Controlling The Snippet Google Shows

Meta descriptions are advisory. For a large share of queries Google discards them and lifts whatever body text matches the query, which is often a cookie notice, a breadcrumb line or the middle of a list. Rewriting the tag again does nothing, because the tag was never the source. The fix is to improve the passage Google is already choosing, or to put a stronger candidate near the top of the page.

The skill file

---
name: snippet-source-control
description: Use when Google ignores your meta description and lifts a stray sentence from the page into the result instead.
---

# Controlling The Snippet Google Shows

Meta descriptions are advisory. For a large share of queries Google discards them and lifts whatever body text matches the query, which is often a cookie notice, a breadcrumb line or the middle of a list. Rewriting the tag again does nothing, because the tag was never the source. The fix is to improve the passage Google is already choosing, or to put a stronger candidate near the top of the page.

## What you need first

- The affected queries with the description text as it appears in live results, not as written in the tag
- The rendered page in document order, including anything injected by script before the body copy
- The meta description as served, and the page position for each affected query

## Method

1. Search each affected query in a private window with the location set explicitly, and copy the displayed description verbatim. Trackers and preview tools show the tag, so the live result is the only reliable source.
2. Find that exact string in the rendered page. Where it came from is the diagnosis: a body paragraph means Google preferred your copy, boilerplate means it found nothing better to prefer.
3. Compare the text shown for two or three different queries against the same page. Different text per query is normal passage selection and needs a body fix, not a tag fix.
4. Where boilerplate was lifted, look at what sits between the H1 and the first substantive paragraph. Consent text, share prompts and author blocks in the markup ahead of the body are the usual sources.
5. Write two plain sentences in the first 200 words that answer the query directly and use its phrasing. This is the candidate you want selected, and it has to be visible body copy to be eligible.
6. Keep the meta description anyway, around 155 characters, and treat it as the version that survives on brand and navigational queries where the query gives Google nothing to match against.
7. Recheck after a recrawl on the same queries from the same location and device, since snippet selection varies by both.

## What this produces

A per-query table of lifted text, its source element, and the body edit made to replace it, usually a handful of rows per page.

## Where this goes wrong

- Rewriting the meta description repeatedly when the displayed text never came from it
- Reading the snippet from a rank tracker or a social preview tool, both of which echo the tag rather than the live result
- Judging the fix on one query when selection happens per query, so the page improves for one and stays wrong for the rest

---

From the QuQi skill library - https://www.quqi.io/skills/snippet-source-control

What you need first

• The affected queries with the description text as it appears in live results, not as written in the tag
• The rendered page in document order, including anything injected by script before the body copy
• The meta description as served, and the page position for each affected query

Method

1. 01 Search each affected query in a private window with the location set explicitly, and copy the displayed description verbatim. Trackers and preview tools show the tag, so the live result is the only reliable source.
2. 02 Find that exact string in the rendered page. Where it came from is the diagnosis: a body paragraph means Google preferred your copy, boilerplate means it found nothing better to prefer.
3. 03 Compare the text shown for two or three different queries against the same page. Different text per query is normal passage selection and needs a body fix, not a tag fix.
4. 04 Where boilerplate was lifted, look at what sits between the H1 and the first substantive paragraph. Consent text, share prompts and author blocks in the markup ahead of the body are the usual sources.
5. 05 Write two plain sentences in the first 200 words that answer the query directly and use its phrasing. This is the candidate you want selected, and it has to be visible body copy to be eligible.
6. 06 Keep the meta description anyway, around 155 characters, and treat it as the version that survives on brand and navigational queries where the query gives Google nothing to match against.
7. 07 Recheck after a recrawl on the same queries from the same location and device, since snippet selection varies by both.

What this produces

A per-query table of lifted text, its source element, and the body edit made to replace it, usually a handful of rows per page.

Where this goes wrong

• Rewriting the meta description repeatedly when the displayed text never came from it
• Reading the snippet from a rank tracker or a social preview tool, both of which echo the tag rather than the live result
• Judging the fix on one query when selection happens per query, so the page improves for one and stays wrong for the rest

Use this skill in your own AI

The download is a plain markdown file with the name and trigger in its frontmatter. Where an assistant supports skills it can load itself, that frontmatter is what it reads to decide this one applies.

Questions about this skill

More in On-page SEO