People routinely search for a "word counter with frequency" as if it were one thing. It is two jobs that happen to sit next to each other. Counting tells you whether your text fits; frequency tells you what your text is made of. You usually want both, in that order.
Two tools, two questions
Word count answers: is this the right length? It produces totals — words, characters with and without spaces, sentences, paragraphs, reading time — which you check against an external constraint: a submission limit, a character cap, an editor's brief.
Word frequency answers: what am I actually saying? It produces a ranked list of every distinct word and how often it appears, which you read against your own intent: did I lean on one word twelve times, is my target keyword present at all, is this paragraph 40% filler.
A counter can tell you your draft is 1,847 words. Only a frequency list can tell you that 22 of those words are "however".
Common word and character limits
These are the constraints people actually write to. Platform limits change, so check the current spec before anything important goes out.
| Where | Limit | Counted in |
|---|---|---|
| X / Twitter post | 280 | Characters (higher on paid tiers) |
| SMS, single segment | 160 | Characters (drops to 70 if any emoji or non-Latin character is used) |
| Meta description | ~155 | Characters — a display convention, not an enforced cap |
| Page title tag | ~60 | Characters before truncation in results |
| LinkedIn post | 3,000 | Characters, with a "see more" fold after the first few lines |
| Instagram caption | 2,200 | Characters, truncated in feed at around 125 |
| Common App essay | 650 | Words (250 minimum) |
| UCAS personal statement | 4,000 | Characters — check the current format, it was restructured recently |
| Journal abstract | 150–300 | Words, varies by publication — always check the specific guidelines |
| Blog post, SEO | 1,000–2,000 | Words — a convention among content teams, not a ranking rule |
Two deserve a caveat. The 155-character meta description is a rough guide: search engines truncate by rendered pixel width, not character count, so wide capitals get cut earlier than narrow lowercase letters. And the 1,000–2,000 word blog range is an industry habit, not a documented ranking factor — length correlates with covering a topic thoroughly, which is the thing that actually matters.
A worked example
Take this paragraph:
Our new dashboard makes reporting faster. The dashboard pulls data from every connected source, and the dashboard refreshes automatically. Teams that used to spend hours on manual reporting now spend minutes. Because the dashboard is fast, teams actually use it, and because teams use it, the data stays fresh.
Run it through the Word Counter and you get:
| Metric | Value |
|---|---|
| Words | 49 |
| Characters (with spaces) | 310 |
| Characters (without spaces) | 262 |
| Sentences | 4 |
| Reading time | ~13 seconds |
Useful for one thing only: it fits in a tweet with room to spare, and it is far too short for the abstract it was probably meant to be.
Now run the same text through the Word Frequency Counter:
| Word | Count | % of text |
|---|---|---|
| dashboard | 4 | 8.2% |
| the | 4 | 8.2% |
| teams | 3 | 6.1% |
| reporting | 2 | 4.1% |
| data | 2 | 4.1% |
| spend | 2 | 4.1% |
| because | 2 | 4.1% |
| use | 2 | 4.1% |
That list says three things the word count could never say.
"dashboard" at 8.2% is too dense. In 49 words the subject appears four times, three of them as "the dashboard" starting a clause. That is not keyword optimisation, it is a writer who forgot pronouns exist. Two of the four should become "it".
"because" twice in one sentence is a structural tell. The last sentence runs "Because... and because...", which reads as circular. The list surfaces the repetition; deciding it is a problem is still your job.
The keyword is present but the framing is not. If this text is meant to rank for "reporting dashboard", "reporting" appears twice and never adjacent to "dashboard". Frequency confirms the ingredients are there; it cannot tell you they are combined.
Notice also that "the" ties for first place at 8.2%. That is normal, and it is why stop word filtering exists.
Why stop words dominate every raw frequency list
In English the most common words are function words: the, of, and, to, a, in, is, it. They carry grammar rather than meaning, and they top the frequency list of a legal contract, a recipe, and a love letter alike. A raw list therefore opens with several rows of noise before it says anything about your subject.
Filtering stop words pushes content words to the top. The caveat: stop word lists are editorial choices, not standards. Tools filter different sets, and a word that is noise in one context ("just" in casual prose) is meaningful in another ("just" in a contract). If a word you care about vanishes, the filter ate it.
What counts as a "word"?
Every tool draws these lines slightly differently, which is why two counters disagree about the same text:
- Hyphenated compounds. "well-being" is usually one word, because it functions as one unit. Some tools split on the hyphen and report two. Em-dashes used without spaces — like this — trip up naive splitters entirely.
- Contractions. "don't" is one word everywhere sensible. A tool that splits on the apostrophe reports "don" and "t".
- Numbers. "2026" and "$1,500" are normally counted as one word each. Whether "1,500" survives as one token depends on whether the tool treats the comma as punctuation.
- URLs and email addresses. Almost always one token, occasionally shredded into several.
Word processors and browser-based counters differ by a percent or two on the same document for exactly these reasons. If you are near a hard limit, count with the tool your reader will use — the university portal or the platform's own field — and treat everything else as an estimate. Formatting artefacts compound this: double spaces and stray line breaks pasted from a PDF inflate character counts and split words in half, so run text through a text cleaner first.
The reading time formula
Reading time is a straight division:
Minutes = word count ÷ reading speed
The conventional speed is 200–250 words per minute for silent adult reading, which is where the "5 min read" labels come from. Treat it as a convention, not a fact about your particular reader: dense technical prose, code blocks, and text in a second language all read considerably slower, while skimmable listicles read faster. Useful for setting expectations, not for scheduling.
A combined workflow
- Count first. Cutting 400 words is a different job from cutting 40, and you want to know which one you have before you start editing.
- Cut to roughly the target. Get within about 10% before analysing frequency, since heavy cutting changes the distribution anyway.
- Run frequency with stop words filtered and read the top 15 rows.
- Look for three patterns: a content word above about 4–5% in a short text (probably overused), filler adverbs high in the list (very, really, actually, just, quite), and your intended subject missing from the top rows — which usually means the piece is not about what you think it is.
- Recount after editing. Replacing repetition with variety often adds length.
- Visualise if presenting. A word cloud makes the same distribution legible to an audience that will not read a table.
Frequently asked questions
Can one tool do both? The Word Counter handles the totals and reading time; the Word Frequency Counter handles the ranked list. Paste the same text into each — both run in your browser, so nothing is uploaded.
What is a good keyword density? There is no target number, and chasing one produces worse writing. Use frequency as a sanity check: is the term present at all, and does it appear so often that a human would notice? Both failure modes are visible in the list.
Why does my word processor give a different count? Different tokenisation rules — usually hyphens, contractions, or whether headers, footnotes, and captions are included. A gap of one or two percent is normal.
Should I filter stop words? Yes if you want to know what a text is about. No if you are studying writing style, where function word patterns are the interesting signal.
Does frequency analysis work on other languages? Counting does, for any language with spaces between words. Stop word filtering is language-specific, and a list built for English will not filter anything useful from French or German.