Spanish, Portuguese, Italian, Japanese, Turkish and Hindi
| Source | Share |
|---|---|
| ChatGPT | 0.06 |
AI referrals (Semrush): ChatGPT 80 (0.06%)
207 keywords · brand ~100% · organic 93.68% / paid 6.32%
| Similarweb top 5 keywords |
|---|
| retain cards |
| retain |
| retrain |
| retain cards gratuit |
| metodo retain |
| Category | Share of referrals |
|---|---|
| Computers Electronics and Technology - Other | 5.5% |
| Education | 1% |
About 12,894 last month (Similarweb estimate), #270 of 606; the month before was 4,712. Opening checkout is not paying, so this is not revenue.
New-user source: Viral (Built-in sharing). The site has invite or share mechanics, and direct plus brand search is the largest source Traffic is all direct visits (71.3%) and brand search (28.7%, "retain cards", "retain cards gratuit", "metodo retain"); new users are classified as viral: users can share decks by link and others clone them in one click; the homepage also has a Creator Program. There is no referral or social source data, and ChatGPT brings a small 0.06% of traffic. Returning users come back directly to review on schedule.
Free to sign up and use; paid prices not found
Dossier confidence: medium. Parts 1 and 2 come from reading the site and Similarweb data; part 3 from public reporting, founder posts, Wayback snapshots, WHOIS and similar, with speculation marked.
The number: checkout visits is Similarweb's estimate of visits from the site to a checkout payment page; opening checkout is not paying, so this is not revenue. Small sites land on fixed buckets.
Data through last month. Page updated 2026-10-31.