What is a meta description?
A meta description is a tag in the <head> section of a page's HTML that summarizes what the page is about in one or two sentences: <meta name="description" content="…">. Visitors do not see it on the page, but search engines may show it as the short description under the title in search results.
Google states that it does not use the meta description as a ranking signal. Its real job is to tell searchers what they will find on the page and make the decision to click easier. If a search engine decides that another part of the page answers the query better, it may show that text instead of your description. That is why the description should match the page content.
What makes a good meta description
An effective description states, in its first sentence, what the page offers for the thing the visitor is looking for. The principles below are the same rules the tool follows when it drafts options; you can also use them as a checklist when you write your own:
- Make it unique to the page; do not repeat the same description across many pages.
- Lead with the value proposition: what will the visitor get or be able to do on this page?
- Use the focus keyword once, naturally; avoid keyword stuffing.
- Do not promise prices, discounts, guarantees or figures that the page does not contain.
- Aim for roughly 140-155 characters; longer text may be cut off in search results.
Why length is an approximate measure
Google does not publish a fixed character limit for descriptions; it shortens long text based on screen width. A description with many narrow letters therefore takes up less space than one with the same number of wide letters. The cut-off point can also differ between mobile and desktop results.
The tool targets 140-155 characters, a range that usually gives enough information while reducing the risk of truncation. Options of 110-139 or 156-165 characters are marked “Warning”, and anything outside those ranges is marked “Issue”. You can check how an option will look in search results with the SERP preview tool.
How the tool works
When you enter a URL, the tool fetches the page once from our server and extracts the title, the current meta description, the headings and the first part of the visible text. This information is sent to Anthropic's Claude model, which writes options using five different angles. Our server, not the model, counts the characters; options outside the range are flagged, not hidden.
If the page content is loaded with JavaScript, the HTML returned by the server may contain little text. In that case the tool warns you, and you can continue by writing a short description in the topic tab. We do not store the URL, the text you enter or the results. The options are drafts: check that they match your page before publishing.