Write meta descriptions from the actual page

Summarize the value a reader will receive

A meta description should help someone understand what is on the page before opening it. Begin with the specific task or answer, then add one useful detail that distinguishes the article. Avoid strings of keywords or generic claims about being the best resource. Write the description after the article is stable so it reflects the finished page. If the article changes from an overview into a tutorial, update its summary too. Treat the description as part of the editorial package rather than a field filled automatically at the last minute.

Understand the limits of control

Google's snippet guidance explains that snippets are generated and may use page content or a meta description. Its title-link documentation describes a similarly automated title process. Supplying good metadata is worthwhile, but it does not guarantee identical wording in every search result. Avoid promising clients that a particular sentence will always appear. Instead, improve both the description and the article's opening so either can accurately communicate the page. This also helps human readers who arrive directly and decide whether the article is relevant before scrolling.

Build descriptions from concrete details

Choose details that the reader can verify after clicking: a checklist, an example calculation, a troubleshooting flow, or a comparison of specific options. Do not include unsupported outcomes such as guaranteed rankings. Keep the summary readable without repeating the title word for word. Preview it at different widths, but do not remove essential context merely to meet an arbitrary character count. For a large library, generate drafts from structured article fields and review exceptions. Page-specific summaries are more useful than replacing one keyword inside the same sentence across every article.

Hypothetical example: a source-checking guide

Imagine an article explains how to verify product claims in AI drafts. A weak description says Learn everything about AI content and SEO with our amazing guide. A stronger version says Check AI article claims using primary sources, a simple evidence ledger, and a worked product-feature example. The second summary gives a reader a reason to choose the page and accurately describes its contents. If the worked example is later removed, the description must change. Metadata should not preserve a promise the body no longer fulfills.

Include metadata in publication review

Check that the published HTML contains one intended description and that it is not accidentally copied from another page. Review escaped punctuation, missing variables, and language mismatches after the build. If your CMS has separate fields for social descriptions, decide whether they should share text or serve a different purpose. Store metadata with the article revision so changes can be audited. When reviewing performance, consider query mix and page changes alongside snippet edits. The objective is a useful and truthful introduction, not a claim that a description alone determines traffic.

• Mention the concrete resource, method, or example the article actually contains.
• Check for copied descriptions and unresolved template variables in published HTML.
• Update the summary when a substantial rewrite changes the article's purpose.