Privacy

Chinese Boost is run by Kensio Software. This page says what happens to your personal data when you use the site.

When you have no account

Nothing is collected. The site loads no analytics, no advertising, no social widgets and no fonts or scripts from anywhere else. There is no tracking to opt out of because none was built.

When you create an account

An account stores your email address, a password, an identifier generated for you, and a record of each device you are signed in on.

The password is held by Amazon Cognito and is never visible to this site. The identifier is ours and is what your saved work is filed under.

Why: to give you an account and keep you signed in. The lawful basis is performance of a contract, which is the service you asked for. Keeping the sign-in pages resistant to abuse relies on legitimate interests.

Your email address is used to confirm the account and to reset a password. It is not used for marketing.

Lookup history

Lookup history is optional. You can enable, pause or clear it on your account’s lookup history page. It records visits to dictionary entries and example sentences while you are signed in, and the words you look up in the browser extension once you have connected it. Free history holds up to 100 distinct words and sentences for 30 days after each lookup. Pro history holds up to 5,000 for 365 days. Repeated lookups refresh that period.

We also keep a separate internal archive of pseudonymous lookup events to explore related-word features. These events identify the reader using an internal identifier, the word or sentence, and the time of the lookup. They exclude email addresses, passwords and authentication session identifiers. Archived events are retained indefinitely, including after active history expires, you clear it, or you delete your account. Pausing history stops future collection. Expired active history is hidden immediately. Background deletion and backup expiry happen later.

Grammar practice

When you practise grammar while signed in, we save your progress and a separate record to help us evaluate the exercises. That record identifies the questions and versions used, their formats and teaching targets, first-answer results, hints, answer exposure, retries, completion and when you practised. It does not keep the answers you typed. You can also leave an optional comment about the exercise. Please leave personal details out of comments.

The separate evaluation record and its comment expire 180 days after you open that practice session. They are excluded from the site’s permanent change log and snapshots. Expired records are hidden immediately, while background deletion and backup expiry happen later. Account deletion removes active evaluation records and comments. Your saved progress follows the retention described below.

Sentences you write

When you are signed in, you can write your own example sentences in Chinese on your Sentences page. You can add an English translation if you want. Only you can see your sentences. You can edit or delete them.

Before a sentence is saved, it is checked automatically for anyone’s personal details and for offensive content. This check runs on Amazon Web Services in London. A sentence that fails the check is not saved. Please do not include personal details.

We store your saved sentences with the rest of your saved work. We also store words from these sentences that are added to your word list.

We keep recovery copies. A permanent change log records every version of a sentence, as first saved and after each edit or deletion. Every version stays in the log indefinitely and cannot be deleted. A separate snapshot store keeps the latest version indefinitely and earlier versions for 90 days. Database backups are kept for up to a year. If you start your own sentence from a sentence kept with a saved word, its page address and video time are also kept in these recovery copies as part of your sentence’s versions.

You can ask for an English translation of a sentence. The Chinese is sent to a translation model on Amazon Web Services in London, and the English it writes is saved with your sentence and marked as a machine translation. Amazon does not store what is sent or use it to train models. We count how many translations you ask for each day, since each account has a daily allowance.

You can also ask for a recording of a sentence read aloud. If the sentence is one this site already publishes, you hear the site’s own recording of it. Otherwise the Chinese is sent to BytePlus, a speech service run by ByteDance, which reads it aloud on its servers in Singapore, outside the UK. We store the recording privately with your sentence, where only you can play it. It is removed when you delete the sentence, or when you change the Chinese and a new recording cannot be made. We count how many recordings BytePlus makes for you each day, since each account has a daily allowance. Hearing the site’s own recording of a sentence it publishes is left out of the count.

Your sentences are removed from the active service straight away when you delete them or ask us to delete your account. The copies in the change log and snapshot store stay because those stores cannot be changed. We use these copies only to restore your own work. They are never shown to anyone. Database backups expire on the schedule above.

Why: to store what you asked us to keep. The lawful basis is performance of a contract. Keeping recovery copies relies on legitimate interests so we can put your work back after a mistake or a fault.

Pronunciation practice

With Pro, you can practise saying words aloud on the Pronunciation page. The page records each attempt through your microphone for up to ten seconds. It sends the recording to us to be checked. The site uses your microphone only on that page.

Each recording is sent to BytePlus, a speech service run by ByteDance. BytePlus writes down what it hears on its servers in Singapore, outside the UK. It receives only the sound. It is not sent your name, email address, account or the word you were trying to say.

The same recording is also checked by a pronunciation model that runs on Amazon Web Services in London. The model receives the word’s pinyin with the recording.

We do not keep your recordings. Once an attempt has been checked, the recording is gone. It is not saved to your account or written to our logs.

For each word, we store whether your latest attempt was heard as that word or a different one. We also store when you made it and how many of your attempts had each result. The page shows these details as your progress. They are kept with the rest of your saved work. They are left out of the change log and snapshot store described under “Sentences you write”. Database backups hold them for up to a year.

We count your attempts each day because each account has a daily allowance. Each day’s count is deleted automatically 400 days later. The length of each recording is written to the server logs so we can check what BytePlus charges.

Why: to give you the feedback you asked for. The lawful basis is performance of a contract.

Where you met a saved word

When you save a word, we can keep the sentence you met it in and the page’s address with it. For a video, we also keep the time in the video. The browser extension sends these details as described below. Saving a word from a graded reading passage on this site keeps the sentence and the passage’s address. Saving a word from the word segmenter keeps the sentence but no address.

We keep the address’s scheme, such as https, its host and its path. We remove the query and fragment, except the part that identifies a YouTube or Bilibili video. We do not keep addresses on your own machine, file pages or addresses containing a user name or password.

You can see these details on your word list and study cards. They are included in your word list export and emailed backup. We leave them out of the permanent change log and snapshot store. When you remove a word, we keep these details for the same period as the removed word. You can restore the word until that period ends. We then delete the details with the word. Database backups keep them for up to a year.

You can use a kept sentence to start writing your own example sentence. If you save your own sentence that way, it keeps the page address and video time. Those details become part of its versions in the permanent change log and snapshot store described under “Sentences you write”.

The browser extension

The Chinese Boost extension for Chrome and Firefox looks up Chinese words on other websites. It talks to chineseboost.com, and it loads no analytics. The one exception is the audio clip and still kept with a word saved from a video (see “Video subtitles” below), which it uploads straight to Chinese Boost’s private storage with Amazon Web Services.

When you look a word up. Hovering over a word asks chineseboost.com for that word’s dictionary entry. Opening the new words panel asks for each listed word’s dictionary entry. Each request names only the word. It leaves out the page you are on and the text around the word. These requests reach the site’s server logs, kept for one month, the same as any page of the site.

When you mark new words. The extension works out the words in your browser using its dictionary. With your account connected, it also checks your saved and known words. It tells new, learning and known words apart. It sends nothing from the page to Chinese Boost while marking. The extension downloads the dictionary and an index of HSK levels once.

The extension reads pages where you have turned on hovering or marking. You can turn either feature on for the current page until you leave it. You can also have them run on every page of a website on your list. Marking is off for each website until you turn on its New words switch. The website list and the extension’s other settings stay in your browser.

When you connect your account. The extension uses a session identifier stored in your browser. To mark new words on a page, it downloads your full list of saved words. The same request downloads the words you know from review or have marked as known by hand. It compares the page’s words with these lists in your browser. This check sends nothing from the page. When you hover over a word, it sends that word with your session identifier to check whether you have saved it. When you open the panel listing a page’s new words, it sends the listed words with your session identifier to check which you have saved.

Saving a word from the hover tip sends the word, the website’s name, the sentence around the word and the page’s address. The sentence is cut around the word to at most 80 characters. Saving one word from the new words panel sends the word, the website’s name and the page’s address, without a sentence. The website’s name, such as bbc.com, is kept as a tag on the saved word. The address is shortened as described under “Where you met a saved word”. Save all and Add all to a List send only the listed words and the website’s name as a tag. If you create a new List, Add all to a List also sends its name.

You can end the connection from your account page. Removing the extension deletes the session identifier from your browser.

When you mark a word as known. Pressing K over a word or I know this in the new words panel sends that word with your session identifier to mark it known. It sends nothing about the page.

Video subtitles. On YouTube, Bilibili, Netflix, iQIYI (iq.com and iqiyi.com) and Viki, the extension reads the Chinese subtitle track the video player itself downloads. It reads the track in your browser to draw each line with readings. Reading a track sends nothing to Chinese Boost. This feature is on by default on these websites. You can turn it on or off for each website.

Saving a word from a subtitle line sends the word, the subtitle line around it, the video’s address and the second in the video the line starts at. The subtitle line is cut to at most 80 characters. Pressing R opens the site’s example sentence form in a new tab with the line and the video’s address filled in.

With Pro, saving one word from a subtitle line also keeps the line’s audio and a still of the video frame, where the player allows it. The extension records each line in your browser as it plays and keeps the last few in memory. Nothing is sent unless you save a word from one of them. The clip and the still are uploaded to storage only your account can read, and are shown with the word on your saved words page and in review. Removing the word deletes them. They are never shared or published. Netflix and iq.com block this, and there only the line is kept. Add all to a List keeps no audio or stills.

Lookup history. With lookup history on, the extension adds a word when you rest the pointer on it for a moment. It sends the word and nothing about the page. Marking words and using the panel add nothing to lookup history. With history off, nothing is added.

Firefox lists what the extension sends before you install it, in three categories. Website content is the words you look up and the lines, clips and stills saved with them, browsing activity is the website a saved word was read on, and authentication information is the session identifier.

Where it is held

With Amazon Web Services, in their London region. AWS acts on our instructions and does nothing else with it.

Two things are sent to BytePlus, a speech service run by ByteDance, on its servers in Singapore. One is the Chinese text of a sentence of your own that you ask to have recorded and that this site does not already publish, as described under “Sentences you write”. The other is each recording you make in pronunciation practice, as described under “Pronunciation practice”. Nothing else is transferred outside the United Kingdom.

Nothing is sold. Nothing else is shared with anyone.

How long it is kept

A sign-in record expires thirty days after it was created and is deleted automatically. Server logs are kept for one month. Your account and the work saved against it are kept until you ask for them to be deleted, except for the recovery copies described under “Sentences you write”.

Cookies

Three, all of them necessary to sign you in and keep you signed in. One carries the identifier for your session, one holds a sign-in in progress for the few minutes it takes, and one tells the page whether to draw a sign-in or a sign-out link. None of them identifies you to anyone else and none is used for tracking, so there is no consent banner to click through.

Your rights

You can ask for a copy of your data, ask for it to be corrected, ask for it to be deleted, ask for it in a portable form, or object to how it is used. Write to privacy@chineseboost.com and it will be dealt with within a month.

No decision about you is made automatically, and nothing here builds a profile of you.

If you are unhappy with the answer you get, you can complain to the Information Commissioner’s Office at ico.org.uk.

Changes

This page is updated when what the site does changes. The date below is when it last did.

Last updated: 13 November 2026.