Free Word Counter & Keyword Density Checker

Paste the article body or another passage into the text box, or import supported UTF-8 TXT, Markdown or HTML locally. Review the extracted text before counting it. Add focus phrases, one per line, when you want independent counts for exact wording. Leave the field blank to review length and the broader frequency tables.

Text to analyze
Paste the part of the document you want included in the count. The default browser input limit is 50,000 UTF-16 units; a displayed symbol can use more than one unit. Check the active limit in the workspace.

Focus phrases — optional
Enter up to ten words or phrases, one per line, such as “content brief.” Each phrase is limited to 180 UTF-16 units. Optional features and limits follow the active admin settings.

Exclude common English stop words
Keep this selected to make frequent content words easier to inspect. Choose the default all-stop-word rule or the stricter any-stop-word rule, and add custom stop words or exact exclusions. These display filters preserve the full-draft word total and focus-phrase denominator.

Buttons: Run check · Clear

Processing note: The instant preview and complete report both run inside the isolated browser workspace, using the same counting engine. Run check does not submit the supplied draft to EduGuestPost or a paid provider. Nothing is saved automatically. Only an explicit save action passes the draft to the surrounding page for storage on this browser, with restore and delete controls. The surrounding website still makes normal page requests; separate enquiry submissions follow the privacy policy.

1. Choose the text you want counted. For an article-length requirement, use the representative article body and include headings, captions or references only when the brief requires them.
2. Paste plain text or review an import. A clean text copy makes the count easier to interpret than raw markup or copied menus. Markdown and HTML imports offer readable-text extraction previews. Choose the plain-text import format when you want markup counted literally; the main text box always counts supplied text literally.
3. Add optional focus phrases. Use up to ten phrases, one per line. Choose relevant wording rather than an unrelated keyword list. Each phrase has its own exact count and evidence.
4. Run the check. Read the completed length metrics, focus-phrase result and frequent-word tables.
5. Review the wording in context. Keep necessary terms and edit repetitions that add no useful meaning.
6. Refresh after editing. The live preview uses the same method as Run check. Editing the draft, focus phrases or analysis settings invalidates the previous completed report, so you cannot export stale results as if they describe the latest draft.

Frequency tables cover the complete returned data under the active analysis and display filters, rather than silently returning only a few leading rows. The 25-row page is a display limit, not the vocabulary size. Processing and export limits remain visible in the workspace.

Consider this short passage:

A content brief defines the audience and scope. The editor uses the content brief to plan useful sections. Clear evidence helps readers compare the available options. A conclusion answers the main question without repeating a sales claim.

With content brief as the focus phrase and the default lexical settings, the shared analyzer returns:

The phrase appears twice and contains two words. The calculation is therefore 2 × 2 ÷ 37 × 100 = 10.81%, rounded to two decimal places.

This short example explains the arithmetic. Its percentage is not a recommended target, a quality grade or a reason to add more occurrences.

For a single word, the completed tool uses:

Single-word density = occurrences ÷ total words × 100

For a multi-word phrase, it uses:

Weighted phrase density = occurrences × words in the phrase ÷ total words × 100

Some other tools count each phrase occurrence as one unit, regardless of how many words it contains. Their occurrence-only percentage can be different even when both tools find the same number of uses. In the example above, occurrence-only density would be 2 ÷ 37 × 100 = 5.41%.

Compare percentages only when the analyzed text, token rules, phrase length and denominator are the same.

Phrase tables can include overlapping sequences

The broader phrase tables examine consecutive tokens. Their sequences can overlap and can span sentence boundaries under the default frequency policy. Separate controls can disable overlapping windows and restrict sequences to a detected sentence. Exact focus phrases use non-overlapping matches by default; their overlap setting is named separately.

For example, content content content contains two consecutive two-word windows of content content. The frequency table returned two occurrences and a weighted value of 133.33%. The separate focus-phrase matcher returned one non-overlapping occurrence and 66.67%.

These are different counting paths. An overlapping weighted percentage is not the share of unique words covered, and table percentages should not be added together to describe the document. Read phrase counts as clues for editorial review.

Stop-word filtering changes the table view

Common words such as “the,” “and” and “of” can dominate a frequency list. The English stop-word setting helps surface other terms.

Under the default all-stop-word rule, phrases consisting only of common function words can be hidden, while useful sequences containing one may remain. A phrase such as alpha and can still appear. The stricter any-stop-word rule is an explicit alternative. Neither rule removes “and” from the original sequence and joins previously separated words into a new phrase.

The draft’s total word count and focus-phrase density remain based on the full analyzed text.

A repeated term may be necessary. An article about a specific qualification, product or technical process can use its name frequently and still read naturally.

Look for repetitions that weaken the passage:

• Several sentences opening with the same phrase.
• A paragraph repeating a sales message without adding evidence.
• A generic expression replacing a concrete explanation.
• An awkward exact phrase inserted where a simpler sentence would work.
• Multiple headings covering the same question with little new information.

Read the sentence before replacing a term. A synonym can change a technical meaning, obscure a product name or make a passage less clear. Keep the language your readers need and remove repetition that contributes nothing.

For sentence difficulty and reading-level estimates, continue with the Readability Score Calculator.

Use the current report after each edit

The instant summary and completed report share one analyzer. Exact focus matching respects the selected token boundaries, so art does not count partial or artful. Matching keys normalize composed and decomposed accents to NFC and fold curly apostrophes; the original supplied text is retained.

Context evidence comes from the same exact offsets as the occurrence count, rather than a separate substring search. Each phrase reports the total uses and the number of retained contexts and offsets. Rerun the completed check after an edit; stale reports are marked out of date and cannot be exported.

Plain text gives a clearer input

Analyze supplied text or supported local UTF-8 TXT, Markdown and HTML files. Import is local and shows a reviewable extraction preview. There is no page-URL fetcher, PDF/Word decoder or OCR in this workspace.

HTML extraction preserves separators between supported block elements and omits script/style content. Markdown extraction offers a readable passage without link destinations or formatting syntax. Review either preview before accepting it for analysis. The plain-text import option preserves literal markup, and the main text box never silently detects or interprets a document format. Keep original formatting elsewhere.

Counts depend on token and character rules

Hyphens, apostrophes, URLs, numbers, emoji and combining accents can be handled differently by word processors and web tools. This workspace counts characters in Unicode code points, not grapheme clusters; a displayed symbol may contain several code points. The full input is preserved, with NFC-normalized matching keys and UTF-16 evidence offsets into that input.

The default lexical mode counts letter/number runs with the disclosed apostrophe and hyphen rules. Optional browser language-aware segmentation uses Intl.Segmenter only when the browser supports the selected locale. Availability is shown rather than assuming every language is supported. Case handling is explicit; hyphen and apostrophe controls apply to lexical mode. A separately named counting setting can exclude numeric tokens. Numeric-frequency display filters never change the full counted-word denominator. If a publisher or application specifies an authoritative counter, check the final submission in that system.

Reading time and sentence totals are estimates

The default estimates use 220 words/minute for reading, 130 words/minute for speaking and 100 letters/minute for handwriting. You can adjust the available pace controls. These are disclosed planning assumptions, not measured performances. Sentence splitting is a punctuation-based estimate that can be affected by abbreviations and unusual formatting.

Use the completed result toolbar to keep the analysis with your editorial notes:

• Copy result: Copy the completed report summary. If clipboard access is unavailable, use the selectable-text fallback.
• Download CSV: Keep the visible-table action for rows in the displayed page and sort order. Choose the separately labeled full-frequency CSV when you need all retained rows across phrase lengths. Review the export scope before saving.
• Download JSON: Keep the full returned report, including metrics, retained phrase tables, tracked phrases, contexts, method settings and notices. The original draft is excluded unless you explicitly include it.

JSON is the full returned analysis, not automatically a copy of the original draft. Additional actions provide TSV, a TXT report or draft, native report printing and inert printable HTML. Export labels distinguish visible rows, full frequency data and the report summary. The default downloadable output cap is 10 MiB; check the active limit in the workspace. Contexts and phrases can reveal parts of the text, so choose what to share with your editor or client.