Typing Speed Test

A fast typing score can miss an employment requirement when it comes from a shorter test or uses a different error penalty. Clerical typing tests assess how much supplied text a candidate can copy in a set time and how accurately they reproduce it. The duration, scoring rule and permission to correct mistakes are part of the result.

Words per minute, or WPM, normally uses a standard word of five characters, including spaces and punctuation. It does not simply count the ordinary words in a passage. This convention prevents a passage full of short words from receiving an inflated speed just because it contains more word breaks. Ordinary words can still be the unit used to count errors, so a five-character speed word and a word error describe different things.

Gross WPM

The rate of text production before an error penalty is applied.

Net WPM

A speed after the stated error deduction. The name alone does not identify the formula. Subtracting errors per minute and subtracting the total number of errors give different results.

Accuracy

The proportion of text counted as correct under a particular scoring definition. It describes correctness rather than speed, and different definitions can treat omissions differently.

Error rules can change the outcome substantially. Several incorrect letters in one word may count as one word error or several character errors. A separate error allowance can reject a fast sample even when its accuracy percentage looks high. Correcting a mistake removes it from the remaining text when corrections are allowed, while the time spent deleting and retyping still belongs to the test.

Comparisons are most useful when the conditions match. Several differences deserve attention before treating a higher score as progress.

• Duration. A rate calculated from one minute does not demonstrate completion of a required five-minute sample. Pauses and correction time remain part of the timed task.
• Passage. Capitals, punctuation, numbers and unfamiliar wording change what must be typed. Repeating a familiar passage and copying an unfamiliar passage are different practice conditions.
• Keyboard and input method. A physical keyboard, a phone's soft keyboard and a different keyboard layout require different actions. Autocorrect, word suggestions, gesture typing and input-method composition can also change the text produced.
• Correction policy. Permission to revise earlier words gives a different task from correcting only the current word or leaving all mistakes in place.

Useful preparation follows the employer's current instructions, including the speed basis, sample length, error allowance and permitted assistance. Short sessions can isolate recurring mistakes, followed by complete samples under the required conditions. Consistently meeting those conditions is more informative than selecting the highest score from unlike tests.

Employment requirements vary by employer, title and jurisdiction. A practice result supports preparation and any permitted self-assessment; it does not establish that an employer will accept the result as certification. Confirm the requirement and accepted testing arrangements with the hiring organization.

How to Use This Tool:

Have the typing requirement from your posting available. The initial plan is a five-minute federal comparison with per-word errors and Backspace allowed.

1. Choose Employer rule. Use the federal or New Jersey choice when it matches your instructions, Custom rule from my posting for another supported combination, or Practice only (no rule) when you want practice without a pass/fail target.
2. Enter your posting's required speed for New Jersey or a custom rule. For a custom rule, also select Speed basis and enter Maximum errors if the posting sets a cap.
   Speed requirements must be numbers from 0 to 1000. Maximum errors must be a whole number in the same range. Correct an invalid value to restore Start. A blank speed requirement permits practice but gives no verdict; a blank maximum means no error cap.
3. Select Test duration from 1, 2, 3 or 5 minutes. Federal and New Jersey comparisons need a completed five-minute sample. A custom comparison uses the selected duration, which should match your posting.
4. Open Advanced to match Error counting and Corrections to the test conditions. You can also choose a starting Passage. The six passages are original text written for this test and continue in a fixed cycle. Rotate advances the starting passage between samples from a randomly chosen initial position.
5. Select Start and type the displayed passage into Response. The clock begins with your first typing action. Keep the window active until time expires. Text entry and permitted deletions stay at the end of the response; paste, drop, cut, undo and redo are blocked.
   Stop or Escape ends with partial evidence. Hiding the tab or switching away from the window interrupts the sample rather than pausing it. Restart discards the unfinished attempt and prepares a new sample.
6. Read Rule check and use the marked words to choose your next practice task. Each sample uses the settings selected when you press Start; later changes apply to the next sample. Take a fresh sample after changing a requirement or correction policy.

Interpreting Results:

Read the speed label beside the large figure before comparing it with a posting. Gross WPM, Net WPM and Gross WPM minus errors represent different calculations. Rule check shows the conditions that determine the verdict.

• Pass means the completed sample met the selected practice criteria. Confirm the employer's error counting and correction policy before using it to support self-certification.
• Fail means at least one criterion was missed. Work on the deciding criterion, such as the error allowance, rather than assuming that a higher gross speed will solve it.
• No verdict means the evidence is incomplete, the required duration was not used or a posting threshold is missing.
• Practice only identifies a completed session without an employer pass/fail target.

Accuracy concerns the characters that remain in your response. It can show 100% even when words were skipped, because omitted text is absent from its denominator. Check Word ledger for missed and extra words before treating a high percentage as a clean transcription.

The Key error map groups errors by the character that was expected. Use repeated substitutions and omissions to choose characters to practise. Extra inserted characters have no expected key and are excluded from this map. Its US QWERTY layout is a reference, not detection of your keyboard layout or evidence that a physical key is faulty. Letters share a key regardless of case, and shifted symbols share their displayed key, although case still matters to the score.

Use Session comparison to compare samples with matching duration, error unit, correction policy and input conditions. Rule comparison recalculates the same sample under both error units and the available rules; it cannot show how you would have typed under another correction policy. A dip in Speed over time can include pauses or correction work, so the chart alone does not establish its cause.

Copy the summary or download a result table when you need a record for later comparison. Table exports and session JSON retain exact score values, while chart data uses numerical approximations for plotting.

Technical Details:

Typing speed depends on the text remaining at the end of a timed interval. Error-adjusted speed also depends on the unit used to count mistakes and the penalty attached to each one. The following calculations make those practice conventions explicit.

Formula Core

Let C be the number of normalized characters remaining in the response, including spaces and punctuation, and let t be elapsed time in minutes. Let E be the selected word-error or character-error count. For positive elapsed time, gross speed G converts characters into five-character words and divides by time.

Net WPM, N, deducts errors per minute and cannot fall below zero. The gross-minus-errors score, S, deducts one WPM for each error in the whole sample. That second score can be negative.

The two deductions agree only when a sample lasts exactly one minute or contains no errors. Deleting and retyping consumes elapsed time, but deleted characters do not contribute to the final character count or remain as final errors. Corrections are reported separately as characters deleted and the subset identified as wrong when entered.

Accuracy A uses the number of correctly matched typed characters, M, divided by all retained typed characters. It is independent of whether word or character errors are selected for the speed penalty.

An empty response has zero speed and no accuracy percentage. An omission increases the error count without adding a typed character to this fraction, so accuracy cannot replace the error ledger or be used by itself to calculate net WPM.

How text becomes an error count

Character and word errors come from separate minimum-edit comparisons with the passage. Each substitution, omission or insertion costs one unit in its own comparison. Matching resumes after a skip instead of automatically marking every later character or word wrong. Untyped passage text beyond the attempted portion is excluded from the penalty.

Per-word counting treats each run of non-whitespace text as a word. Case and punctuation belong to that word, while per-character counting also assesses spaces. Several character errors can therefore belong to one wrong word, and extra spacing can affect character errors without adding a word error.

The incomplete-word exception ends when whitespace follows the partial word. A partial word containing a wrong character does not qualify. Where several comparisons have the same minimum error count, the chosen alignment determines the marks; it does not reconstruct the typist's intention.

For a short illustration, compare the passage red blue green yellow with the response red green yellow. The response contains 16 characters and all of them match, giving 100% accuracy. Skipping blue and its following space still produces one word error or five character errors. The later words remain matched.

Characters are counted as Unicode code points after normalization. Curly single and double quotes become straight quotes, en and em dashes become hyphens, and nonbreaking spaces become ordinary spaces. Canonically equivalent Unicode forms are normalized before comparison. Other case and punctuation differences remain significant. Passage whitespace is reduced to single spaces, but repeated spaces typed in the response are retained.

Employer comparisons and inclusive boundaries

A pass requires a completed sample and every applicable criterion. Equality meets a minimum speed or maximum error allowance. The error count below always uses the selected word or character unit.

The federal comparison draws on the US Office of Personnel Management (OPM) proficiency requirement for Clerk-Typist GS-2/4 and specified typing and office-automation positions. OPM permits self-certification and allows agencies to verify proficiency by testing. Its cited standard does not specify word-versus-character error units, gross-versus-net arithmetic, correction permissions or rounding. Using gross speed and the selectable error units above is a disclosed practice convention.

The New Jersey comparison follows the five-minute, WPM-minus-errors wording quoted in the Civil Service Commission's September 10, 2025 decision. That decision does not establish one minimum speed for every title. The required score and detailed error policy must come from the relevant announcement.

An interrupted or stopped sample has no verdict. A completed shorter federal or New Jersey sample also has no verdict, as does a completed New Jersey or custom sample without its required speed. These conditions are checked before assigning Pass or Fail. Completing a shorter custom sample can produce a verdict when that duration and the other entered criteria match the intended rule.

Corrections, timing and precision

Backspace allowed permits deletion backwards from the response's end through earlier words. Current word only prevents deletion across the most recently entered space, following the correction pattern in the Missouri Office of Administration's published typing-test instructions. Immediately after a space, that policy allows no deletion. No corrections blocks deletion altogether. These choices do not enable moving the cursor into earlier text to edit it.

The timer starts on the first typing action, including the start of input-method composition. Only committed composition text enters the response; intermediate text is excluded. Input arriving at or after the deadline is excluded. A completed sample uses the full selected duration, while a partial sample uses elapsed time rounded up to a millisecond. That precision is a counting convention, not a calibration of physical key timing. The response is limited to 25,000 normalized characters.

Pass/fail comparisons use exact values and the entered decimal requirement. Textual speed and accuracy figures are rounded down to one decimal place, including downward rounding for negative scores. A displayed decimal is therefore not the value used to decide a close boundary.

Cumulative speed and interval pace

Cumulative gross and net speed are calculated from the start of the sample at five-second checkpoints. The gross-minus-errors score is also plotted when it is the selected scoring basis. The first checkpoint needs five elapsed seconds.

The 15-second pace uses the growth in retained characters over each complete 15-second interval. If C₁ and C₂ are the retained character counts at its beginning and end, the interval pace P₁₅ is:

Dividing by five converts characters to conventional words, and multiplying by four converts a 15-second rate to WPM. This pace has no separate error deduction. Deleted and retyped text is reflected only in the change in retained length, and negative growth gives zero. A final interval shorter than 15 seconds does not create another interval point.

Input Assistance and Privacy:

The response field requests that autocorrect, automatic capitalization, spellcheck and autocomplete be disabled. Browser or operating-system keyboards can still offer assistance, and dictation is not blocked. For an unaided employment practice sample, turn off assistance in the keyboard's own settings and use the expected input method. A phone score should be compared with other phone samples taken under the same conditions.

Typing capture and scoring take place in your browser; the response is not submitted to a scoring service. The typing test does not request camera or microphone access. Page and chart resources can still require a network connection, and copy actions depend on browser clipboard access.

• Up to 20 session summaries are saved in this browser, including their settings, scores, error counts, accuracy and completion status. Raw responses and timed edits are not automatically saved in that history. Labelled example sessions are not saved.
• Download session JSON includes the passage, full response and timed edits as well as the results. Review that detailed evidence before sharing the file. A page URL carries settings rather than your response or result.
• If browser storage is unavailable, export the comparison before leaving the tab. Unreadable or unsupported saved history is left untouched until you choose Clear saved sessions. Clearing history keeps the current result.

Worked Examples:

References:

• General Schedule Qualification Standards: Clerical and Administrative Support Positions, US Office of Personnel Management.
• In the Matter of Tammy Shupp, New Jersey Civil Service Commission, September 10, 2025.
• Merit System Typing Test, Missouri Office of Administration.
• Evaluation of Text Entry Techniques, I. Scott MacKenzie, Text Entry Systems, 2007.
• How do People Type on Mobile Devices? Observations from a Study with 37,000 Volunteers, Palin, Feit, Kim, Kristensson and Oulasvirta, ACM MobileHCI, 2019.