Keyword Extractor: Get Keywords from Text or a URL

Last updated: What’s new

Extract keywords from text online or analyze a public website URL. Review word and phrase frequency, source evidence, and context, then switch to Relevance to shortlist topical phrases.

Find words and phrases already present in your text or a public webpage. Use Frequency to inspect repetition or Relevance to review topical phrases, with context for each result.

0 words · stays in this browser
Sort by

For keyphrase discovery, try Relevance with 1+ occurrence and All lengths. Auto stop words only apply when one supported language has enough evidence.

How the Keyword Extractor Works

The report keeps measurement separate from interpretation:

  1. Choose text or URL. Text is processed locally. URL mode fetches public HTML and separates title, meta description, H1 evidence, and cleaned body text.
  2. Segment Unicode words. Modern browser and server segmentation preserves diacritics and supports many writing systems; a Unicode-regex fallback is used when needed.
  3. Apply an explicit stop-word rule. Auto mode supports English, Lithuanian, Spanish, German, and French and falls back to removing nothing when evidence is uncertain.
  4. Count within sentence boundaries. One-, two-, and three-word strings are normalized and counted without building phrases across line or sentence breaks.
  5. Show evidence. Every row includes exact frequency, density, detected source zones, and a first-occurrence context snippet.
  6. Optionally sort by relevance. The Relevance view ranks phrases of up to four words with a statistical estimate based on position, frequency, capitalization and spread in this text alone. It is not search volume.

What Keyword Frequency Can and Cannot Tell You

Frequency is useful evidence about a document, but it is not a verdict about SEO value:

  • Focus review, compare recurring language with the page's intended subject and audience.
  • Repetition review, inspect context when a word or phrase appears unusually often; there is no universal “safe” density.
  • Page-element alignment, use source badges to see whether language appears in the title, H1, description, or only the body.
  • Comparable-page analysis, compare like-for-like pages, then investigate meaningful differences instead of copying a competitor's wording.
  • Research boundary, frequency provides no search volume, difficulty, ranking, traffic, conversion, or semantic-quality data.

Use the Meta Tag Checker for a dedicated metadata review and the Website Word Counter for readability and length. If you need new seed variations rather than terms already present in a sample, use the separate Keyword Generator.

How to Use Extracted Keywords

Start with the evidence in each row, then make an editorial decision:

  • Use 1+ for discovery, especially on short copy. Switch to 2+ or higher when you specifically want repeated language.
  • Choose the view for your task, Relevance for a shortlist of topical phrases, or Frequency for exact repeated wording. Frequency stays the default view.
  • Read the snippet, because the same string can be a core topic, navigation residue, a disclaimer, or an incidental mention.
  • Check source zones, but do not force a phrase into every element. Titles, descriptions, headings, and body copy have different jobs.
  • Export a reproducible view, using the same phrase-length, frequency, and row settings when comparing pages.
  • Validate outside this report, with search-demand, intent, SERP, customer, and subject-matter evidence before changing content.

For a complete workflow, follow our keyword research guide. It explains the research steps that a document-frequency analyzer cannot perform.

A Worked Pasted-Text Example

The Try sample text button runs this example locally with English stop words and a minimum frequency of 1+:

Boiler repair starts with a diagnosis. Our boiler repair service checks pressure and leaks. Book boiler repair after reviewing the fault report.

Checked with this page’s extraction function on 28 October 2026: 22 analyzed words. Frequency keeps the exact phrase count; Relevance also puts “boiler repair” first for this sample.

PhraseOccurrencesSource zoneContext
boiler repair3 (13.64% of the 22-word denominator)Pasted text“Boiler repair starts with a diagnosis.”

Select 2-word phrases and 2+ occurrences in Frequency to isolate repeated phrases. The summary, table and exports then use that same selection. For a public webpage, check its source badges to see whether a phrase occurs in the title, H1, meta description or body.

The sample demonstrates extraction and counting. It provides no search-volume or commercial-value evidence and sets no desirable keyword density.

TaskSuitable evidence
Extract phrases already presentThis tool: counts, source zones and snippets.
Brainstorm new variationsKeyword Generator: unvalidated pattern ideas.
Find queries that reach a page you ownSearch Console: filter Search results by the exact page, then review its Queries tab.
Prioritize search opportunitiesKeyword research: combine search demand, intent, audience needs and a current results review.

What’s new

  • CSV exports protect text cells from spreadsheet formulas, and failed requests show readable explanations with cancellation support.

  • The keyword summary follows your selected phrase length and minimum frequency, and a local sample lets you try the report.

  • Reads only the page’s main content, so CSS and layout words no longer appear as keywords, and ASP.NET pages keep their text.

Next steps

Keyword Extractor related tools and articles

Continue with the closest follow-up checks and guides based on this tool's topic, crawl intent, and optimization workflow.

How to Do Keyword Research (Without Expensive Tools)

How to Write Content That Ranks: SEO Copywriting Guide

On-Page SEO Checklist: 15 Checks Before You Publish

Keyword Generator

Keyword Density Checker

Website Word Counter

Keyword Extractor: FAQ

What does the keyword extractor measure?
It counts normalized one-word, two-word, and three-word strings in the analyzed text. Count is raw frequency and density is count divided by total analyzed words. Neither metric is an importance, search-volume, ranking, competition, or intent score.
What does the Relevance view show?
Relevance is a statistical estimate from this text alone, not search volume. It ranks phrases of up to four words with a deterministic, YAKE-style score that weighs where a phrase first appears, how often it repeats, capitalization, and how evenly it is spread. Capitalized names such as Google Search Console stay together, English singular and plural forms are counted together, and a shorter phrase found almost only inside a longer one is listed once. Pasted text and URLs are ranked with the same code. Frequency stays the default view.
Can I extract keywords from short text?
Yes. Choose a minimum frequency of 1+ to keep phrases that occur once. The 2+ and 3+ settings are useful for repetition analysis, but can hide every result in a short sample.
Should I use Frequency or Relevance to find keyphrases?
Use Relevance with 1+ occurrence and All lengths to shortlist topical phrases, including names of up to four words. Use Frequency to inspect exact repeated words and two- or three-word strings. Phrase length, minimum frequency and row count filter the displayed results and the rows you copy or export. Review the context before choosing a phrase for your content.
Does it support accented and non-Latin text?
Yes. It uses Unicode-aware word segmentation when the browser or server supports it, preserving diacritics and scripts such as Cyrillic, Greek, Arabic, CJK, and Hangul. Segmentation quality varies by script and runtime, so review phrases in their displayed context.
How does the stop-word filter work?
Auto mode compares common-word matches for English, Lithuanian, Spanish, German, and French. It only applies a list when one language has enough evidence; otherwise it removes nothing. You can select a language explicitly or turn filtering off. These are compact lists, not linguistic parsers.
How do I get keywords from a website?
Choose Extract from URL and enter the page address. The tool fetches that one public page and counts words and phrases in its title, meta description and main content (the main element or a single article, otherwise the cleaned body, with scripts, navigation, header, footer and similar boilerplate removed), and source badges show whether a match was in the title, H1, meta description or body. It lists the language on that page, not the keywords the site ranks for, and it rejects HTTP errors, empty responses and non-HTML files.
Does URL mode render JavaScript?
No. It analyzes the HTML returned to the server-side request, reading at most the first 2 MB; a longer page is labeled a partial report. Client-rendered, authenticated, interaction-gated, and consent-gated text can be absent, while bot or CDN responses can differ from your browser.
Can it find every keyword a website ranks for?
No. URL mode extracts terms present in one public webpage. It does not crawl the whole site or query a search-ranking database. To see search queries that already bring impressions or clicks to a site you own, use Google Search Console and filter the report to the relevant page.
Why might a source badge and frequency count differ?
H1 is shown as an evidence zone but is already part of body text, so it is not counted twice. Source badges indicate where a phrase was found; they are not extra occurrences.
Are extracted terms automatically good SEO keywords?
No. The report describes language in one sample. Validate candidates against real search demand, intent, SERP context, business relevance, and editorial judgment. Use a keyword generator separately when you want new variations rather than extraction.
What happens to my text or URL?
Pasted text is analyzed in your browser and is not submitted by this tool. URL mode sends the public URL to Web Aloha so the server can fetch and parse the page; the endpoint has no application logic that stores the page or report.