Case study · Platform

Squawk

Published

The problem

Buyers now ask ChatGPT, Perplexity, Gemini and Grok as well as Google. Nobody could say how often a brand gets named there, why it doesn't, or whether a fix moved anything beyond the noise.

What we built

A platform that asks the AI engines a brand's real buyer questions and scores how often it gets named, alongside its Google ranking and how readable the site is to AI. Failed checks become fixes, and every change is measured against a noise band and a control.

Results

53
readiness checks
7
AI engines queried
465
automated tests

Counted from Squawk's own code: its readiness check catalogue, its engine adapters and its test suite.

How it’s built

05 · PROVEShows whether the change movedAgainst a noise band and a control04 · FIXTurns each failed check into a fixEvery check carries its fix, written out03 · AUDITFlags what keeps AI from reading youBlocked AI crawlers, no Wikidata entry, no llms.txt02 · SCOREScores how often you're namedSearch, readiness and mentions in one number01 · ASKAsks the AI engines your buyers' questionsChatGPT, Gemini, Perplexity, Grok and AI Overviews

Swipe sideways to see all of it.

Tech stack

Built with

  • Next.js
  • React
  • TypeScript
  • PostgreSQL
  • Drizzle ORM
  • Azure
  • Docker

Connected to

  • ChatGPT
  • Gemini
  • Perplexity
  • Google AI Overviews
  • Copilot
  • Claude
  • Grok
  • DataForSEO
  • Google Search Console
  • Google Analytics
Next case studySwitchyardSoftware / Engineering Teams