Privacy

SharpTyper sells no data, shows no ads and uses no third-party trackers. Usage statistics are only used to improve the application.

My choice

Always stored (needed for the service)

  • Account: username and hashed password (argon2), and a session cookie to stay logged in.
  • E-mail address, only if provided: it is only used to confirm the address and to send a password reset link, never for marketing. Messages go through an e-mail delivery provider, which only receives the address and the message.
  • Test and lesson results of logged-in accounts: they feed the statistics, progress and leaderboards.
  • Preferences (theme, keyboard, test language): kept in the browser. For accounts with the Supporter plan, keyboard settings are also kept on the account, so they follow you to another device.
  • Interface language: kept in a cookie, and in the account so it follows you to another device.

Cookies and browser storage

SharpTyper only uses cookies needed for the site to work: they require no consent and are never used for tracking.

  • sharptyper_session: keeps you logged in to your account. 30 days, removed when logging out.
  • sharptyper_lang: remembers the chosen interface language. 1 year.
  • Browser local storage (not a cookie, never sent automatically): light or dark theme, test settings, keyboard layouts and themes, course progress without an account, your choice about usage statistics, and a visitor identifier only if they are accepted. Erased by clearing the site's data in the browser.

Retention

  • Accounts: deleted automatically after 2 years without a login, with all their data. An e-mail warns 3 months before if an address is linked to the account.
  • Usage statistics: deleted after 13 months.
  • Confirmation and reset links: valid for 48 hours and 1 hour, then unusable.
  • Messages sent from the Contact page: received by e-mail by the publisher, kept while the request is handled.

Only if statistics are accepted

  • A random identifier, kept in the browser, to count visitors without an account.
  • Page views, with the country (derived from the browser's time zone, without the IP address), the operating system and browser, the chosen keyboard layout and the theme.
  • Results of tests and lessons taken without an account (speed, accuracy, mode), without the typed text.

No IP address is stored. This data is grouped in a dashboard reserved for the administrator and deleted after 13 months. If you decline, none of this is sent and the site works exactly the same way.

My rights

From the My account page: download all your data, change or remove the e-mail address, and delete the account with its whole history. For any other request: contact the site publisher. A complaint can also be lodged with the CNIL, the French data protection authority.

Understanding the statistics

Every measurement is recomputed by the server from each recorded keystroke (which key, at what time): the same rules apply everywhere.

Speed

WPM (words per minute)
Net speed. By convention a “word” is 5 characters, space included, whatever the actual length of the words: speed stays comparable from one text to another. WPM = characters of correct words ÷ 5 ÷ minutes. Only words that are correct when Space is pressed count (with their space; a mistake corrected before does not penalise them here), plus the correct start of the last word in progress.
Raw
Speed ignoring mistakes: all keystrokes ÷ 5 ÷ minutes, right or wrong, spaces included (Backspace is not counted). The gap between raw and WPM measures what mistakes cost.
Result chart
Second by second: cumulative WPM since the start (thick line), instantaneous raw speed of each second (thin line), and crosses for the seconds with mistakes.

Accuracy and consistency

Accuracy
correct keystrokes ÷ all keystrokes. A corrected mistake is still a mistake: accuracy measures what the fingers typed, not the final text.
Consistency
How steady the rhythm is, from 0 to 100%. It is computed from how much the raw speed varies from one second to the next (coefficient of variation), with the same formula as Monkeytype so that values are comparable. 100% = a perfectly steady rhythm.
Words ✓ / ✗
Words confirmed with Space while correct / wrong.
Keystrokes ✓ / ✗
Number of correct / wrong keystrokes, spaces included.
Confusions
“expected → typed”: the key that should have been pressed and the one pressed instead. A frequent confusion between two neighbouring keys often points to a misplaced finger.

Detailed analysis

Bigrams, trigrams
Sequences of 2 or 3 letters inside a word (“th”, “ing”). They are the real movements of typing: people rarely stumble on a single letter, often on a sequence.
Delay (ms), “slowest”
Time between two keystrokes. Only “clean” sequences count (a correct keystroke right after a correct keystroke, with no correction in between) and pauses longer than 2 seconds are ignored, to measure speed rather than hesitations or corrections.
Smoothed error rate
Rankings do not sort on the raw rate, otherwise a bigram missed 1 time out of 2 would rank above one missed 30 times out of 200. Each rate is pulled towards the overall average, as if 20 “average” attempts were added: it takes a lot of data for a unit to move away from the average. A bigram is only ranked from 10 attempts, a trigram from 8, a word from 2.
Most missed / slowest words
A word is “missed” if it had at least one mistake, even corrected. Its speed is measured from the previous space to its own space, then scaled to its length, so that a long word is not always the slowest.
Keyboard (heatmap)
Each key is coloured by its error rate or its average delay, relative to the other keys. A key is only coloured from 5 keystrokes.
Per finger, per hand
Key statistics grouped by the finger that should press them in touch typing (colours of the “fingers” mode), even if another finger is used in practice.

Progress

WPM (last 10), average WPM, record
Average of the last 10 tests, average over the chosen period, best test. Records are kept separately per test duration and mode: a numbers test is not compared with a words test.
WPM / 10 tests (trend)
The pace of progress: the slope of a line fitted on the last 50 tests, expressed in WPM gained (or lost) every 10 tests. Shown from 10 tests.
Progress chart
One dot per test; the line is the moving average over 10 tests, which smooths out good and bad days.
Minutes of typing
Total duration of the tests over the period.
Weak spots
This mode builds the text from the last 90 days: the 6 most missed bigrams, the 3 slowest and the most missed words, mixed with common words so it still reads like real text.

Course

Stars
The number of stars is the lower of two tiers: accuracy (★ from 90%, ★★ from 94%, ★★★ from 97%) and speed (★★ from the lesson's minimum speed, ★★★ from its target speed). One star passes the lesson and unlocks the next one: speed never blocks.
Stop on error
In lessons, a wrong key counts as a mistake but the cursor waits for the right key: no wrong movement is repeated and there is nothing to delete.
Skipped
A lesson moved past without being passed.
Lesson statistics
Keystrokes from lessons do not count in the statistics: the early exercises (“fff jjj”) would distort the analysis of real text.

Leaderboard

What is ranked
Each player's best test in each category (language, duration, mode), from 85% accuracy. The “weak spots” mode is not ranked: its text depends on each player's weaknesses.
Verified results
Every test is analysed when it is sent. It stays in the player's statistics but only enters the leaderboard after a review if it shows a sign of automation: keystrokes simulated by a script or pasted text (ignored by the page), a rhythm too regular for a human, repeated identical delays, an impossible sustained speed, 100% accuracy above 160 WPM, or a jump of more than 50% over the player's average.
Position, top %
The rank among the players ranked in the category; “top 10%” means being among the fastest 10%.