How these tools work

Reading-speed calculations

Words per minute (WPM) = total words ÷ elapsed minutes, with seconds converted to fractions of a minute (90 seconds = 1.5 minutes). The WPM Calculator rejects samples under 50 words (too noisy), elapsed times of zero (division by zero) and results above 3,000 WPM (outside any plausible reading speed — almost always a typo).

Time estimates = word count ÷ WPM. The Words-to-Minutes Calculator applies this with preset paces: silent reading at 238 WPM (the measured adult average — see sources), reading aloud at ~150 WPM and presentations at ~130 WPM (approximate common figures for clear deliberate speech), and audiobooks at ~150 WPM narration times your playback multiplier. Presets are estimates, labeled as such; real texts and real voices vary.

Benchmarks and their sources

Average adult speed — about 238 WPM for silent non-fiction reading (roughly 260 for fiction), from Marc Brysbaert's 2019 meta-analysis of 190 studies in the Journal of Memory and Language. This is the most defensible estimate available and the anchor for every benchmark band we show.

The comprehension ceiling — about 400–500 WPM. Rayner, Schotter, Masson, Potter & Treiman's 2016 review in Psychological Science in the Public Interest found comprehension declines as speed is pushed, with genuine reading topping out around this range. Above it, the honest word is skimming. Our benchmark bands ("average", "strong", "past the research ceiling") are approximate interpretations of these two sources — bands, not verdicts.

Improvement targets — roughly 30–60% over about six weeks of short daily practice is what the published training literature supports as a typical outcome. It is not a guarantee, and no tool on this site will set you a target above the 400–500 WPM ceiling, because no honest tool can.

Book and PDF time estimates

The book calculator computes total minutes as words ÷ WPM, sessions as total minutes ÷ your daily minutes (rounded up), and calendar time by spreading those sessions across your reading days per week. When you enter pages instead of words, total words = pages × a words-per-page estimate you can edit (novel ~275, general non-fiction ~300, technical ~400, dense academic ~450 — typical figures, not facts about your copy). Estimated values are labeled as estimates everywhere they appear; the improved-speed comparison uses +30%, the low end of the training literature's typical range, capped at 500 WPM.

The PDF calculator extracts text with pdf.js entirely in your browser and counts the words it finds — a measured count, not an estimate. Scanned or image-only PDFs have no text layer; the tool says so (it does not do OCR) and falls back to your manual page count at ~300 words per page, labeled as the estimate it is.

Readability formulas

The level checker computes the published Flesch equations verbatim: Reading Ease = 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words); Flesch–Kincaid Grade = 0.39 × (words ÷ sentences) + 11.8 × (syllables ÷ words) − 15.59. Syllables are counted with a standard heuristic (vowel groups with silent-e handling), which is approximate by nature. Three honest limits: the formulas are calibrated for English only (the tool declines other languages rather than mis-scoring them); they see only sentence and word length, not meaning; and they describe the text's demand — never a reader's intelligence.

How comprehension changes the interpretation

A WPM number with low comprehension mostly measures skimming. When you give the WPM Calculator a comprehension score under about 70%, it says so instead of congratulating you — speed only counts when understanding comes with it. That threshold is a practical rule of thumb we apply consistently, not a published constant. The full speed test measures comprehension directly with questions, which is why we call it the better measurement.

Limitations, honestly

• Self-timed inputs inflate: people round down time and round up word counts. The test measures; the calculators trust you.
• Reading speed is material-dependent — one person legitimately has different speeds for fiction, news, and technical text. A single number is a simplification.
• Benchmark bands are approximate by construction; being 10 WPM either side of a boundary means nothing.
• Speaking and narration presets vary by speaker, language, and text. Rehearse anything that will be timed for real.

Privacy and data retention

• All tool calculations run in your browser. Nothing you type into a tool is sent to or stored on our servers — with one opt-in exception: if you explicitly create a public link, the previewed result strings and your optional display name are stored for 90 days (see the sharing section below).
• PDFs are parsed locally by pdf.js in your tab — never uploaded, never stored; the filename and contents never reach us, analytics, or any share output.
• Pasted text in the level checker stays in your browser; only the resulting difficulty band (e.g. "Standard") is sent to analytics, never the text.
• Analytics receive only which tool was used and generalized labels (e.g. a benchmark band) — never your exact inputs, text, names, or files.
• Result cards are drawn in your browser and downloaded straight to your device; they never pass through us.
• Share links point at the tool page with anonymous attribution tags (ref=result-share) — they contain no identity and nothing about you. The share message contains exactly the numbers you saw, and you compose where it goes.
• Retention: since we don't receive tool data, there is nothing to retain or delete — except opt-in public results, which auto-expire after 90 days.

Comprehension test and improvement plan

The comprehension test uses passages written originally for this site — self-contained, so every question is answerable from the passage alone. Comprehension % = correct answers ÷ questions; speed = passage words ÷ your reading time (the timer runs only between Start and Finished, and the questions are untimed). Implausible times — under 3 seconds, or an implied speed above 1,500 WPM — are flagged and the speed score marked unreliable rather than celebrated. Below ~70% comprehension we say plainly that the speed number mostly measured skimming.

The improvement plan scales its target from the training literature's ~six-week 30-60% anchor: roughly 8-20% for 14 days, 15-35% for 30 days, 30-60% for 60 days — typical outcomes, never guarantees, always capped at the 400-500 WPM ceiling. A comprehension score under 70% pivots the whole plan to comprehension-first. The plan is educational guidance generated from your inputs, and works fully without an account.

How public sharing works

• Public result links are opt-in and preview-first: before anything is created, you see exactly the strings that will be public — the tool name, the headline result, and one secondary line. Nothing else exists to leak.
• Anonymous by default; a display name is optional and capped at 40 characters.
• Links use unguessable 22-character random IDs, expire automatically after 90 days, and are excluded from search indexing (noindex).
• The share/card/link text never includes your inputs, pasted text, file names, or answers — only the previewed result strings.
• Creation is rate-limited per network to prevent abuse.