Privacy
Policy
Last updated: 23 October 2026
1. Overview
This privacy policy explains how I collect and process personal data on smartergerman.com and the services at courses.smartergerman.com, learn.smartergerman.com, bot.smartergerman.com, conversation.smartergerman.com, assess.smartergerman.com, write.smartergerman.com, articles.smartergerman.com, community.smartergerman.com and the tool pages linked from my Toolbox, for example wizard.smartergerman.com and dative.smartergerman.com.
Controller:
Michael Schmitz, Friedelstraße 38, 12047 Berlin, Germany
Email: [email protected] | VAT-ID: DE288541978
Data I receive from third parties
Some data reaches me from others rather than from you: Teachable (account, enrolment and progress data of my course platform), Stripe (payment status and billing details) and Facebook Lead Ads (the name and email you submit in a lead form there). This policy covers that data in the same way.
Adults only
My courses, my learning app and my AI practice tools are intended exclusively for adults. Under my terms of use, you must be at least 18 years old to create an account, buy a course or use these services. If I learn that a person under 18 is using an account, I close the account and delete the personal data, unless I am legally required to keep it.
2. Hosting
Main Website
My website is hosted on my own server at IONOS SE, Germany.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR)
Teachable course platform
My courses at courses.smartergerman.com are hosted on Teachable. For users in the EEA, Teachable names Hotmart B.V., Spuistraat 168, 1012 VT Amsterdam, Netherlands, as its controller. Teachable processes account and contact data, payments through its payment providers, course progress, communications and technical usage data to provide the course platform.
Teachable, not SmarterGerman, selects and operates Wistia for course-video delivery and Heap for platform analytics. I do not integrate those two services independently. My own learner community is described below. Teachable determines the concrete data categories, purposes, retention and international-transfer safeguards for that platform processing; those details and the current providers are in Teachable’s privacy policy and subprocessor information.
Legal basis for my use of the course platform: Contract performance (Art. 6(1)(b) GDPR). Teachable describes its own legal bases and transfer safeguards.
Teachable Privacy Policy
Teachable cookies
Teachable controls the cookies sent by the course platform at courses.smartergerman.com. When a course or checkout page is requested, Teachable may set session and platform cookies such as _session_id and site_preview, visit-analytics cookies such as ahoy_visitor and ahoy_visit, and affiliate-attribution cookies such as affiliate_code, _afid and aid. Observed lifetimes range from the session to two years. Some are sent with the initial platform response, before a Teachable cookie choice is made. The consent choice on smartergerman.com controls my own marketing storage but cannot prevent cookies sent directly by Teachable. Teachable describes its cookie categories and available choices in its Cookie Policy and provides its own cookie interface on supported checkout pages.
Teachable Cookie Policy
Community
My learner community is part of my learning app at learn.smartergerman.com and runs on my own server. Course learners use it with their learning account. Anyone else can apply through the form on my homepage: I use the application, with your name, email address and answers, only to decide on admission, and I delete every application 90 days after it arrives. If I admit you, my server creates a learning account for your email address that opens the community only. The community stores your profile, your posts, replies and likes; other members see your posts under your public name. When your account is deleted, your posts and replies are deleted with it.
Until 23 October 2026 the community ran on Circle (CircleCo, Inc., USA) under my subdomain community.smartergerman.com, which now forwards to the learning app. What was written there stays stored with Circle until I close that account, and I keep an archive copy of its posts and comments on my own server. Transfers to Circle rest on the Standard Contractual Clauses in the Circle data processing agreement.
Legal basis: contract performance and steps before a contract (Art. 6(1)(b) GDPR).
Circle Privacy Policy
Cloudflare
Cloudflare protects the services and delivers content. It processes IP addresses, request and header data, device and browser signals and security events for delivery, attack detection and abuse prevention.
Provider: Cloudflare, Inc., USA. International transfers are covered by Cloudflare’s Data Privacy Framework participation and contractual safeguards, including Standard Contractual Clauses where applicable.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR) in secure and reliable delivery.
Cloudflare Privacy Policy
Bunny.net (video delivery and encrypted backups)
Course videos in my learning app and on some course pages are streamed through Bunny Stream. When a video loads, Bunny receives your IP address, browser and device information and playback data, for example which video was requested and when, to deliver the stream and to protect it against abuse. Bunny also stores my server backups; they are encrypted on my server before upload and Bunny cannot read their content.
Provider: BunnyWay d.o.o., Dunajska cesta 165, 1000 Ljubljana, Slovenia (EU). Bunny processes this data on my behalf (Art. 28 GDPR).
Legal basis: Contract performance (Art. 6(1)(b) GDPR) for delivering course videos; legitimate interest (Art. 6(1)(f) GDPR) in reliable delivery and disaster recovery.
Bunny Privacy Policy
Amazon Web Services (encrypted database backups)
The databases of my learning app and of the Listening Lab are additionally backed up continuously to Amazon S3 in eu-central-1, Frankfurt, so that they can be restored after a server failure. These backups contain the account and learning data stored in the learning app and in the Listening Lab, including your email address and practice progress. They are encrypted on my server before upload; the key needed to read them is kept neither on my server nor at Amazon, so Amazon cannot read their content. Each backup is kept for up to 42 days and then deleted automatically.
Provider: Amazon Web Services EMEA SARL, Luxembourg. AWS processes this data on my behalf (Art. 28 GDPR).
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR) in data security and disaster recovery.
AWS Privacy Notice
Country-based payment routing
I use the country code provided by Cloudflare (based on your IP address) to display the payment platform appropriate for your region and to comply with tax obligations. Your IP address is not retained for this purpose.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR)
Checkout security and trial-abuse prevention
When you start or complete a checkout, I process your contact details, account and checkout identifiers, IP address, country, timestamp, and technical request data to prevent fraud, repeated free-trial abuse, and unauthorized access. Several warning signs can generate an internal alert for manual review. An unusual name or IP address alone does not cause an automatic block. If an account was already blocked after human review, a later zero-cost trial may be canceled automatically and course access will not be granted.
The internal sales dashboard displays checkout data already held in secured payment and consent records. It does not create a second IP database. Security evidence is kept only as long as needed to investigate abuse, enforce or defend legal claims, and meet legal recordkeeping duties. Access is restricted to the controller.
Legal basis: Legitimate interest under Art. 6 para. 1 lit. f GDPR in fraud prevention, service security, and enforcing the one-trial rule.
Right to object: email [email protected].
Stripe Checkout
When you buy through my Stripe checkout, for example as a buyer in the EU or EEA, my server sends you to a payment page hosted by Stripe. You enter your payment details there, never on my website. Stripe receives your name, email address, billing address and the payment details you enter, and processes your IP address and device data for fraud prevention; card data never reaches my server. I receive the payment status, your name, email address, billing address and invoice data to grant access and, for sales outside Managed Payments, to issue invoices and meet tax obligations. Stripe also provides the customer portal in which you manage your subscription. For payment processing, fraud prevention and its own legal obligations Stripe acts as an independent controller.
For buyers in some countries outside the EU, including Switzerland, Norway and Liechtenstein, I sell through Stripe’s Managed Payments service. Sold through Link, LLC, a Stripe company, is then the seller of record. Link sends your receipt, collects and pays the sales tax, and handles payment support and refunds. Your bank statement shows LINK.COM. Link’s own privacy policy, linked on the payment page, applies to this.
Provider: Stripe Payments Europe Limited, Ireland; Stripe, Inc., USA, is certified under the EU-US Data Privacy Framework.
Legal basis: Contract performance (Art. 6(1)(b) GDPR) and legal obligations (Art. 6(1)(c) GDPR, tax law).
Stripe Privacy Policy | Link Privacy Policy
Paddle checkout
For a purchase through Paddle, I send Paddle your email address, country code and selected offer. Paddle is the merchant of record and collects billing and payment details in its checkout. I receive payment, refund and purchase references to provide the purchased access and keep payment records. Paddle processes this information under its privacy notice.
Accounting and tax
Invoice and payment records (name, address, amount, date) go to my tax advisor and my bank as far as bookkeeping, tax filings and payment processing require it, and to the tax office on request. They are kept for the statutory periods, up to ten years.
Legal basis: legal obligation (Art. 6(1)(c) GDPR).
3. Cookie Consent
I use CookieConsent v3 by Orest Bida, a self-hosted cookie consent tool, to manage your tracking preferences. The proof record is stored on my own server in Germany and contains a consent ID, selected categories and services, policy and banner version, language, timestamps and the type of choice. New records do not retain your IP address, an IP hash or browser information. Older records can still contain browser information and a pseudonymous IP hash. Records are kept for three years to document permission during the analytics retention period and allow a limited period for handling disputes; this is my retention policy, not a fixed period required by GDPR. Your browser remembers your choice for 182 days. If a proof record cannot be sent, a pending copy is kept in necessary local storage and retried; it is removed after acknowledgement or on a later visit after 182 days.
Legal basis: Legal obligation (Art. 6(1)(c) GDPR)
You can change your preferences at any time:
4. Analytics & Marketing
Aggregate website traffic
I use existing aggregate Cloudflare request statistics to understand website demand and improve content and reliability. This adds no analytics script, tracking pixel, cookie or visitor identifier. My reports retain daily estimated successful HTML request totals, published public-page counts for 24 months. They do not retain IP addresses, visitor or session identifiers, query strings, full referring URLs or learner page addresses. Low-volume page categories are combined. The current Cloudflare plan does not expose referrer breakdowns. Cloudflare processes the underlying requests as my delivery and security provider, as described above. Common automated clients are filtered, but estimates are not counts of people. Legal basis for the limited processing used to produce these statistics: legitimate interests under Article 6 paragraph 1 point f GDPR. You may object by contacting [email protected]; already anonymous aggregate totals cannot be linked back to you.
Optional browser analytics - Plausible, self-hosted
Detailed Plausible analytics on my server in Germany runs only with your Analytics permission, in every country. The consent banner appears in every country when you have no current saved choice. Browser analytics stays off until you enable Analytics. You can change or withdraw your choice in Cookie Preferences. Plausible sets no analytics cookies. It briefly processes IP address and browser information to generate a daily-changing pseudonymous identifier; the raw IP address is not stored in Plausible events. Reports include pages, referring websites and campaigns, country and browser or device categories, scroll depth, time spent, outbound clicks and course or checkout events. No account name or email is sent. The learning app uses the same permission and excludes administrative and sensitive learner-record pages. Pronunciation tools and the Hyama reader send pageviews only. Event and session records are kept for 24 months. Hashing salts can remain for up to 72 hours; these records are pseudonymous, not immediately anonymous. Refusal, expiry, withdrawal and the browser objection stop future browser analytics. Legal basis: your consent under Article 6 paragraph 1 point a GDPR and section 25 paragraph 1 TDDDG.
Google Analytics 4 and Google Tag Manager
My website smartergerman.com loads neither Google Analytics nor Google Tag Manager, and my server sends no data to Google Analytics. My course platform at courses.smartergerman.com is run by Teachable, which loads its own Google Analytics tag there. Teachable controls that tag and the data it collects, and I have no access to that data. Details are in Teachable’s privacy policy.
Google Ads
I advertise my courses with Google Ads. My website loads no Google Ads tag, and neither my website nor my server sends Google Ads any conversion data about your visit, signup or purchase.
Provider: Google Ireland Limited, Dublin, Ireland.
Local attribution storage
My site keeps the first traffic source, medium, campaign and landing page in your browser’s local storage for up to 90 days, and the same values in the cookie __Host-sg_attr for the same period. In the configured marketing-consent regions, this happens only after you accept marketing storage. Before that choice, a source found on the current landing page is held only in page memory and is not written to or read from local storage or a cookie. In all other regions the values are stored automatically unless you refuse marketing storage in the consent settings. When you start a purchase in my Stripe checkout, these values are added to your Stripe checkout record and to my own sales records, so I can see which channels bring paying customers. In the configured marketing-consent regions, this happens only for a consent given after 17 October 2026, 15:00 UTC. The cookies __Host-sg_ftref and __Host-sg_ftflag then remember which of your recent checkouts carry the values. Adding them is optional and never affects your purchase. I delete them 14 days after a checkout you did not complete and 24 months after a completed one. Refusing or withdrawing marketing permission in the cookie settings removes the values in your browser at once and, within a day, the values added to your recent purchases from the same browser. For older purchases or from another device, write to me and I delete them.
Legal basis: Your consent in the configured marketing-consent regions (Art. 6(1)(a) GDPR and Sec. 25(1) TDDDG). In all other regions, my legitimate interest in knowing which channels bring paying customers (Art. 6(1)(f) GDPR).
Right to object, or to have the values deleted: email [email protected].
Partner commissions - affiliate attribution
Some partners recommend my courses and receive a commission when someone they sent buys. If you arrive through a partner link, my site stores the partner code and, where available, the referring website on your device, in the cookies __Host-sg_affcode and __Host-sg_affref and with the same values in local storage, for up to 360 days. In the configured marketing-consent regions, this happens only after you accept marketing storage; without that consent, cookie-based affiliate attribution is not stored or passed to Stripe. In all other regions it is stored automatically unless you refuse marketing storage in the consent settings. At checkout the partner code is added to your Stripe payment record so I can pay the partner. It stays with the payment and its invoices as evidence for the commission I pay, for the retention periods of German tax and commercial law, and I delete it from my records ten years after the end of the year of purchase, for a subscription ten years after the end of the year it ended. A personal coupon or your optional referral answer can provide separate evidence; your answer alone does not award commission. Affiliate cookies that Teachable sets on smartergerman.com are deleted once you answer the cookie banner without marketing consent.
Legal basis: Your consent in the configured marketing-consent regions (Art. 6(1)(a) GDPR and Sec. 25(1) TDDDG). In all other regions, my legitimate interest in paying partners for the customers they send (Art. 6(1)(f) GDPR). Keeping the partner code with the payment: my legal accounting obligations (Art. 6(1)(c) GDPR with Sec. 147 AO and Sec. 257 HGB).
Affiliate invoice portal
As an affiliate, you can save your business name, address, country, tax details and optional IBAN or Wise contact in your private dashboard. I use these details to prepare commission statements, let you issue invoices and arrange payment. Only you and I can access your portal records. Invoice creation uses a fixed PDF template on my server, with no AI service. Profile changes affect future invoices; issued invoices and their supporting records remain unchanged for accounting and are retained for the applicable statutory periods, up to ten years. Contact [email protected] to correct profile details or arrange an invoice correction. Legal bases: performance of the affiliate agreement and legal accounting obligations, Art. 6 para. 1 lit. b and c GDPR.
Optional learner recommendations
On your first learning dashboard visit, you can choose a partner, name a public website or person, choose No one, or skip the question. If you save an answer, I store it with your account and its date on my server in Germany. I compare it with referral links, personal coupons and payment records to review partner attribution. Partners see only their own aggregate figures, not your name, email or free-text answer. A recommendation does not change your price or access, overwrite an existing credited partner, or automatically award commission. Your answer stays with your account until it is deleted or corrected; financial decision records are kept separately with my accounting records. You can contact [email protected] to correct or delete your answer.
No Facebook or Meta tracking
I do not use the Meta Pixel, the Meta Conversions API, Facebook or Instagram lead forms, or any other Meta advertising tool, and I share no data with Meta. This is a deliberate choice: I would rather your data help you learn German than fund another $900,000 watch, so I do not advertise on their platforms. You may still reach this site from a Facebook or Instagram link, but nothing about your visit is sent back to them.
5. Plugins & Tools
Cloudflare Turnstile
Turnstile protects forms from spam and bot submissions by processing IP, browser, device, interaction and security signals. My active widgets use Turnstile’s default single-use verification token and do not issue the optional reusable cf_clearance cookie.
Provider: Cloudflare, Inc., USA.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR) in service security.
Cloudflare pre-clearance information
Video FAQ search and contact
The FAQ search runs in your browser. When no answer matches, the search text may be stored separately from contact messages for up to 30 days to improve the FAQ. Email-like text and long number sequences are rejected. Contact messages process your email address, message, and one optional screenshot so I can reply. In regions where consent is required, I also store a consent record containing a hashed email address, country, timestamp, the exact consent text, a monthly rotating IP hash, and browser information for up to three years. Contact records and unmatched searches are not linked. Storage is self-hosted in Germany.
Legal basis: Your consent where requested and my legitimate interest in answering enquiries and improving the FAQ.
5b. AI Conversation Practice
AI features: basics
All AI features are optional practice tools. None of them identifies you by your voice or face, and none makes decisions about you that have legal or similarly significant effects (Art. 22 GDPR). Where a tool gives feedback, points or a level, it rates the German in that exercise or test, not you as a person. I receive no mood or emotion values from any AI feature. Apart from the optional German Learning DNA questionnaire, no AI feature assesses your personality or behaviour, and I use no AI result to set prices, make offers or grant access.
AI conversation practice
The conversation tool sends your voice, transcript and conversation context to Google Gemini on Vertex AI in the EU, through Google’s EU multi-region, to run the practice conversation, give you feedback on your German and translate each finished transcript line into English, which you can show with a click. Google does not use this data to train its models. This service has an approved exception from Google’s abuse-monitoring prompt logging. On my German server I store your account email and internal identifier, session timing, topic, session summaries of your German, feedback, error categories and technical provider, token and cost data. Full conversation transcripts are not stored in the session-analytics table. Session analytics are kept for no more than 12 months. Markers of refunded AI credit purchases are kept without personal data (transaction id only) to prevent duplicate credits.
Provider: Google Cloud EMEA Limited, Ireland. Google uses Google LLC, USA, as a subprocessor for data-centre operations, maintenance and technical support; in doing so, Google LLC may access this data from the USA, also when EU endpoints are used. Such transfers rely on the European Commission’s adequacy decision for the EU-U.S. Data Privacy Framework, under which Google LLC is certified; where that decision does not cover a transfer, Google’s Cloud Data Processing Addendum provides for EU Standard Contractual Clauses. You can request a copy of these safeguards at [email protected].
Legal basis: Contract performance (Art. 6(1)(b) GDPR). A privacy notice is shown before first use.
Free level assessment
The level assessment uses a random session identifier and sends your typed or spoken answers, including writing text and transcripts, to Google Gemini on Vertex AI in the EU, through Google’s EU multi-region, to score the test. My German server stores item responses, scores and result data, session state and device information for up to 12 months. Raw writing text is removed from item-response records after 48 hours. Google does not train its models with this data. Google may process short-lived logs for abuse prevention or legal obligations under its service terms.
Provider: Google Cloud EMEA Limited, Ireland. Google uses Google LLC, USA, as a subprocessor for data-centre operations, maintenance and technical support; in doing so, Google LLC may access this data from the USA, also when EU endpoints are used. Such transfers rely on the European Commission’s adequacy decision for the EU-U.S. Data Privacy Framework, under which Google LLC is certified; where that decision does not cover a transfer, Google’s Cloud Data Processing Addendum provides for EU Standard Contractual Clauses. You can request a copy of these safeguards at [email protected].
Speaking part of the level assessment
Short read-aloud recordings are sent to Microsoft Azure Speech in the Germany West Central region for pronunciation scoring. Microsoft states that real-time speech and pronunciation-assessment audio is not retained or stored. My server keeps the transcript and pronunciation scores with the pseudonymous assessment session for up to 12 months.
Provider: Microsoft Ireland Operations Ltd.
Legal basis for the assessment and speaking part: Contract performance (Art. 6(1)(b) GDPR).
German Learning DNA quiz
The quiz at smartergerman.com/dna works without an account. A fixed calculation in your browser turns your answers into one of my learner types. For the personal part of the report, your learner type and your answers, without name or email address, are sent to Google Gemini on Vertex AI in the EU; Google does not use this data to train its models. My server in Germany keeps an anonymous statistic of each completed quiz: learner type, scores, country and device type, without email or IP address. If you ask for your report by email, my server stores your email address, country, newsletter choice and results, sends you a link to your report and deletes these records 12 months later. Anyone who has that link can open the report. You receive my newsletter only through the separate newsletter checkbox, see section 5c. To limit misuse, my server keeps a request counter under a hashed form of your IP address for up to two days. I do not use your learner type for prices, offers or access.
Provider: Google Cloud EMEA Limited, Ireland, with the safeguards described above.
Legal basis: Contract performance (Art. 6(1)(b) GDPR) for the report and the email; consent (Art. 6(1)(a) GDPR) for the newsletter; legitimate interest (Art. 6(1)(f) GDPR) for the anonymous statistic and misuse protection.
Important: Do not share sensitive personal data (passwords, financial info, health data, ID numbers) during AI conversations.
Model-training program - closed
Until 18 July 2026 learners could opt in to donate anonymized session recordings to train my own German speech-recognition and grammar models. This program is closed: recording stopped on 18 July 2026, all stored recordings and derived training data were deleted, and the opt-in checkbox has been removed from the tool. I keep the audit log of past grants and revocations (email, timestamp, exact consent text, IP address, user agent) for 5 years under Art. 7(1) GDPR; it is used for no other purpose.
Legal basis for the audit log: Legal obligation and legitimate interest (Art. 6(1)(c) and (f) GDPR).
Learning memory in the conversation tool
Self-hosted Whisper and LanguageTool on my German server derive grammar-error categories for adaptive practice. The ordinary learning-memory record contains the error category and rule linked to an internal identifier; a separate identity table maps that identifier to your account email. It normally does not store the flagged words. If you rate a correction Helpful or Wrong, the feedback record also contains the flagged text, session identifier and account email. These records are not used for model training. Learning memory is deleted after 12 months without a session unless you have explicitly asked me to keep it longer; correction-rating records are kept for no more than 12 months.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR) in adaptive practice and quality improvement.
Right to object and erasure: email [email protected].
Additional Conversation credits
When you make the optional one-off purchase of additional Conversation credits, I record the email address, timestamp, IP address, user agent, product identifier and the immediate-activation acknowledgement shown at checkout. This checkout record is stored on my German server for three years as evidence of the transaction and the information presented. It is not a separate consent to the Conversation tool’s AI processing.
Legal basis: Contract performance (Art. 6(1)(b) GDPR) and legitimate interest in establishing or defending legal claims (Art. 6(1)(f) GDPR). This paragraph describes data processing only; it does not remove any mandatory consumer right.
5c. Email Marketing & Lead Forms
Sendy subscriber database and Amazon SES delivery
The subscriber database for newsletters is self-hosted on my server in Germany. It can contain your email address, name if supplied, IP address, country, signup source and method, subscription and confirmation status, campaign and delivery records, bounces and complaints. Newsletter records remain while you are subscribed. After unsubscribe or erasure, I may retain a limited suppression and consent record where needed to honour the opt-out or establish legal compliance.
Email delivery is a separate step: Sendy sends the recipient address, message content and headers, and receives delivery, bounce and complaint data through Amazon Simple Email Service in eu-central-1, Frankfurt. The active endpoint is email.eu-central-1.amazonaws.com. Provider: Amazon Web Services EMEA SARL, Luxembourg.
Legal basis: consent (Art. 6(1)(a) GDPR) for newsletters; contract performance (Art. 6(1)(b) GDPR) or legitimate interest (Art. 6(1)(f) GDPR) for necessary service and transactional messages.
AWS Privacy Notice
Emails to existing customers
As a customer you may receive emails about your plan and about similar offers of mine without a newsletter subscription (Section 7(3) of the German Unfair Competition Act). You can object to such emails at any time, free of charge, via the unsubscribe link in each email or by writing to [email protected].
Legal basis: legitimate interest (Art. 6(1)(f) GDPR).
Newsletter signup inside the learning app
Signed-in learners can subscribe to or unsubscribe from my newsletter at Settings → Email in my learning app. For learners in the EU/EEA, UK, and Switzerland the signup first requires ticking a separate explicit-consent checkbox; this is a single opt-in (no confirmation email). I record an audit entry for each subscribe or unsubscribe (email, list, action, the exact consent text shown, IP address, user agent, country, timestamp), kept for 5 years to demonstrate consent under Art. 7(1) GDPR. You can unsubscribe at any time from the same screen or via the link in any email.
Legal basis: Your consent (Art. 6(1)(a) GDPR).
5d. Purchase Verification (Article Buster)
The Article Buster entitlement ledger stores a SHA-256 hash of the purchase email so the app can verify access through Restore Purchase. The ledger is self-hosted in Germany and is retained while Article Buster is supported and the entitlement is needed. A separate diagnostic log stores timestamps, verification results and limited webhook details such as event type and product or course identifiers for troubleshooting and fraud prevention. It does not store the purchase email or the full webhook payload. Diagnostic logs are kept for no more than 12 months.
Legal basis: Contract performance (Art. 6(1)(b) GDPR) for the entitlement and legitimate interest (Art. 6(1)(f) GDPR) for diagnostics and fraud prevention. The entitlement ledger is not retained under a claimed ten-year AO or HGB tax rule.
5e. Writing Tool (write.smartergerman.com)
Usage, quality and error logs
The Writing Tool stores the raw IP address, a direct Teachable-prefixed user identifier, action, token use, success and timestamp for abuse prevention and cost control. It also stores correction, evaluation, explanation and error-analysis records that can contain your original and corrected text, surrounding context, scores, highlighted words and technical error details, generally linked by a session hash. The user identifier is not described as an anonymous or pseudonymized hash. These records are self-hosted in Germany and are kept for no more than 12 months.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR) in service security, cost control and quality improvement.
Google Gemini via Vertex AI
The text you submit is sent to Google Gemini on Vertex AI in an EU region to generate corrections, explanations and evaluations. Google does not use the content to train its models without permission. Provider security or abuse processing can apply under Google’s service terms.
Provider: Google Cloud EMEA Limited, Ireland. Google uses Google LLC, USA, as a subprocessor for data-centre operations, maintenance and technical support; in doing so, Google LLC may access this data from the USA, also when EU endpoints are used. Such transfers rely on the European Commission’s adequacy decision for the EU-U.S. Data Privacy Framework, under which Google LLC is certified; where that decision does not cover a transfer, Google’s Cloud Data Processing Addendum provides for EU Standard Contractual Clauses. You can request a copy of these safeguards at [email protected].
Legal basis: Contract performance (Art. 6(1)(b) GDPR).
5f. Pronunciation Scoring (learning app)
Reading Out Loud and Grade
Preview and local speech processing run on my German server. When you request Grade, the audio from that recording and its reference sentence are sent to Microsoft Azure Speech in the Germany West Central region to score pronunciation. I do not store the audio. Microsoft states that real-time speech and pronunciation-assessment audio is not retained or stored.
Provider: Microsoft Ireland Operations Ltd.
Legal basis: Contract performance (Art. 6(1)(b) GDPR) because the transfer provides the scoring you request. A privacy notice is shown before first use.
5g. AI Practice Chatbot
AI chatbot for German practice
The Practice Bot at bot.smartergerman.com lets you practise German conversations and role-play with AI characters. It is an AI system, not a person - even when it plays a named character.
Sign-in: you use the chatbot with your course account. Your account - email and name - identifies your chats, usage and saved content.
Data processed: the messages you type or speak, and any personal data you choose to include in them.
Providers: OpenAI - current GPT models - generates the answers and transcribes voice messages you record. OpenAI may retain API data for up to 30 days for abuse monitoring; limited longer retention is possible when required by law or needed to investigate abuse or a security incident. I do not opt in to model training. Google Gemini normally generates the read-aloud voice for bot messages; OpenAI remains the fallback voice provider. Microsoft Translator translates messages when you use the translate button.
Stored content: conversations are saved only if you switch saving on - you see your saved chats, can delete them anytime, and they expire after 90 days unless you mark up to ten to keep. Individual exchanges where the assistant answers wrongly may be stored for error analysis; these are linked to your account, so a deletion request can find and remove them, and they are deleted after 12 months at the latest.
Please do not share passwords, payment details, health information or ID numbers in the chat.
Legal basis: Contract performance (Art. 6(1)(b) GDPR) for answering and the voice features; your consent (Art. 6(1)(a) GDPR) for saved conversations; legitimate interest (Art. 6(1)(f) GDPR) for error analysis and abuse prevention.
Feedback and bug reports
You can send me feedback or a bug report from Settings. This is optional, and the report is linked to your course account. I store your message, selected category, attached and automatically generated screenshots, and technical details such as app version and browser. When you open the report form, the switch Include data about my current conversation starts on. If you send the report, it includes the conversation in that tab, the latest sentence check and related technical details unless you switch it off first. Nothing is sent until you press Send. Switch it off if the tab contains sensitive information that is not needed for the report. I use reports to fix problems and improve the chatbot. When I use Claude from Anthropic or Codex from OpenAI to analyse a report, its content can pass through that provider’s API. Anthropic and OpenAI state that business API inputs and outputs are not used for model training by default and are normally deleted from their systems within 30 days, subject to limited security, abuse, legal or contractual exceptions. I delete every report, including conversation data and screenshots, no later than 12 months after receipt, whether handled or not.
Legal basis: Legitimate interest (Art. 6(1)(f) GDPR).
Anthropic API retention | OpenAI API data controls
5h. Learning App (learn.smartergerman.com)
One privacy policy
This page also covers my learning app at learn.smartergerman.com. The privacy link inside the app leads here.
Third-party lesson media
Lessons can contain videos, documents, pages or exercises hosted by YouTube, Google Docs, GitHub Pages, LearnClick and TED. The learning app does not contact these providers before you choose Load. If you load an item, the provider receives your IP address and browser data and may store or read information on your device under its own privacy terms. You can remember one provider on this device, stored locally in your browser, or enable all third-party lesson media in Profile under Privacy & data. Switching the profile setting off also clears remembered device choices. YouTube players use the privacy-enhanced youtube-nocookie.com host, but Google still receives data when you load them.
Providers: Google for YouTube and Google Docs, GitHub for GitHub Pages, LearnClick and TED.
Legal basis: Your consent under Art. 6 para. 1 lit. a GDPR and Sec. 25 para. 1 TDDDG for storage or access on your device.
Processing on my own server
Lessons, quizzes, progress, speech transcription for exercises such as the Listening Lab, read-aloud previews and grammar checking run on my own server in Germany with self-hosted tools (Whisper, Vosk, LanguageTool). Nothing of this reaches any third party.
Legal basis: Contract performance (Art. 6(1)(b) GDPR)
Testing a second speech recognition model
During a four-week test in October and November 2026, a second self-hosted speech recognition model, Parakeet, also transcribes the recordings from the speaking exercises of the Learning App, on my own server and next to Whisper. Your result comes from Whisper. Only in the Assisted Preaching drill, when Whisper’s second check would accept an answer it first rejected, Parakeet’s transcript of the same recording is compared: if both models heard the same wrong grammatical form, the answer stays incorrect. This comparison runs on my own server, stores nothing of its own and continues after the test ends. I store both transcripts without your name, email address or account. I delete the transcripts after four weeks as well. Where the two transcripts differ, and for a small random sample of the others, I also keep the recording so I can listen to it and judge which model was right. I also keep recordings that only passed thanks to Whisper’s second check, so I can verify that this check is right. I delete each recording once I have listened to it, and at the latest four weeks after it was made. These recordings are never included in backups, only I can access them, and nothing of this reaches any third party. You can object at any time by writing to me; I will then leave your recordings out of the test.
Legal basis: Legitimate interest in accurate transcription of your exercises (Art. 6(1)(f) GDPR)
AI features using Google Gemini
The writing tool, creative writing suggestions, exam evaluations and the Talkie Talkie speaking coach send the text you write, or for the speaking coach your voice and its transcript, and the German Learning DNA report sends your questionnaire answers, to Google Gemini via Vertex AI in the EU region to generate feedback. The spoken-translation checker sends your transcribed sentence only after you switch that checking on. Google does not use this data to train its models. Corrections from your practice are kept in your learning history on my server, so the app can target the mistakes that keep coming back; they are part of your account data and are removed with it.
Provider: Google Cloud EMEA Limited, Ireland. Google uses Google LLC, USA, as a subprocessor for data-centre operations, maintenance and technical support; in doing so, Google LLC may access this data from the USA, also when EU endpoints are used. Such transfers rely on the European Commission’s adequacy decision for the EU-U.S. Data Privacy Framework, under which Google LLC is certified; where that decision does not cover a transfer, Google’s Cloud Data Processing Addendum provides for EU Standard Contractual Clauses. You can request a copy of these safeguards at [email protected].
Legal basis: Contract performance (Art. 6(1)(b) GDPR).
Pronunciation scoring
See section 5f above. When you request Grade, one recording and its reference sentence are sent to Microsoft Azure in Germany. A privacy notice is shown before first use.
Vocabulary translation and pronunciation
When you tap a word, the translation comes from my own server; no word is sent to a translation service. Pronunciations are audio files on my server, and your browser loads them only from there. In the app they are prepared in advance; if a word has no audio yet, my server notes only the word, without your account, and I later create its audio with Google Cloud Text-to-Speech. On the course pages, the first time a word needs a pronunciation, my server fetches the audio for that single word once from the speech function of Google Translate and stores it. The word is normally from my lesson text; if you saved a word from your own letter in the writing tool, it is that one word. Nothing else about you reaches Google - not your name, email address, IP address, account or anything else you wrote or said.
Providers: Google Cloud EMEA Limited for Text-to-Speech and Google Ireland Limited for Google Translate, both Dublin, Ireland.
Study assistant
The study assistant answers your questions about German from my own course material. It is open only to learners whose plan includes it. To write an answer, your question, your last three questions and answers, your course level and excerpts of my lessons are sent to Google Gemini via Vertex AI. This processing takes place in the EU. Google does not use this data to train its models. I do not store the text of your question. For each question I keep a short neutral topic label, your course level, whether my material could answer it, and your rating if you give one; none of this is stored with your name, and I delete it after 12 months. With your account I keep how much the assistant has cost in each month, to apply the monthly limit; I delete these figures after 12 months. If you press Tell Michael, that one question and its answer are stored with your account so that I can reply and improve my material; I delete these reports after 12 months. If you switch on Saved for practice, I store the rule and topic of every practice task you mark Not yet with your account, so that you can practise those rules again. I do not store what you type. A saved rule expires 90 days after you last missed it, and switching the function off deletes everything saved at once. The answers are written by an AI system, not by a person, and can contain mistakes. Please do not enter personal details about yourself or other people. The study assistant is intended for adults.
Provider: Google Cloud EMEA Limited, Ireland. Google uses Google LLC, USA, as a subprocessor for data-centre operations, maintenance and technical support; in doing so, Google LLC may access this data from the USA, also when EU endpoints are used. Such transfers rely on the European Commission’s adequacy decision for the EU-U.S. Data Privacy Framework, under which Google LLC is certified; where that decision does not cover a transfer, Google’s Cloud Data Processing Addendum provides for EU Standard Contractual Clauses. You can request a copy of these safeguards at [email protected].
Legal basis: Contract performance (Art. 6(1)(b) GDPR); for Saved for practice your consent (Art. 6(1)(a) GDPR); for the anonymous topic records my legitimate interest in improving my material (Art. 6(1)(f) GDPR).
Study buddy
If you want practice partners suggested, you fill in a matching profile: level, practice times and goals, and optionally time zone, country, languages, gender, age range and religion, plus whether your partner should be of a certain gender or share your religion. I use the profile only to suggest partners; if you took the German Learning DNA quiz, your learner type also counts. Other learners see your name, level and country, which practice times, goals and languages you share and whether your learner types complement each other, never your gender, age range or religion. Because matching follows the preferences, learners who asked for a certain gender or the same religion are only shown partners who fit. The profile is stored on my own server in Germany and deleted when you turn matching off or delete your account.
Legal basis: Contract performance (Art. 6(1)(b) GDPR); for religion your explicit consent by choosing to answer (Art. 9(2)(a) GDPR), which you can withdraw at any time by changing the answer or turning matching off.
Your data, your controls
In the app you can export your data as a PDF and request full deletion; deletion removes your account and learning data. Where a consent or transaction record must be retained to demonstrate compliance or defend a claim, it is kept separately only for the period stated in this policy.
Search in the learning app
What you type into the search box is sent to my server only to find lessons, glossary entries, course questions and posts, and is not stored. Server logs record that a search happened, never its text. I count which help entries are opened and how many searches find nothing, per day and without any link to you. If you choose Suggest a topic, I store the text you send, without your name, account or IP address, for up to 12 months to add missing topics. Please do not include personal details.
5i. Live class (Zoom, YouTube)
Registration and the class
You register for the free weekly live class with a first name and your email address on smartergerman.com. My server passes both to Zoom, which sends you the joining details and runs the class. Zoom receives the usual meeting data: the name shown, audio, video if you switch your camera on, chat messages and connection data.
Provider: Zoom Video Communications, Inc., USA. Zoom is a registered participant of the EU-US Data Privacy Framework and provides Standard Contractual Clauses for other transfers.
Legal basis: contract performance (Art. 6(1)(b) GDPR) for registration and class.
Recording and publication
The class is streamed to and published on my YouTube channel. The registration page says so before you sign up, lets you choose the name that is shown, tells you that this name is public, and asks you to confirm that you understand the class is recorded. If you do not want to appear, keep your camera and microphone off. The recording stays online for as long as I keep it published.
Provider of the platform: Google Ireland Limited (YouTube), Dublin, Ireland; Google LLC, USA, is certified under the EU-US Data Privacy Framework.
Legal basis: your confirmation at registration (consent, Art. 6(1)(a) GDPR) for appearing in the published class.
6. Your Rights
Under GDPR you have the right to: access your personal data, rectify inaccurate data, erase your data ("right to be forgotten"), restrict processing, data portability, object to processing, and withdraw consent at any time. You may also lodge a complaint with a supervisory authority.
Deletion and backups
When I delete your data, it is removed from live systems within the periods described in this policy. Encrypted backups are kept for up to 12 months and used only for security and disaster recovery, not ordinary processing. A disaster recovery can temporarily restore data that was deleted after the backup was created; any continuing deletion obligation then remains and the affected live record must be removed again. The backup itself expires within 12 months.
Contact: [email protected]
7. Data Security
This website uses SSL/TLS encryption. You can recognize an encrypted connection by "https://" in the address bar and the lock icon.