Perplexity cited pooyagolchian.com on 14 of 30 queries
Google's first page listed pooyagolchian.com on 7 of 30 queries. A Node.js tool counted the AI citations.
The code is private. The tool runs on SearchApi, whose Developer Ambassador Program I take part in.
Measured on 22 August 2026
- Cited by any AI engine
- 18 of 30 queries
- Cited by Google AI Mode
- 10 of 30 queries
- Cited by ChatGPT
- 9 of 20 queries it searched
The problem
Google Search Console reports clicks and impressions on Google's own results. It says nothing about citations inside ChatGPT or Perplexity answers. So I could not tell whether AI engines cited my site at all. I wanted a dated count I could re-run and compare.
Constraints
- The baseline run took 150 metered API calls, five for each query
- ChatGPT skipped the web search on 10 of the 30 queries
- Google showed an AI Overview on only 5 of the 30 queries
- Every figure on this page includes four queries that name me or my site
- Without those four, Perplexity cited my site on 10 of 26 queries
What I built
- I built a Node.js script that sends queries through SearchApi, a search data service. The 22 August 2026 baseline sent 30 queries to ChatGPT and Perplexity, and to Google's AI Mode, AI Overview and first page of results.
- Google's first page, searched from the US, is the benchmark. I treat a page that ranks there but earns no AI citation as a quoting problem, not a reputation problem.
- Only a clickable link to my site counts as a citation. The tool logs a mention of my name, or a link to my GitHub or LinkedIn, as a separate result.
- A query drops out of the score when no citation is on offer, as when ChatGPT answers without searching the web. An early version misread Google's empty AI Overview placeholder as a real overview and scored each one as a loss.
- Every run saves the raw engine responses, so a scoring fix re-applies to past runs at no API cost. Partial runs merge into the day's file. A failed call counts as an error, never a silent skip.
- ChatGPT decides per question whether to search at all, so one run cannot pin down its score.
Languages
- JavaScript
Stack
- Node.js (ES modules)
- SearchApi
- Vitest
More case studies
Open source
An open-source method for governed agents
SpecForge blocks Claude Code from editing your code until you approve the spec and plan.
Read case studyOpen source
15,246 businesses from 1,400 Google Maps requests
An open-source toolkit that crawls Google Maps into a business directory. No paid API call runs without an explicit yes.
Read case study
Engagement
Want work like this?
I scope every engagement fresh. Request a free call to test the fit.
Most Generative Engine Optimization engagements cost $14,000 to $25,000.
About AED 51,500 to 92,000.
Entity graph, agent-readable index and passage-level rewrites aimed at retrieval and citation. A fixed query set on live ChatGPT and Perplexity measures citation rate before and after, over 6 to 10 weeks. Optional citation monitoring afterwards costs $1,500 a month (about AED 5,500).