Keyword Research Dataset: 366 Queries With Their SERPs

Keyword research dataset, built 2026-11-19: 366 queries with the top 7 results each, ads, AI Overview, People Also Ask and autocomplete. 1,219 different domains. Cost: $1.9 of API calls.

Ran
Sample
366
Tools
None, built to order
Method
Google results and autocomplete for 366 queries, one geo (US), one day, through a SERP API and our fetch layer. We labelled each row by hand and by script.

What we ran

We needed to know what to build on this site, so we ordered ourselves a dataset. It is the kind of feed we sell: a list of targets in, a clean table out.

  • Input: 366 queries across 8 topics (fake shops, ad quality, merchants, bot traffic, regulators, social, data feeds, brand).
  • Per query: the top 7 Google results (domain, page type), whether ads and an AI Overview showed, the People Also Ask questions, the phrase’s position in autocomplete, and whether a free tool or page already answers it.
  • Where: Google, US, one day (2026-11-19 and 20), through a SERP API and our fetch layer. Raw rows in the research folder of this project.

What we found

queries 366
different domains in the top 7 1,219
queries with People Also Ask 284
queries where the phrase is in autocomplete 106
queries with an AI Overview 92 (25%)
queries with ads 43 (12%)
queries rated priority 1 43
page type we planned queries
landing page 174
free tool 80
guide 53
stats page 23
glossary 20
comparison 16

Reddit is in the top 7 for 205 of 366 queries (56%). YouTube 52, Facebook 48, Google’s own help pages 39. Cost: about $0.81 for 380 SERPs and $1.1 for 1,100 autocomplete calls.

The table drove the site: the first free tools and pages are the rows that scored priority 1.

How you’d use it

  • Marketing team: the same pull for your market, with a gap column against your own pages. Refreshed monthly, with a diff of what moved.
  • Agency: one dataset per client, in your template, delivered as CSV or into your sheet.
  • Data team: any list of targets in, a clean table out. That is a custom feed: you name the data, we build it, run it and deliver it.
Run this on your own list

Send us the sites, domains or ads. We run it, and send back the table.

Get this run

Questions

What is in a keyword research dataset?

One row per query: who ranks in the top 7, what kind of page each is, whether ads or an AI Overview show, the People Also Ask questions, where the phrase sits in autocomplete, and our gap and priority call.

How much does a dataset like this cost to build?

This one cost about $1.9 in API calls: roughly 380 SERPs for $0.81 and 1,100 autocomplete calls for $1.1. The labelling and the reading was the work.

Does it show search volume?

No. Google does not give volume. We use autocomplete position, SERP type, ads and who ranks as demand signals. Volume tools estimate.