Culture · Craft
People scan. A readability score can't tell you if they understood.
This page covers how people actually read on a screen, the structural habits that match that behaviour, and a working Flesch–Kincaid calculator you can run on your own writing — with an honest account of exactly what that score does and doesn't measure.
Every figure above is sourced below — see Sources for the exact studies and reports.
01 — How People Actually Read
Eye-tracking says people scan, not read.
What you're browsing for: a "readability score" you can hit and call the writing done.
What you actually need to know: a score only checks sentence and word length — whether anyone understood, believed, or acted on the writing is a separate question the formula was never built to answer.
Nielsen Norman Group's original eye-tracking research, recorded across 232 users looking at thousands of web pages, found that reading on the web commonly follows an F-shaped pattern: a horizontal stripe across the top of the content, a second shorter horizontal stripe further down, and a vertical stripe scanning down the left edge. A later eye-tracking study, and NN/g's own decades of repeat testing, confirmed the pattern is still present — including on mobile.
The headline finding isn't that people refuse to read; it's that on an average web page, people read at most around 28% of the words, with 20% being more typical. Whatever sits inside that fraction — the first line of a paragraph, a bolded phrase, a heading — is disproportionately what actually gets read.
That is a statement about attention, not intelligence or effort. It means structure carries more of the communication burden on a screen than it would in a printed book read start to finish.
02 — Structure
Write for the fraction that gets read.
- Front-load the point. Put the conclusion or answer in the first sentence of a paragraph, not the last — the F-shaped pattern means that first line gets disproportionate attention.
- One idea per paragraph. A paragraph that argues two things forces a scanning reader to either read all of it or misunderstand it; neither is the outcome you want.
- Headings as a table of contents. A reader deciding whether to keep going is scanning headings, not prose — a heading that only makes sense after reading the section beneath it has failed at its one job.
- Bold and lists sparingly, but deliberately. These are the visual anchors the eye catches inside the F-shape; overusing them removes the contrast that made them useful.
- Short sentences are a means, not the goal. A short sentence that's vague still fails; the goal is a sentence a scanning reader can finish before their attention moves on.
None of this is a claim that long-form writing is wrong — the pillar pages on this desk's own Content cluster are proof otherwise. It's a claim that structure is what decides whether the other 72–80% of a long piece ever gets read at all.
03 — The Calculator
Paste your own text. See the actual formula.
This computes the real Flesch Reading Ease and Flesch–Kincaid Grade Level formulas, live, from whatever you paste below. Nothing is sent anywhere — it runs in your browser and the score updates as you type.
Your text
What that computes to
- Words
- –
- Sentences
- –
- Avg. words / sentence
- –
- Avg. syllables / word (approx.)
- –
- Flesch Reading Ease
- –
- Flesch–Kincaid Grade Level
- –
Paste text above to see a reading-ease band.
What this score can't tell you: whether the words are familiar to your specific reader, whether the ideas are logically ordered, whether the layout helps a scanning eye, or whether anything in it is true. It counts sentence length and syllables — nothing else. The grade-level formula itself was calibrated in 1975 against a sample of U.S. Navy enlisted personnel reading technical training manuals; it was never a claim about your reader.
04 — Plain Language, Officially
What the U.S. federal guidelines actually say.
The Federal Plain Language Guidelines, produced under the U.S. Plain Writing Act of 2010, describe plain language as writing that lets the intended reader "find what they need, understand what they find, and use that information" — a definition about the reader's outcome, not the writer's vocabulary.
- Write for your audience. Know who's reading before deciding what counts as "simple."
- Use common, everyday words over formal or technical alternatives, unless the technical term is the one your reader actually needs.
- Prefer active voice and address the reader directly ("you") where it fits the context.
- Organize for the reader's task, not the order the information happened to be gathered in.
- Keep sentences reasonably short — but never at the cost of meaning. The guidelines are explicit that clarity, not brevity for its own sake, is the actual goal.
Notice that none of these is "hit a target readability score." The guidelines describe habits and a test (can the intended reader find, understand and use it), not a number to optimize.
05 — FAQ
Common questions.
Is a low Flesch-Kincaid grade level the same as "good writing"?
No. A low grade level only means shorter sentences and shorter words; it says nothing about whether the content is accurate, well-organized, or genuinely easier to understand for a specific reader.
Why does the calculator need my own text instead of giving me an average?
Because there's no honest "average" to give you — the score depends entirely on the sentence and word choices in the specific text you paste, computed the same way the original formula computes it.
Does a high syllable count always mean a word is hard to understand?
No. Familiarity matters more than length. A common three-syllable word like "computer" is easier for most readers than an unfamiliar one-syllable word from an unfamiliar domain, but the formula can only count syllables — it cannot tell familiar from unfamiliar.
Did Nielsen Norman Group's research say people don't read online at all?
No. Their research says people scan rather than read word-for-word, reading roughly a fifth to a bit over a quarter of the words on an average web page. Structure — headings, front-loaded key points, short paragraphs — determines which words those are.
Should I trust a readability score as the final word on whether my writing works?
No single score should be the final word. Treat it as one quick, mechanical check among several — reading the piece aloud, or having an actual member of your intended audience read it, will tell you more than any formula.
06 — Sources
Where the research, the formula and the guidance come from
- Flesch, R. (1948). “A New Readability Yardstick.” Journal of Applied Psychology, 32(3), 221–233 — the original Reading Ease formula.accessed 27 Jul 2026
- Kincaid, J.P., Fishburn, R.P., Rogers, R.L., & Chissom, B.S. (1975). “Derivation of New Readability Formulas for Navy Enlisted Personnel.” Research Branch Report 8–75, Chief of Naval Technical Training — the grade-level recalibration used by the calculator above, including its original validation population.apps.dtic.mil · accessed 27 Jul 2026
- DuBay, W.H. (2004). “The Principles of Readability.” Impact Information — the review used for this page's honest account of what readability formulas do not measure.files.eric.ed.gov · accessed 27 Jul 2026
- Nielsen Norman Group, “F-Shaped Pattern For Reading Web Content” — the original 2006 eye-tracking study.nngroup.com · accessed 27 Jul 2026
- Nielsen Norman Group, “F-Shaped Pattern of Reading on the Web: Misunderstood, But Still Relevant (Even on Mobile)” — the current-relevance update and the 20–28%-of-words figure used in Section 01.nngroup.com · accessed 27 Jul 2026
- Plain Language Action and Information Network (PLAIN) / Digital.gov, “Federal Plain Language Guidelines” — the official U.S. federal guidance referenced in Section 04.digital.gov · accessed 27 Jul 2026
About the calculator: word, sentence and syllable counts are computed live from the text you paste, using the same published Flesch and Flesch–Kincaid formulas as sources 1–2. Syllable counting from spelling alone is inherently approximate — this page uses the same kind of heuristic most readability tools use, not a dictionary lookup. No score interpretation band, benchmark or comprehension claim beyond Flesch's own published scale (source 1, as reviewed in source 3) appears anywhere on this page.
Related
Keep reading
How to Build a Content System
Pillar/cluster structure, a repurposing matrix, and a cadence planner from your own capacity.
Commerce · CalculatorUnit Economics for a Small Business
Contribution margin, CAC payback, cohort retention and cash conversion cycle, computed from your numbers.
Intelligence · GuideHow to Read a Market
A sortable ledger of seven verified free public data sources, and the biases to watch for.
purposed2 · HubAll of Culture
Every culture guide on the desk, in one place.
The Concierge
Talk through your writing.
The purposed2 concierge can walk through what the calculator found, in plain words — and will say when a working session with Amit is the better next step.