This English version is the authoritative one. 中文版
Terms · Privacy · Refund
Last updated: 7 November 2026 · Operated by: PTE Study Pty Ltd (Australia)
- Who we are
- Terms of service
- Green light (and the former guarantee)
- Refund policy
- Privacy policy
- Classes and teachers
- Contact us
The most important thing to say first: NAATI CCL is an independent exam-preparation platform.
We are not affiliated with, endorsed by, or in partnership with NAATI (the National Accreditation Authority for Translators and Interpreters).
We do not represent NAATI, and we cannot register you for the exam, change your score, or influence your actual exam result.
"NAATI CCL" appears on this site only to identify which exam we are helping you prepare for.
1. Who we are
| Service name | NAATI CCL (naaticcl.com) |
|---|---|
| Operating entity | PTE Study Pty Ltd (an Australian-registered company) |
| What we provide | Practice dialogues, audio, AI-based marking, and simulated exams for the NAATI CCL interpreting test |
| Billing currency | Australian dollars (AUD), processed by Stripe — we never see your card number |
2. Terms of service
2.1 What we do promise
- Practice dialogues generated in line with NAATI's published test specifications (12–14 segments, approximately 300 words, no segment longer than 35 words, covering the official twelve topic areas)
- Marking and predicted scores based on NAATI's published assessment framework, with our prediction error rate published openly
- Your marking records belong to you, and you can request deletion of them at any time (recordings themselves are not kept — see section 5)
2.2 What we do not promise
- We do not promise you will pass the real exam. Predicted scores are a statistical estimate, not a guarantee
- Practice dialogues are originally generated to the official specifications — they are not copies of actual exam questions
- AI marking can get things wrong. If you believe anything has been marked incorrectly, you can flag it with one click and request remarking
2.3 Account and usage rules
- Your account is for your personal use only. Sharing your account with others may result in your access being suspended
- Paying without signing in: you can buy a plan without signing in first. Your plan goes on the email you enter on the payment page, so use the email you sign in with. If that email already has an account, the plan is added to it (remaining time is kept and the new time starts after it) and we send a sign-in link to that email. If it is a new email, this browser is signed in once, straight after payment; signing in later with that email ends that automatic sign-in on other devices. If you paid with a different email from the one you sign in with, sign in with the email on your receipt
- Each paid plan covers one CCL language. If you buy while practising a language, the plan covers that language; otherwise it covers the first language you practise after paying. Marking for other languages is not included. Plans bought before 7 October 2026 are not limited to one language
- Free: your first two full dialogues are marked free, which is one practice test in the same structure as the real test (you can sit them back to back as a mock exam once you are signed in). Without an account you can practise one dialogue a day; its score opens after a free email sign-in and counts as one of the two
- The 7-day pass (A$19) covers up to 10 full dialogues marked in total within 7 days of purchase. If you buy the 90-day plan within 7 days of buying the 7-day pass, the A$19 you paid is taken off its price (you pay A$50). A 7-day pass bought before 30 October 2026 at A$29 has its A$29 taken off instead (you pay A$40). If the 7-day pass was refunded, this does not apply
- One full mock exam (A$9) covers one mock exam sitting (two dialogues back to back), to be taken within 7 days of purchase. A mock exam bought before 19 October 2026 (A$49) keeps its two sittings. If you buy the 90-day plan within 7 days of buying the A$9 mock exam, the A$9 you paid is taken off its price (you pay A$60). If the mock exam was refunded, this does not apply
- Second check: on a segment that has been marked, a signed-in student can ask a stronger AI model to mark it again with the same rules, up to 5 times a day. If the two markings differ by more than 10 points, the second one is used for that segment. It is still an AI estimate, not a human review and not a NAATI result
- Marking is subject to a daily usage cap to prevent automated abuse
- You must not scrape, bulk-download, or resell content from the platform
3. Green light and the former Green Light Guarantee
The Green Light Guarantee is no longer offered. It was sold as a separate $199 plan and was withdrawn in September 2026; no account holds it, so sections 3.1 onward do not apply to any current plan. The green light itself remains as a readiness indicator in "My profile".
What "green light" means: the green light is our own readiness
standard, computed from segment-by-segment marking of your real recordings.
Six conditions must all hold. Across your last 20 marked segments: an
average information-retention of 85% or more and no more than 2 severe losses.
Across your practice in the last 30 days: at least 6 different dialogues,
at least 2 different domains, spread over at least 3 days.
You can see exactly where you stand on each condition in "My profile".
The green light is not a prediction of your NAATI score. It is a practice
benchmark we set ourselves, and it says nothing about how you will do on the day.
3.1 What the guarantee covers
If you reach green light and then do not pass the real exam within the following 30 days, we will extend your full-feature access free of charge until you pass.
Why extended access rather than a cash refund: what you want is to pass, not to get the money back. Extending your access means we finish the job with you — you keep practising, at no further cost, until you pass.
3.2 Conditions for eligibility (all must be met)
- Your purchase must be made before your declared exam date
- You must have completed marking on at least 30 live-voice recorded segments (typed-text submissions do not count)
- You must have completed at least 3 exam-condition mock tests (no reference translation shown, timed, listened through in one go)
- Your green light must have come from an exam-condition mock test — results from practice mode do not count
- You must submit your official result notice (including your candidate number and exam date) within 30 days of the exam
3.3 How extensions work
Access is extended by exam cycle: each extension requires proof of your booking for the next exam sitting (including your name and NAATI candidate number). If you fail to submit your result after an exam, the guarantee is paused, and resumes once you do.
4. Refund policy
| 7-day no-questions-asked | Full refund within 7 days of purchase, provided no more than 10 segments have been marked |
|---|---|
| Service unavailable | If a technical fault on our end makes the service unavailable for 72 consecutive hours, a refund will be issued in proportion to the unused days |
| Failed marking | Segments where marking failed to produce a result are not counted towards usage and are not charged |
| Not eligible for refund | Requests for a refund on the grounds of "I didn't pass" after substantial use of the service (this situation is covered by the Green Light Guarantee — see Section 3) |
Refunds are returned to your original payment method, typically within 5–10 business days. See Section 6 for how to apply.
Nothing in this refund policy affects your rights under the Australian Consumer Law.
Where there has been a major failure in the service, your statutory remedies take precedence over this policy.
5. Privacy policy
This policy covers naaticcl.com, our mobile app, WhatsApp practice and our ChatGPT app. If you only use the ChatGPT app, section 5.6 is the full notice for it.
5.1 What we collect
| Test-date alerts | We record your selected languages, signup source, confirmation and unsubscribe times, using a random signup ID to count confirmations and unsubscribes. Links in alerts carry source tags and a random delivery ID to count return visits and practice starts, not for advertising |
|---|---|
| Email address | Used for login (we don't use passwords — login is via an emailed link), to send receipts, for any test-date alerts you sign up for, and for occasional emails about your practice or an unfinished checkout (one tap to stop) |
| Your recordings | Used for transcription and marking, which is the core of this service |
| Practice records | Scores, weaknesses and progress, used to generate your diagnostic report |
| Payment information | Processed and stored by Stripe, which also keeps your customer and subscription records. We send Stripe your email address and plan details (or, if you pay without signing in, you enter your email on Stripe's page and Stripe sends it to us so we can create or find your account), and keep your plan language and daily practice count in your Stripe customer record to manage access. We receive your customer ID, subscription status, payment confirmation and the last four digits of your card |
| Exam plans and results | If you tell us, your planned exam date or that you have not booked, stored on your account to plan your practice. If you answer our post-exam question, your answer, your language and whether you have paid are stored separately from your account, with a one-way hash of your network address to stop repeat submissions |
| Test results and comments you choose to share | If you tell us whether you passed the CCL test, and optionally your test language, your total score and one sentence about what helped, we store this under your coded account ID and use it in our own statistics. We ask once, from 14 days after your first practice (and again a month later if you say you have not sat the test yet); if you close the question we do not ask again. A sentence is made public only if you tick a separate box, unticked by default, giving permission. We read it first. If we publish it, it appears exactly as you wrote it, signed with your first initial and the test language only, and never with your email, name or score. You can withdraw permission or delete your answer at any time from your account page: the sentence is deleted from our records straight away and any copy on our public pages is taken down within 14 days. Deleting your account deletes it too |
| WhatsApp number (only if you practise on WhatsApp) | Messages reach us through Twilio. We keep only a one-way coded form of your number, your practice progress and your marking results — not the number itself. Voice notes are deleted from Twilio as soon as they have been marked, and we never store their transcripts. We only reply to messages you send us; reply STOP to stop all messages. If you link the number to your account, practice there counts towards the same plan and the same one free dialogue |
| Sign-in with Google or Apple | If you choose it, we receive your email address and an account identifier from that provider |
| Site usage | Pages you open, actions such as starting a practice, your browser type, whether you are on a phone, your interface language, your country and network provider (worked out from your network address), and where you came from (a website or a link tag such as utm_source). We connect these using a random device ID kept in a cookie (section 5.5), and when you sign in they can be linked to a coded form of your account. We use them to count visitors, find faults and improve the site, not for advertising |
| Invitations | If you share or use an invitation link, we connect your account, your email address, a coded device ID and who invited you, so we can credit the invitation and stop the same device claiming it twice |
| Network address | Used when you connect, to apply free-practice limits and stop abuse. Limits are counted under a one-way hash of the address. When you fetch practice audio, or when a connection looks like automated scraping, we log the address, country, browser identifier and the file or page requested |
| Messages you send us | Feedback and emails to our support addresses, so we can reply |
5.2 How your recordings are used
- Sent to speech transcription and marking services (Anthropic, ElevenLabs) to generate your results
- They are never used to train our own models, and are never made public or sold to third parties
- Recordings are not stored. Each one is sent for transcription and marking and then discarded; we keep only the marking result (scores and the kinds of errors found), which is deleted together with your account
- When a segment can't be marked — we couldn't make out the recording, or it seemed to be in the wrong language — we also keep the first 300 characters of what the transcription service heard for that segment, so we can tell whether the problem was on our side and fix it. Segments that were marked normally keep no transcript. This is deleted together with your account
- If you are signed in and stop part-way through a dialogue, we keep the marking of the segments you have done — including what the transcription service heard — on your account, so you can carry on from the same segment on another device. The full text of what was heard is removed when you finish that dialogue. The point-by-point marking, which can quote short phrases of what you said, stays until it is replaced when you start another dialogue, and everything is deleted together with your account
- If you type instead of speaking, your typed answer is sent to Anthropic for marking and handled the same way as a transcript
- Dzongkha volunteer reading is the only place we store recordings. If you tick the consent box on /naati-ccl/dzongkha/read and read the sentences we wrote, the recordings are stored privately under your account (Cloudflare R2) and a person on our team listens to each one; approved recordings become the Dzongkha voice in practice dialogues on this website and in the app. We don’t show your name unless you ask to be credited, never sell them, and never use them to train models. Delete your account before they go live and they are all deleted; to withdraw after they are live, email us and we take them out of the dialogues and delete them within 30 days
5.2.1 CPI role-play practice
CPI practice uses the same sign-in with a separate practice record and purchase allowance. The CCL recording and transcript descriptions above apply to CCL. For CPI, your browser keeps your recordings, notes and unfinished role-play on this device so you can sign in and continue in the same browser. Recordings are sent to the transcription service after you finish a role-play. Before sign-in, transcripts are saved privately against a device recovery token so you can continue after sign-in. On successful account binding, these transcripts move to your account and the anonymous transcript copies are removed. Unclaimed transcripts remain associated with that recovery token until claimed or removed following a request to support@naaticcl.com. Clearing browser data removes your local copy and recovery token, and does not itself remove the server copy. We retain the CPI transcript, feedback, band estimates, repeat requests and progress in private Cloudflare R2 storage under your account so you can retry interrupted marking and read your reports again. These server records are retained until you delete your account through your account page. Browser copies of recordings and notes remain on this device until you clear this site's browser data. A sign-in on another device can retrieve saved server reports, while recordings awaiting marking remain on the original device. Purchase records and encoded counters needed to prevent repeated free claims are retained separately. CPI's 90-day allowance resets each day at 00:00 UTC.
5.3 Where your data is stored and who receives it
Practice records (marking results — not audio) are stored in Cloudflare R2 (Sydney region). We share data only with the services that run this site, and only for the purpose shown:
| Vercel | Hosts the website and our server. Every request passes through it, including your network address and browser type |
|---|---|
| Cloudflare | Stores our records and the practice audio |
| Anthropic | Marks your interpretation (typed text or a transcript, with the practice segment) |
| ElevenLabs | Transcribes your recording |
| Stripe | Takes payments |
| Resend | Sends and receives our email |
| Sign-in with Google, if you choose it, and our support mailbox | |
| Apple | Sign in with Apple, if you choose it |
| Twilio | WhatsApp practice, if you use it |
| YouTube | Only when you press play on a video on our pages |
| OpenAI | Tool results, if you use our ChatGPT app (section 5.6) |
| Your teacher | If you join a class (section 6) |
Some of these services process data outside Australia, including in the United States. We do not sell your data.
5.4 Your rights
- To access, export, correct, or delete your data
- To withdraw consent and close your account (closing your account deletes all your practice records; see how to delete your account)
- To complain about how we handle your personal information; if your complaint is not resolved to your satisfaction, you may complain to the Office of the Australian Information Commissioner (OAIC)
5.5 Cookies
We use our own small cookies and no advertising trackers. They keep you logged in, remember your interface language, avoid counting you twice in a day, hold a random device ID (ccl_device) for the usage data in section 5.1, remember where you came from, remember an invitation or homework link you opened, remember prompts you have already seen, help tell real visits from automated downloading, and, if you pay without signing in, link this browser to that checkout for up to 2 days so it can be signed in after payment (__Host-ccl_gp). You can block or delete cookies in your browser, but sign-in and some features may then stop working.
If you join a class, two more cookies are stored; see section 6. Google's sign-in button and YouTube videos are provided by those companies, and they may set their own cookies.
5.6 If you use our ChatGPT app
This is the full notice for our NAATI CCL Practice app in ChatGPT and its server at https://naaticcl.com/mcp.
The app works without an account on this site. It does not ask for your name, email address, password, payment details or voice recording, and it cannot read the rest of your ChatGPT conversation. It receives only the text ChatGPT includes when it uses the app. If that text contains personal information, we process it in the same way; we do not remove it automatically.
What each tool receives and returns
| List languageslist_languages | Receives: nothing. Returns: the languages we cover, how many practice dialogues and model answers each has, and links to our site. |
|---|---|
| Find dialoguesfind_dialogues | Receives: a language; optionally a topic or word to search for, and a page number. Returns: matching practice dialogues with title, topic, number of segments and links to our site. |
| Play a segmentget_segment | Receives: a dialogue ID and segment number; optionally a request to show the text. Returns: the situation, speaker, languages, audio link and text of that practice segment, and links to our site. |
| Mark an interpretationmark_interpretation | Receives: a dialogue ID, a segment number and the interpretation you typed or dictated. Returns: your interpretation repeated back, the share of information kept, point-by-point feedback, notes on language and delivery, the model answer and examiner points where ready, and how many free markings you have left today. |
| Check test datestest_dates | Receives: a language. Returns: upcoming NAATI sessions and seat numbers from our copy of NAATI's public search, when it was checked, and a link to our site. |
Tool results contain practice content and the text you sent. We do not add your name, contact details or account details, or anything about other users. Results go back to ChatGPT, where OpenAI handles them as part of your conversation.
What we handle, why, who receives it and how long we keep it
| Your interpretation | Why: to mark it and show you the feedback. Who receives it: our server on Vercel and Anthropic (our marking model), with the practice segment and no user ID. It also goes back to ChatGPT in the result. How long: we do not store it or write it to our logs; it is discarded once the result is returned. Anthropic deletes API inputs and outputs within 30 days, unless it must keep them longer to enforce its usage policies or comply with the law. |
|---|---|
| One-way hash of the dialogue, segment and your text | Why: so a network retry of the same answer does not use another free marking. Who receives it: our storage on Cloudflare R2. How long: deleted by a daily clean-up once it is 30 days old. It is a salted hash: it does not contain your text and we cannot read your text from it. |
| Anonymous user and conversation IDs that ChatGPT attaches | Why: to apply 3 free markings per person per day and to slow down rapid repeated requests. Who receives it: our server on Vercel. Only a salted one-way hash is stored, on Cloudflare R2. How long: the IDs themselves are held only briefly in server memory and never written to storage. The hash and the daily count are deleted by a daily clean-up once they are 30 days old. We use the conversation ID or network address only when no user ID is sent. |
| Network address of the request (normally an OpenAI server, not your device) | Why: to stop abuse and limit how much marking one network can use. Who receives it: Vercel, which hosts our server and firewall. A salted one-way hash is stored on Cloudflare R2. How long: the hash is deleted by a daily clean-up once it is 30 days old. Vercel's logs are kept for up to 30 days. An address that breaks our anti-abuse rules can be blocked as described in section 5.7. |
| Usage log | Why: to find faults and count use. Contents: the request time, path and status from Vercel, and one line from our server with the tool name, language, whether it worked and the marking status. It contains no text and no user IDs. Who receives it: Vercel. How long: up to 30 days. We keep daily totals, which do not identify anyone. |
| Other details ChatGPT may attach, such as locale, device type or approximate location | Why: we do not use them. Who receives it: they reach our server on Vercel with the request. How long: we do not store them or write them to our logs. |
| Audio and links | When you press play, your device fetches the practice audio from naaticcl.com and Cloudflare R2, so your network address and browser type reach Vercel and Cloudflare, as with any web request. To protect the audio from automated downloading, we log the address, country, browser identifier and audio file requested, in Vercel's logs and our own security records, for 12 months (section 5.7). These records are not linked to your ChatGPT IDs. Links to our site carry the tag utm_source=chatgpt_app. If you open one, sections 5.1 and 5.5 apply to that visit. |
We use this data only for the purposes above. We do not sell it, use it for advertising, build profiles from it or use it to train models.
Your controls
- Choose what you send. Leave names, contact details and other personal information out of your answers. Practice answers never need them.
- Stop at any time. You can disconnect or remove the app in ChatGPT's settings. After that it receives nothing from you.
- Questions and requests. For a question, or to ask for access, correction or deletion, email hello@naaticcl.com. We reply within 2 business days. Because we keep only hashes, we usually cannot link a daily count to you, and it is deleted automatically once it is 30 days old. You can also complain to the OAIC (section 5.4).
- Your ChatGPT conversations, including tool results, are managed in ChatGPT under OpenAI's privacy policy and your ChatGPT data settings. We cannot delete them for you.
5.7 How long we keep data
| Account and practice records | Until you delete your account |
|---|---|
| Records kept after you close your account | A coded record of free allowances, refunds and rewards already used, so closing an account does not reset them, and a coded record that stops old sign-ins working. We keep these for as long as we offer free allowances |
| Recordings | Not stored, apart from Dzongkha volunteer reading (section 5.2) |
| Payment records | At least 5 years, as Australian tax law requires. They are not deleted when you close your account |
| Site usage data and post-exam answers | 24 months, then deleted automatically |
| Test results and comments you shared | Until you delete them or your account. A comment published with your permission comes down from our public pages within 14 days of withdrawal or deletion |
| Network-address counters, audio logs and anti-scraping records | 12 months, then deleted automatically. An address or network caught scraping can also stay blocked for as long as the abuse is a risk |
| Emails and feedback you send us | As long as we need them to help you and keep business records. Tell us if you want them deleted |
| Records of emails we sent you, and unsubscribes | For as long as we send email, including after you close your account, so we do not email you again after you opt out |
| ChatGPT app | See section 5.6 |
| Service providers' own logs | Kept under each provider's own policy |
5.8 Young people
This service is for people preparing for the NAATI CCL test and is not directed at children under 13. If you think a child under 13 has given us personal information, email us and we will delete it.
5.9 Changes to this policy
When the way we handle data changes, we update this policy and the date at the top of this page.
6. Classes and teachers
A CCL teacher can set up a class on this site and invite students with a link, a class code or a QR code. Teachers are independent: they are not employed by us, and we are not responsible for their teaching. Using a class is free for the teacher. Students pay for their own plan at the normal price, if they choose to buy one.
6.1 If you join a class as a student
- Always, without names: your teacher sees class totals: how many students joined, how many practised in the last 7 days, how many dialogues the class has practised, and the kind of error the class makes most. The error breakdown and the class average are only shown once three students have practised, unless every student who practised has chosen to share their results.
- Only if you tick "Let my teacher see my practice results": your teacher also sees a row for you: the name you entered (or the first two letters of your email address followed by ***), how many dialogues you practised in the class, your average score, the kind of error you make most and when you last practised. The box is not ticked unless you tick it.
- Never: your recordings or the words you said (see 5.2).
- In a class whose teacher we invited or verified, signed-in students get three full dialogues marked free instead of two. If you leave the class, the normal free allowance applies again; dialogues you have already used still count.
- You can withdraw that consent, or leave the class, at any time under My progress. Your row disappears from the teacher's page straight away.
- Joining stores two cookies in your browser for 90 days: the class code, so that practice on that device counts towards the class, and your teacher's referral code. If you practise without signing in, your practice only adds to the class totals.
6.2 If you set up a class as a teacher
- We store your email address, the name you want students to see, your classes and your referral code.
- When a student who joined through your link pays for a plan, you earn a referral share. The current share is 50% of what the student pays, on first purchases and renewals, net of refunds. On the 10th of each month we pay out the previous month, with no minimum amount. We may change the share for future purchases and will show the current terms on your class page.
- Purchases you make yourself through your own link do not earn a share, and we may withhold a share for payments that were not made by genuine students of yours.
- A class name or teacher name must not suggest that the class is run or endorsed by NAATI, and must not promise a result.
- If your class is verified (we invited you, or we checked that you teach), you can choose to list yourself on our page for your class's language. It shows the name, the one line and the one contact link you enter, and that you practise with your class on naaticcl.com; never your students or their results. It is off unless you turn it on, you can turn it off on your class page at any time, and we may hide a listing that breaks these terms.
- If you delete your account, your classes are closed and their join links stop working.
7. Contact us
For refund requests, data deletion, or account issues, please use the "Feedback" button on the site, or email support@naaticcl.com. We will respond within 2 business days.