AI Overview & SERP Visibility Monitor

What you get

Does Google's AI recommend your brand? This Actor answers that question with evidence instead of guesswork. For every keyword you track, it checks whether Google rendered an AI Overview, whether that AI Overview cited your site, who owns the featured snippet, where you rank organically, and which competitors are being surfaced instead of you.

One dataset row per keyword. Built for weekly monitoring by SEO agencies, in-house marketing teams, and anyone whose organic traffic is being absorbed by AI-generated answers.

Why this exists

Google's AI Overview answers the question at the top of the page. The user reads it and leaves. Your rank-3 blue link never gets the click. Traditional rank trackers still report "position 3" and call that a good week.

Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) are the response to that shift, and both need the same missing input: a measurement of whether you are being cited in the answer surfaces themselves. That is what this Actor produces.

What it measures

For each keyword:

Field	Meaning
keyword	The query that was checked
country	Country code the SERP was localized to
aiOverviewPresent	Whether Google rendered an AI Overview at all
brandCitedInAiOverview	Whether the AI Overview cited your brand — null when no AI Overview existed
aiOverviewSources[]	Every source the AI Overview cited (url, domain, title)
organicRank	Your 1-based organic position — null when you are genuinely absent
featuredSnippetOwner	Domain that owns the featured snippet, if any
brandOwnsFeaturedSnippet	Whether that owner is you
peopleAlsoAsk[]	The People Also Ask questions on the page
competitorsCited[]	Per competitor: AI Overview citation and organic rank
checkedAt	UTC timestamp of the check
status	checked or failed

Diagnostic fields are included too: aiOverviewDetectedBy, brandCitedMatchedBy, organicRankMatchedBy, matchStrategy, organicResultsParsed, and error.

The honesty rule this Actor is built around

Google does not render an AI Overview for every query. Many monitoring tools quietly report that absence as "your brand was not cited", which is a different and much more alarming claim.

This Actor keeps the two facts separate:

• No AI Overview was rendered → aiOverviewPresent: false, brandCitedInAiOverview: null
• An AI Overview was rendered and did not cite you → aiOverviewPresent: true, brandCitedInAiOverview: false
• An AI Overview was rendered and cited you → aiOverviewPresent: true, brandCitedInAiOverview: true
• The results page could not be retrieved → status: "failed", every measured field null, and no charge

The same discipline applies to organicRank: absent is null, never 0 and never a 999 sentinel that sorts like a real rank.

How brand matching works

Supply domain whenever you can. Domain matching is the reliable signal, and it matches subdomains too (help.notion.so counts as notion.so). Brand-name matching is the fallback and uses word-boundary matching, so Notion does not match Notionally Ltd.

Every row tells you which method produced the answer via brandCitedMatchedBy and organicRankMatchedBy, so you can audit any result you find surprising.

No LLM APIs

This Actor measures public Google SERP surfaces only. It does not call OpenAI, Perplexity, Anthropic, or any other model provider, and it does not resell output from an API it does not own. What it reports is what Google actually served on a public results page — verifiable by hand, and stable as a compliance story.

Input

{
  "brand": "Notion",
  "domain": "notion.so",
  "keywords": [
    "best note taking app",
    "notion alternatives",
    "team wiki software"
  ],
  "country": "us",
  "language": "en",
  "competitors": ["obsidian.md", "Evernote"]
}

Input	Required	Default	Notes
brand	yes	—	Brand name to track
domain	no	—	Your domain. Strongly recommended for accuracy
keywords	yes	—	One row is produced per keyword
country	no	us	Two-letter country code
language	no	en	Two-letter language code
competitors	no	[]	Names or domains; a dotted entry with no space is treated as a domain
proxyConfiguration	no	preconfigured	Leave at its default for reliable access to Google Search results

Output

{
  "keyword": "best note taking app",
  "country": "us",
  "brand": "Notion",
  "domain": "notion.so",
  "status": "checked",
  "aiOverviewPresent": true,
  "aiOverviewDetectedBy": "attr:data-subtree=aio",
  "brandCitedInAiOverview": true,
  "brandCitedMatchedBy": "domain",
  "aiOverviewSources": [
    { "url": "https://obsidian.md/", "domain": "obsidian.md", "title": "Obsidian" },
    { "url": "https://notion.so/product", "domain": "notion.so", "title": "Notion" }
  ],
  "organicRank": 2,
  "organicRankMatchedBy": "domain",
  "featuredSnippetOwner": "obsidian.md",
  "brandOwnsFeaturedSnippet": false,
  "peopleAlsoAsk": ["What is the best free note taking app?"],
  "competitorsCited": [
    { "competitor": "obsidian.md", "citedInAiOverview": true, "organicRank": 1 }
  ],
  "matchStrategy": "domain",
  "checkedAt": "2026-08-03T12:00:00+00:00"
}

Use cases

• Weekly GEO/AEO reporting for retainer clients — show whether client brands are being cited in AI Overviews, not just where they rank.
• AI Overview coverage mapping — find which of your keywords now trigger an AI Overview at all. That subset is where classic rank tracking has stopped predicting traffic.
• Competitive citation share — see which competitors Google's answer surfaces prefer, and on which queries.
• Content prioritization — pages that rank well but are never cited are candidates for restructuring toward extractable, answer-shaped content.
• Featured snippet defense — track snippet ownership changes over time.
• Post-migration verification — confirm a domain migration did not drop you out of the answer surfaces.

Notes on reliability

Google does not publish a stable format for the AI Overview block, and it changes the page layout from time to time. Detection checks several independent signals rather than one, and every row records which signal fired in aiOverviewDetectedBy, so any drift is visible in your data rather than silently reported as "no AI Overview".

If a run approaches its time limit, the Actor stops cleanly with the keywords it has already finished rather than being cut off mid-run, and one keyword failing never fails the whole run.

Limitations

• Google does not render an AI Overview for every keyword; when it doesn't, brandCitedInAiOverview is honestly null rather than a false negative.
• Brand-name matching (used when no domain is supplied) can occasionally mismatch similarly named brands; supplying domain avoids this.
• Reflects Google's public results page at the moment of the check — AI Overviews and rankings can change between runs.
• Covers Google Search only; it does not query any AI chat assistants directly.

Related Actors

• Google Search Results Scraper — Thirdwatch
• Google Trends Scraper - Interest, Slope & Related Queries — Thirdwatch
• Bulk SEO Audit - Titles, Meta, Headings, Schema, Links — Thirdwatch

Last verified: 2026-09

More scrapers at thirdwatch.dev.