Text analysis basics: word counts, frequency, keyword density and readability

For everyday writing, text analysis means four measurements: the length of the text, the words it repeats, the share of the text a keyword takes up, and a readability score built from sentence and word length. You do not need special software for any of them. Paste the draft into the word counter for length, the word frequency counter for repetition, the keyword density checker for one term’s share and the readability score calculator for difficulty. The sections below explain what each number means, how it is calculated, and what it cannot tell you.

The four measurements at a glance

Question	Measurement	Tool	Typical use
Is it the right length?	Words, characters, sentences, paragraphs	Word counter	Essay limits, meta descriptions, speech length
What does it repeat?	Count of every word or phrase	Word frequency	Finding overused words, comparing drafts
How much of it is one term?	Keyword density (%)	Keyword density	Checking a page for stuffing or for a missing topic
How hard is it to read?	Flesch reading ease, grade levels	Readability score	Plain-language writing, web copy, school material

All four share the same definition of a word: a run of characters between spaces that contains at least one letter or digit. “Well-known”, “don’t” and “3.5” are one word each, and a lone dash is not a word. Because the tools agree on this, the totals match from one tool to the next.

Word and character counts: the baseline

Length is the first thing most limits are written in. The word counter reports words, characters (with and without spaces), sentences and paragraphs, plus a few derived figures:

• Unique words: how many different words appear, ignoring case. A low ratio of unique words to total words means a lot of repetition.
• Average word length in letters and average sentence length in words. Both feed the readability formulas.
• Reading time and speaking time. The defaults are 238 words per minute for silent reading and 183 for reading aloud, the averages for adult English readers in Brysbaert’s 2019 meta-analysis of reading-rate studies. You can change the reading speed in the options. The guide on reading and speaking speed covers the range around those averages.

A sentence ends at a period, question mark, exclamation mark or ellipsis followed by a space or the end of the text. An empty line also ends a sentence, and each list item (a line starting with a dash, bullet or “1.”) counts as its own sentence, so a heading or a bullet without a period is not run into the next line. Common abbreviations (“Dr.”, “e.g.”, “U.S.”), decimals and single initials do not end a sentence. Paragraphs are blocks separated by at least one empty line, so a text with single line breaks counts as one paragraph.

For limits counted in characters rather than words, such as posts and meta tags, use the character counter. It counts an emoji or an accented letter as one character, the way a reader sees it.

Word frequency: what the text repeats

A frequency list counts every word and sorts them from most to least used. Before counting, the tool lowercases each word, strips punctuation from its start and end, and turns curly apostrophes into straight ones, so “Cold,” “cold” and “COLD” are the same entry. Two- and three-word phrases never run across a sentence end or a comma, semicolon or colon: in “is cold, cold brew”, “cold cold” is not counted as a phrase.

The word frequency counter has four options:

Option	What it changes
Count	Single words, two-word phrases or three-word phrases
Skip common words (the, and, of…)	Leaves out about 180 English function words; for phrases, drops any phrase that starts or ends with one
Minimum count	Hides words used fewer times than this
Text result format	Tab-separated for a spreadsheet, CSV, or “word: count”

The “Share of words” column is always a share of all words in the text, including the common words you chose to hide. Skipping “the” and “and” removes them from the list but does not inflate the percentages of what remains.

Frequency lists are most useful for finding a word you lean on (“really”, “just”, “very”), checking that a report uses one term consistently instead of three synonyms, and comparing two versions of a draft. Two- and three-word phrases catch repetition that single words hide, such as a product name or a stock phrase you open every paragraph with.

Keyword density: frequency as a percentage

Keyword density is a frequency expressed as a share of the whole text. The keyword density checker uses this formula:

density = times used × words in the term ÷ all words × 100

Multiplying by the number of words in the term matters for phrases. A two-word phrase used 4 times in a 61-word text covers 8 of the 61 words, so its density is 13.11%, not 6.56%.

Enter a target keyword or phrase to see its density and how often it appears. Below that, the checker lists the top 15 single words, two-word phrases and three-word phrases used at least twice, with “Skip common words” switched on by default.

Density is a diagnostic, not a target. Google’s spam policies define keyword stuffing as filling a page with keywords or numbers to manipulate rankings, and give unnatural repetition as an example. They do not name a safe or ideal percentage. If one term’s density looks high, read the sentences that use it: when they sound forced, rewrite them. The keyword density guide covers what search engines say about it in more detail.

Readability: how hard the sentences are

Readability formulas turn sentence length and word length into a score. The best known, Flesch reading ease, was published by Rudolf Flesch in 1948. The readability score calculator reports six formulas:

Score	Based on	Read it as
Flesch reading ease	Words per sentence, syllables per word	Higher is easier; most text falls between 0 and 100
Flesch–Kincaid grade	Same inputs	US school grade
Gunning fog	Words per sentence, share of words with 3+ syllables	Years of schooling
SMOG	Words with 3+ syllables per 30 sentences	Grade; reliable only with 30+ sentences
Coleman–Liau	Letters per 100 words, sentences per 100 words	Grade; uses no syllable count
ARI	Characters per word, words per sentence	Grade

The tool counts syllables with English spelling rules and a list of common exceptions, so its syllable count is an estimate, as it is in every program. It also lists the sentences over 25 words (the first 20 of them), which are usually the quickest thing to fix. Flesch scores have no fixed floor or ceiling: very dense text can score below zero. The guide to Flesch reading ease scores explains the bands and what score to aim for.

One paragraph through all four tools

Here is a 61-word paragraph, run through each tool with default settings:

Cold brew coffee is coffee steeped in cold water for 12 to 24 hours. Because the water is cold, cold brew tastes smoother and less bitter than hot coffee. To make cold brew at home, mix one part coarse coffee with four parts water. Leave it in the fridge overnight, then strain it. Cold brew concentrate keeps for about a week.

Tool	Result
Word counter	61 words, 327 characters (267 without spaces), 5 sentences, 1 paragraph, 42 unique words, 12.2 words per sentence, 15 sec reading time, 20 sec speaking time
Word frequency	cold 6 (9.84%), brew 4 (6.56%), coffee 4 (6.56%), water 3 (4.92%)
Keyword density, target “cold brew”	13.11%, used 4 times
Readability score	Flesch reading ease 89 (Easy), Flesch–Kincaid grade 3.9, Gunning fog 6.2, Coleman–Liau 6.4, ARI 4.7

The paragraph is easy to read, but “cold” makes up almost a tenth of it. Three uses of “cold” in the second sentence (“the water is cold, cold brew”) are the obvious candidates for a rewrite.

Now the same idea written as one long sentence:

Cold brew, a beverage produced through prolonged extraction of coarsely ground coffee in cold water, is generally considered considerably less acidic and noticeably smoother than conventionally prepared hot coffee, particularly when the concentrate is subsequently diluted.

This version has 36 words in 1 sentence, and 13 of the words have three or more syllables. The Flesch reading ease drops to -13 (Very difficult) and the Flesch–Kincaid grade rises to 24. The tool also lists it under long sentences. Splitting it into three sentences and swapping “subsequently diluted” for “watered down” would do more than any other edit.

What the numbers cannot tell you

Every measurement here comes from counting, so none of them can judge meaning:

• A word count says nothing about whether the length suits the topic. Google’s helpful-content guidance says it has no preferred word count.
• A frequency list cannot tell deliberate repetition, such as a defined term in a contract, from careless repetition.
• A low density does not mean a page covers a topic, and a high one does not mean it covers it well.
• A readability score rewards short words and short sentences, even when they are out of order or unclear. A technical term the reader needs can add syllables without making the text harder for that reader.

A practical routine for a draft: check the length against your limit, run a frequency list to catch overused words, check the density of your main term if the page is meant for search, then fix the long sentences the readability tool lists. Read the result aloud before you publish; the numbers point to problems, and your ear decides whether they are real.