Google search snippets: how they're written, how to control them, and the AI link

What a snippet is, and where Google gets the words

Look at any normal Google result and you'll see a title link, the page address and a few lines of text below. Those lines are the snippet. Google's snippet documentation says it primarily uses the content on the page to work out the right snippet, and sometimes uses the meta description if that gives a more accurate summary. The same page can show different snippets for different searches, because Google picks the text that fits each query most closely.

That surprises a lot of owners. You can write a careful meta description and still see Google quote a sentence from halfway down the page. You don't get to choose the words. What you do control is the material Google picks from: clear opening lines, headings that say what each section covers, and a meta description that sums up the page in plain words. Google says there's no limit on how long a meta description can be, and that identical or similar descriptions on every page of a site aren't helpful.

The word snippet also gets used for several different features, and they behave differently. A featured snippet flips the usual layout so the text sits above the link. Google's help says its systems decide when a page would make a good featured snippet, and that you can't mark a page as one. Rich results add details such as ratings or event dates, and they depend on structured data. A supporting link in an AI Overview is different again: the answer is written by Google's AI, and your page appears as a link beside it.

The table sorts out who writes what. The important column is the last one, because it shows that the same few settings reach every row.

How Google picks the text, and how to write descriptions it can use

Google's snippet documentation names two sources for the text. The main one is the visible content of the page. The second is the meta description, which Google uses when it might describe the page more accurately than a passage would. The documentation doesn't list structured data as a source for the text snippet. Because the choice is made search by search, a broad query about your business might pull the meta description, while a narrow question pulls the one sentence on the page that answers it.

That gives you two jobs. The first is a meta description that describes the whole page in plain words, so Google has an accurate summary to fall back on. The second is a page whose sections each open with a sentence that makes sense on its own. If the first line under a heading is a scene-setter about your history, that's what Google has to work with when it looks for a passage. If it states what the section is about, Google has something it can lift without the searcher losing the meaning.

Google's guidance on meta descriptions is short and practical. These are the points it makes, with the local service version of each.

• Give every page its own description. Google says identical or similar descriptions on every page of a site aren't helpful. A template sentence repeated across thirty service pages tells Google nothing about any of them.
• Use a site-level description on the home page and other aggregation pages, such as a services index, and page-level descriptions everywhere else.
• Include the facts a searcher would want. Google's example for news and blog posts is the author, the date and the byline. For a service page the equivalent is the service, the area and what the page covers.
• Generate descriptions from page data if you have a large site. Google says programmatic generation can be appropriate and is encouraged, as long as each one is truly descriptive and human-readable.
• Don't write a string of keywords. Google says descriptions made of long strings of keywords don't give users a clear idea of the page's content.
• Don't chase a character count. Google says there's no limit on how long a meta description can be. Write as much as the page needs to be described, and put the most useful words first.

Featured snippets, rich snippets and image previews

A featured snippet is the box where Google reverses the normal layout and shows the descriptive text first, with the link underneath. Google's documentation says its systems decide whether a page would make a good featured snippet for a search, and that there's no way to mark a page as one. Clicking a featured snippet takes the searcher directly to the section of the page that appeared in it. Google doesn't publish a recipe for winning one. In practice, the pages that get picked tend to have a heading that matches a common question and a sentence straight after it that answers it.

Opting out is less tidy than most owners expect. Google says nosnippet removes every snippet from the page, featured and regular. If you only want the featured snippet gone, it suggests experimenting with lower max-snippet values, but says a low max-snippet setting doesn't promise Google will stop showing a featured snippet for your page. Text marked with data-nosnippet is kept out of both featured and regular snippets, and if a page carries both nosnippet and data-nosnippet, nosnippet takes priority. Before opting out, weigh what you'd lose: your normal snippet goes with nosnippet, and so does eligibility as a direct input for AI Overviews and AI Mode.

Rich snippets are a different thing. Many SEO guides use that name, while Google's documentation calls them rich results: listings with extra details such as review stars, event dates or recipe times. They come from structured data, and Google is clear that correct markup doesn't promise anything. Its structured data guidelines say Google doesn't guarantee your structured data will show up in search results. They also say not to mark up content that isn't visible to readers, and that the structured data must be a true representation of the page content. A page that breaks those rules can get a manual action, which removes its eligibility for rich results without changing its ranking in web search.

Image previews have their own setting. max-image-preview takes three values: none, standard and large. Google says large allows a preview up to the width of the viewport, and that if you don't set the rule, Google may show an image preview of the default size. Even with nosnippet, Google says a static image thumbnail may still appear if it gives a better experience. If your service pages carry strong job photos, leaving image previews at large is usually the sensible choice.

To set up rich results without guessing, work through the steps in order.

1. Pick a structured data type that matches something a visitor can see on the page, such as an FAQ block or a business's opening hours.
2. Add the markup so every value matches the visible text exactly.
3. Test the page with the Rich Results Test before publishing, and fix errors first.
4. After the page is live, check it with the URL Inspection tool in Search Console.
5. Wait for Google to recrawl, and accept that a valid page may still show as a plain result.

What each snippet control costs in the normal listing

Every snippet control removes something from your result, and that something is usually the reason a searcher clicks. A result with a title and no description gives a person less to go on than the listing beside it. So before adding any rule, ask what you're trying to hold back, and pick the narrowest control that does it.

Google's robots meta tag documentation is specific about scope. It says nosnippet applies to all forms of search results at Google: web search, Google Images, Discover, AI Overviews and AI Mode. max-snippet sets a character limit on the text snippet, where 0 means no snippet and -1 lets Google choose the length. Google adds that the limit also caps how much content may be used as a direct input for AI Overviews and AI Mode, unless the publisher supplies content as in-page structured data or has a licence agreement with Google. A short limit chosen to keep results tidy can therefore cost more than it saves.

The table sets out each control, what it does and what it costs. Product screens change, so check Google's documentation before acting; this reflects it at the time of writing (October 2026).

Header-level and template rules: where snippet settings hide

Snippet rules don't only live in the page source. Google's documentation says any rule that works in a robots meta tag can also be sent as an X-Robots-Tag in the HTTP header. That's the only way to apply rules to files without an HTML head, such as PDFs, and Google's own example is a server rule that adds a header to every PDF on a site. It's also why header rules get missed: they never appear in the source a browser shows you.

Rules can also be aimed at one crawler. A meta tag named robots applies to all crawlers, while one named googlebot applies to Google's text results and googlebot-news to news results. A header can name a crawler the same way. When several tags apply, Google says it uses the sum of the negative rules, and when rules conflict, the more restrictive one wins. A page carrying both max-snippet:50 and nosnippet ends up with nosnippet. And if robots.txt blocks a page from crawling, Google can't see any of these rules at all.

data-nosnippet has a few traps of its own. Google says it's a boolean attribute, so any value you give it is ignored. It only works on span, div and section elements, inside valid HTML. Google also says not to add or remove the attribute on existing page elements with JavaScript.

The bigger risk is scale. Themes and SEO plugins usually apply robots settings by page type, such as all posts, all tag archives or all attachments, so one checkbox can reach hundreds of URLs at once. Checking one page proves little. Check one page of every type, using these steps.

1. List the page types on your site: home page, service pages, suburb pages, guides, archives, PDFs and any landing pages for ads.
2. Pick one live URL of each type.
3. Open URL Inspection in Search Console, run a live test, and open View tested page to read the HTML and the HTTP response headers.
4. Search each one for robots, googlebot, nosnippet, max-snippet, max-image-preview and data-nosnippet, and note what you find.
5. Compare what you found with the theme, plugin and server settings for that page type, so you know where each rule comes from.
6. Fix the rule at its source, then run a live test on one URL of each type to confirm the change.

Worked example: the snippets that said nothing

The following business is made up for illustration. The suburb is real; the site, the settings and the decisions belong to the invented business.

A Buderim electrician launches a new website built with a popular SEO plugin. A month later the owner searches for their switchboard service and finds the result looks thin. The description reads like a slogan, and it's the same slogan that appears under their lighting page and their smoke alarm page. When they search a more specific question about switchboards, Google quotes a line about when the business started, cut off halfway through.

The first finding is the meta descriptions. The plugin filled every service page with one template sentence, so Google either ignored it or showed the same words everywhere. The owner writes a new description for each service page, naming the service, the area and what the page covers, along the lines of the table earlier in this guide.

The second finding is in the headers. A live test in URL Inspection shows the plugin sending a short max-snippet value on every page, set during the build to stop long descriptions. That explains the cut-off text. It also means every page on the site has a cap on how much of its content can be used as a direct input for AI Overviews and AI Mode. The developer removes the rule at the plugin level, and the owner runs a live test on one page of each type to confirm it's gone.

The third finding is the page itself. The switchboard page opens with two paragraphs of company history, and the useful part starts much further down. The owner moves a plain answer to the top of each section, so a passage Google lifts makes sense on its own. While they're editing, they wrap the staff mobile numbers on the contact page in data-nosnippet, because they'd rather those weren't quoted in results.

Then they wait for Google to recrawl, without promising anyone a featured snippet or an AI link. The settings no longer get in the way, and the page now gives Google better material. Whether it's chosen still depends on whether it's the most useful answer on offer. For the next round of work on the page, our guide on how to rank on Google covers what moves rankings once the technical basics pass.

Where snippet settings meet AI eligibility

Snippets and Google's AI features share a gate. Google's guide to optimising for its generative AI features says that to be eligible, a page must be indexed and eligible to be shown in Google Search with a snippet, and must meet the Search technical requirements. The site must also be included in the Search generative AI setting in Search Console, which is on by default. Google's AI features page says there are no additional technical requirements beyond that.

So the snippet decisions in this guide are AI decisions as well. nosnippet stops the page's content being used as a direct input for AI Overviews and AI Mode. A max-snippet limit caps how much can be used. data-nosnippet appears in Google's list of ways to limit what its AI features show from your pages. A rule added to tidy a blue-link result can quietly take a page out of AI answers, which is why it pays to treat every snippet control as a decision about both.

Passing the gate makes a page eligible, not chosen. Google says a page meeting all the requirements still might not be crawled, indexed or served, and that its AI features rely on its core Search ranking and quality systems to pick the links. If you want to stay in normal results but out of AI answers, there's a separate Search Console setting for that, and it leaves your snippets alone. Our guide to how Google AI Overviews work covers that setting, the full page-by-page eligibility check and how to measure AI visibility, so this guide doesn't repeat them.

If your snippets look thin or your key pages never appear in AI answers, our AI SEO work starts with these settings and the content behind them, and the technical SEO side covers headers, templates and indexing. You can request a strategy review to have your service pages checked against the points in this guide.