Page Word Counter
Enter any URL to count the words on it. The body copy is separated from the navigation, header and footer, so the number describes the page rather than the template around it.
Short answer
A page word counter reads the HTML a URL returns, strips the markup, and counts the visible text. The number that matters is the body copy alone: navigation, headers and footers are words on the page that no reader came for, and a site-wide footer of 400 words adds 400 to every page on the site. This tool reports both figures plus the gap between them, and states which boundary it drew, in order of preference a main element, an element marked role="main", the longest article, or the body with the furniture removed. It counts the server HTML rather than the rendered browser view, which is what a crawler reads first, so a page whose text arrives via JavaScript will report close to zero here and is worth investigating rather than dismissing.
Why every word counter gives you a different number
Run the same URL through four tools and you will get four answers, sometimes varying by a factor of three. None of them is broken. They are counting different things, and almost none of them say which.
| What is being counted | Typical result | When it is the number you want |
|---|---|---|
| Everything in the HTML | Highest, inflated by nav and footer | Almost never. It measures the template as much as the page |
| The browser selection, pasted into a word processor | High, and different on every viewport | A quick sanity check, nothing you would report |
| The rendered DOM after JavaScript | Matches what a reader sees | Auditing the reader experience, not the crawl |
| The server HTML before JavaScript | Lower on client-rendered sites, sometimes zero | Anything about indexing. This is what a crawler reads first |
| The declared content wrapper only | Lowest, and the most comparable across pages | Comparing pages on one site, or against a competitor |
This tool counts the last two and shows them side by side. The body copy figure comes from the content wrapper, the whole-page figure from the server HTML, and the difference is reported as page furniture with the share of the page it takes up.
The number that surprises people
A page built with a client-side framework can return almost no text in its HTML while looking full in a browser. The crawler queue that renders JavaScript runs behind the one that fetches HTML, and it fails often enough to matter. A near-zero count here is a finding, not a bug in the counter.
How long a page should be, by what the page is for
There is no correct word count, and Google has said so more than once. What exists is a rough sense of what each kind of page needs in order to do its job, which is a different claim and a more useful one.
| Page type | Usual range | What decides it |
|---|---|---|
| Product page | 200 to 500 words | Specification, use case, and the objections a buyer raises |
| Category or collection page | 150 to 400 words | Enough context to be more than a grid of links |
| News or announcement | 300 to 800 words | The facts, once, without padding to a target |
| Comparison or review | 1,200 to 2,500 words | One section per thing being compared, plus the verdict |
| Definitive guide | 1,500 to 3,000 words | Every question the query implies, answered on the page |
| Glossary or definition | 300 to 700 words | The definition first, then the distinctions people confuse |
| Landing page | 600 to 1,200 words | The claim, the evidence, and the objection handling |
Past about 3,000 words, a page is usually two pages wearing one URL. Splitting it into a hub and its parts tends to rank better than extending it further, because each part can then target the query it actually answers.
The only benchmark worth using
Count the pages currently ranking for your query and take the range rather than the average. One 6,000-word outlier drags an average somewhere no page needs to go. The range tells you what clearing the bar looks like.
What word count tells you about ranking, and what it does not
Word count is a symptom you can measure cheaply, which is why it gets treated as a cause. The correlation studies are real, and they are almost always read backwards.
- Longer pages do rank better on average, because covering a subject properly takes words. Length is the consequence, not the mechanism
- Adding words to a page that already answers its query makes the page worse, and the count goes up either way
- Under about 300 words of body copy, thin content becomes the likeliest explanation for a page Google crawls and declines to index
- A page can be 4,000 words and still thin, if all of it restates the same point
- Word count says nothing about whether the words are readable, or whether they demonstrate the experience and expertise a reviewer is looking for. Both are separate measurements
The practical use is as a filter rather than a score. Sort a site by body copy word count, look at the bottom of the list, and you have found the pages most likely to be sitting unindexed. That is a real workflow. Chasing a target number is not.
When a page is short and you want to know whether the words it does have are working, the readability checker scores the writing itself, and the keyword density checker shows which phrases the copy actually leans on.
Where the count becomes a real diagnosis
A page reporting almost no words while looking full in a browser is a rendering problem, not a writing problem. Confirm it against the page indexability checker, which reports the same server-side word count alongside the noindex, canonical and robots signals that decide whether the page can be indexed at all.
How to count the words on a page properly
Six steps, ordered so the ones that change the number come before the ones that interpret it. The tool runs the first three for you; the last three are judgement.
- 01
Count the HTML your server returns, not the browser view
Selecting a page in your browser and pasting it into a word processor counts what rendered after JavaScript ran. A crawler reads the server HTML first, and on a client-rendered page the two numbers can differ by thousands of words.
- 02
Separate the body copy from the page furniture
Navigation, headers, footers and cookie banners are words on the page that nobody came to read. A site-wide footer carrying 400 words adds 400 to every count and tells you nothing about any individual page.
- 03
Find the boundary the page itself declares
A page wrapped in main, marked role=main, or built around an article element is telling you where its content starts and stops. Use that boundary rather than guessing at one, and note which was used so the figure can be checked.
- 04
Read the count next to the page type, not against a target
A category page at 250 words and a guide at 250 words are different problems. Compare the number to what pages of that kind need rather than to a number pulled from a blog post about ideal content length.
- 05
Look at how the words are distributed
Paragraph count, longest paragraph and heading count say more about whether a page is readable than the total does. Two thousand words in nine paragraphs is a wall, and the total alone cannot see it.
- 06
Check the pages already ranking for the same query
The only length benchmark worth having is what currently satisfies the query. Count the top few results, take the range rather than the average, and treat it as a floor to clear rather than a target to hit.
Record the method alongside the number whenever you report it. A word count quoted without one starts an argument that neither side can settle.
What this tool reports
Body copy word count
The words inside the page's own content boundary, with the boundary named
Whole-page word count
Everything in the server HTML, including nav, header and footer
Page furniture share
How much of the page is template rather than content, as a percentage
Unique words
Distinct words, case-insensitive, which catches a page repeating itself
Reading time
At 238 words per minute, the median adult silent reading rate
Paragraph breakdown
Count plus the longest paragraph, which the total cannot see
Heading outline
Every heading in the counted content, with its level
Redirect resolution
Up to six hops, with the URL actually counted reported back
This counts one page, and thin content is rarely one page
Frequently asked questions
How do I count the words on a web page?
Paste the URL into a page word counter and it fetches the HTML, strips the markup, and counts the visible text. Doing it by hand means selecting the page in a browser and pasting it into a word processor, which counts the navigation, the footer and the cookie banner along with the article, and misses nothing about which of those the reader was there for. The number worth having separates body copy from page furniture, which is what this tool reports.
Why does every word counter give a different number?
Because they are counting different things. Some count everything in the HTML including nav and footer, some count only what a chosen content wrapper holds, some count the rendered page after JavaScript runs, and some count the source before it does. None of them is wrong, but a number without its method is not checkable. This tool states which boundary it used, so a figure you disagree with can be traced rather than argued with.
How many words should a page have?
There is no correct length, and Google has said repeatedly that word count is not a ranking factor. What matters is whether the page answers the query fully. In practice a product or category page works at 200 to 500 words, a news item at 300 to 800, and a page trying to be the definitive answer to a question usually needs 1,200 or more because that is what covering the subject takes. Length is the consequence of covering a topic, never the goal.
Is word count a Google ranking factor?
No. Google has stated this directly more than once, and the correlation studies people cite show longer pages ranking well because thorough pages tend to be longer, not because length itself is rewarded. Adding 500 words of padding to a 400-word page makes it worse. Adding 500 words that answer a question the page left open makes it better, and the word count going up is a side effect of the improvement rather than the cause of it.
What counts as thin content?
Thin content is a page that offers little value for the query it targets, and while there is no official threshold, pages under about 300 words of real body copy are the ones that most often show up as 'Crawled, currently not indexed' in Search Console. Word count is a symptom rather than the disease. A 200-word page that answers a specific question completely is fine; a 200-word page that restates its own title six times is not.
Why does the counter say my page has almost no words?
Because it reads the HTML your server returns, which is what a crawler sees on its first pass. If your content is rendered by JavaScript in the browser, that HTML is close to empty. Google can render JavaScript, but rendering is queued and can lag the initial crawl by days, and it fails often enough to matter. Content present in the initial HTML is indexed faster and more reliably, so a near-zero count here is worth acting on.
Does the count include the navigation and footer?
The body copy figure excludes them, and the whole-page figure includes them, which is why both are shown. The gap between the two is reported as page furniture, along with the share of the page it represents. When more words sit in the furniture than in the content, a crawler is reading the same ratio and has very little that is unique to this page to judge it on.
How does the tool know where the content starts and stops?
It looks for the boundary the page declares, in order: a main element, an element marked role="main", then the longest article element. If none of those exist it falls back to the whole body with nav, header, footer, aside and form removed. The method it used is printed with the result, because 'how many words is this page' has no single right answer and a number without its method cannot be checked.
What is a good reading time for a page?
Reading time is a presentation of word count rather than a separate measurement, so it inherits the same answer: it should be as long as the subject needs. The figure here uses 238 words per minute, the median silent reading rate for adult non-fiction from Brysbaert's 2019 meta-analysis of 190 studies, which is a better source than the 200 or 250 that circulates without one.
Should I match the word count of the pages ranking above me?
Use it as a floor, not a target. If the top five results all run past 2,000 words, that is evidence about what the query needs covered rather than about a length Google prefers. Take the range rather than the average, because one 6,000-word outlier drags an average somewhere no page needs to go. Then write what the subject requires and let the count land where it lands.
Does a longer page rank better?
Not by virtue of being longer. Longer pages often rank better because covering a subject properly takes words, and a page that covers it properly satisfies more of the queries that lead to it. The same length reached by padding does the opposite: it dilutes the topic, buries the answer, and increases the chance a reader leaves for a page that got to the point.
Can I count the words on a page that needs a login?
No. The tool fetches the URL as an anonymous visitor, so a page behind authentication returns the login screen and the count reflects that instead. The same applies to pages blocked by a firewall rule or geo-restriction. For a gated page, count the source HTML directly, or run the crawl from inside your own infrastructure where the session exists.
Does the counter follow redirects?
Yes, up to six hops, and it reports the URL it finally landed on when that differs from the one you entered. This matters because the page you counted is the one that gets indexed, and on a site with a trailing-slash or protocol redirect the address you typed is not the address being counted.
Other on-page SEO tools
Canonical Tag Checker
Validate canonical tags for self-referencing, protocol mismatches, trailing slash conflicts, noindex conflicts, and og:url consistency.
Use toolHreflang Validator
Validate hreflang tags for international SEO: self-referencing, return tags, x-default, language codes, and duplicate detection.
Use toolSchema & JSON-LD Validator
Validate structured data markup: required properties, date formats, nested entities, and JSON syntax. Supports all major schema types.
Use toolMeta Tag Checker
Check title tags, meta descriptions, Open Graph tags, Twitter cards, canonical URLs, robots directives, and viewport settings.
Use toolMeta Tags Generator
Write the title, description, and social share tags every page needs. Live character counts plus previews of how it will look in Google and on social.
Use toolKeyword Density Checker
See which words and phrases your copy actually leans on. Single word, two word and three word counts with density and first-mention position.
Use toolSERP Snippet Preview
See your title and meta description the way Google renders them, measured in pixels. Desktop and mobile, with the exact character your snippet is cut at.
Use toolURL Slug Generator
Turn a list of page titles into clean, readable URL slugs. Accents transliterated, filler words optional, and duplicates caught across the whole list.
Use toolAnchor Text Generator
Build branded, partial, long tail and naked URL anchor variations for any page, then audit the anchors you already have against a natural distribution.
Use toolReadability Checker
Score any text on six readability formulas at once, then see which sentences are dragging the grade up. Passive voice and long sentences broken out separately.
Use toolImage Alt Checker
Audit the alt text on every image on a page. Missing attributes kept apart from decorative alt="", plus image links with no name, filenames used as descriptions and duplicates.
Use toolOpen Graph Preview
See the share card Facebook, X, LinkedIn, Slack and Discord will each render from your meta tags, cut at the exact character each one cuts at.
Use toolRead up on on-page content
A tool tells you what is wrong. These explain what to do about it.
Cora SEO Review: What a $250/Month Correlation Tool Actually Tells You
Read guideSaaS SEO: The Complete Guide to Ranking Your Product in 2026
Read guideSEOquake Review: Free, Still Shipping, and Built to Upsell Semrush
Read guideTerms this tool checks
- Content Structure
Content structure refers to how information is organized on a web page using headings (H1-H6), paragraphs, lists, tables, and other formatting elements.
- E-E-A-T
E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trustworthiness.
- Keyword Cannibalization
Keyword cannibalization occurs when multiple pages on the same website target the same keyword or search intent, forcing them to compete against each other in search results.
Find every thin page on the site in one crawl
One URL at a time will not find the template that ships 180 words to a hundred pages. CrawlRaven crawls the whole site, reports body copy word count per page, joins it to your Search Console data, and ranks the thin pages by the traffic they already get.