SERP Keyword Clustering — group keywords by shared top URLs

SEO teams and content planners: turn a big keyword list into topic groups by checking which keywords Google answers with the same URLs, so you write one page per group instead of one per keyword. It reads datasets you already have from apify/google-search-scraper; no Google access, no proxies, no LLM. $2.00 per 1,000 keywords. Reading datasets is free.

Use cases

• Plan one page per topic instead of one page per keyword
• Find keyword cannibalization: keywords that share results but point to different pages
• Group a keyword research export into topics with an intent label (informational, commercial, navigational)

How it works

1. Run Google Search Scraper (apify/google-search-scraper) on your keywords.
2. Run this Actor with that run's dataset ID.
3. Read one row per keyword with its cluster, parent keyword and estimated intent. A full cluster list is saved in the key-value store as CLUSTERS.

For each keyword the top N organic URLs (default 10) are compared. Two keywords with at least minSharedUrls (default 3) URLs in common belong together. URLs are compared without www, query string, fragment or trailing slash.

• strict (default): the most connected keyword becomes the centre of a group, and only keywords that overlap with it join. Tight, focused groups.
• loose: connected components. Any chain of overlaps merges into one group, so groups are larger and can drift across topics.
• parentKeyword: the keyword in the group with the highest total overlap with the others.
• Intent (informational, commercial, transactional, navigational, local) is an estimate from rules: words in the keyword (buy, price, best, vs, how, near me, login, ...) plus URL patterns in the top results (/product, /shop, /blog, wikipedia, youtube, ...). intentConfidence is low, medium or high. Treat it as a hint, not a fact.
• Missing fields in the source data are tolerated. By default each country and device combination is clustered separately.

Input

{
  "datasetIds": ["<dataset ID of your Google Search scraper run>"],
  "topN": 10,
  "minSharedUrls": 3,
  "clusterMethod": "strict",
  "groupByCountry": true
}

Leave datasetIds empty to see a free demo on sample data (rows carry _demo: true).

Output

One row per keyword:

{"keyword": "buy running shoes", "clusterId": 2, "parentKeyword": "running shoes price", "clusterSize": 3,
 "sharedUrlsWithParent": 4, "intent": "transactional", "intentConfidence": "high", "countryCode": "US",
 "device": "DESKTOP", "topUrls": ["https://a.com/shop", "https://b.com/product/1", "https://c.com/store"]}

sharedUrlsWithParent is empty (null) for the parent itself. In strict mode a keyword is joined through the group's centre, which can differ slightly from the parent, so its overlap with the parent may be below your threshold.

Key-value store CLUSTERS:

[{"clusterId": 2, "parentKeyword": "running shoes price", "keywords": ["running shoes price", "buy running shoes"],
  "size": 2, "commonUrls": ["https://a.com/shop"], "countryCode": "US", "device": "DESKTOP"}]

SUMMARY holds counts (keywords, clusters, singletons) and the settings used.

Pricing

$2.00 per 1,000 keywords written (event keyword, $0.002 each). The demo is free.

Notes

This Actor only processes datasets you already own; it does not access Google. You are responsible for how the underlying data was collected and used.

More tools from the same developer

• SERP Analyzer — rankings, share of voice and People Also Ask from Google Search scraper datasets
• Sitemap URL Extractor — list every URL of a website from its sitemaps
• Dataset Merge & Dedup — merge Apify datasets and remove duplicates by the fields you choose
• Google Sheets Export — export an Apify dataset to Google Sheets