GEO Tracker

Automated tracking of Development Seed's prominence across LLM-generated responses. Queries multiple AI platforms daily, analyzes responses for mentions and citations, and stores results as CSV for easy analysis.

Overview

This tool measures how prominently Development Seed and its products appear in AI-generated answers by:

1. Querying LLMs (Perplexity, ChatGPT, Gemini, Claude) with geospatial search terms
2. Analyzing responses for mentions, citations, and recommendation language
3. Scoring prominence (0-100) based on multiple factors
4. Appending results to data/results.csv — committed back to the repo automatically

Tracked queries include:

• Development Seed products (VEDA Dashboard, titiler)
• Geospatial technologies (STAC, Cloud-Optimized GeoTIFF)
• Industry trends (Satellite Imagery, Climate Data)

Data Output

Results are stored in data/results.csv with these columns:

Column	Description
date	Run date (YYYY-MM-DD)
source	LLM source (Perplexity, ChatGPT, Gemini, Claude)
query_name	Human-readable query name
query_id	Machine-readable query ID
category	Query category (product, technology, trend, organization)
search_term	The exact natural-language prompt sent to the LLM
prominence_score	0-100 score based on mention position, citations, recommendations
mentioned	Whether Development Seed was mentioned
recommended	Whether DS was recommended/endorsed
position	Position of first mention (early/middle/late/none)
citation_count	Number of DS page citations found
data_source	Source type: web (search-grounded) or training (knowledge only)
ds_pages	Pipe-separated DS page URLs cited
tokens	Total tokens used for this query

The CSV can be opened directly in Google Sheets, Excel, or loaded into any analysis tool.

Setup

Prerequisites

• Node.js 20 LTS
• One or more LLM API keys (see below)

Local Development

1. Clone the repository:

   git clone https://github.com/aboydnw/ds-geo-tracker.git
   cd ds-geo-tracker

2. Install dependencies:

   npm install

3. Create environment file:

   cp .env.example .env
   # Edit .env and add your API keys

4. Run the tracker:

   node index.js

   Results are written to data/results.csv.

GitHub Actions (Automated Daily Runs)

The tracker runs automatically every day at 9:00 AM UTC via GitHub Actions. After each run, the updated CSV is committed back to the repository.

Required Secrets:

Configure in your repository: Settings > Secrets and variables > Actions > New repository secret

Secret	Description	Required?
PERPLEXITY_API_KEY	Perplexity Sonar API key	At least one
OPENAI_API_KEY	OpenAI API key (ChatGPT)	At least one
GOOGLE_AI_API_KEY	Google AI API key (Gemini)	At least one
ANTHROPIC_API_KEY	Anthropic API key (Claude)	At least one

At least one API key must be configured for the tracker to produce results. Sources without keys are gracefully skipped.

Manual Trigger:

1. Go to the Actions tab in your repository
2. Select "GEO Tracker" workflow
3. Click "Run workflow"

Configuration

Adding/Modifying Queries

Edit src/queries.js to add or modify tracked queries:

{
  id: 'my-query',
  name: 'My Query Name',
  searchTerms: ['search term 1', 'search term 2'],
  category: 'product', // product | technology | trend | organization
}

Schedule

The default schedule is daily at 9:00 AM UTC. To change it, edit the cron expression in .github/workflows/geo-tracker.yml:

schedule:
  - cron: '0 9 * * *'  # Daily at 9 AM UTC

Use crontab.guru to build cron expressions.

Project Structure

geo-tracker/
├── .github/workflows/
│   └── geo-tracker.yml       # GitHub Actions workflow (daily run + CSV commit)
├── data/
│   └── results.csv           # Tracking results (auto-updated by CI)
├── src/
│   ├── analysis.js           # Response analysis & prominence scoring
│   ├── csv-store.js          # CSV storage module
│   ├── orchestrator.js       # Core tracking loop
│   ├── queries.js            # Query configuration
│   └── sources/
│       ├── index.js          # Source registry
│       ├── perplexity.js     # Perplexity Sonar client
│       ├── chatgpt.js        # OpenAI ChatGPT client
│       ├── gemini.js         # Google Gemini client
│       └── claude.js         # Anthropic Claude client
├── index.js                  # Main entry point
├── package.json
├── .env.example              # Environment template
└── .nvmrc                    # Node.js version (20)

License

MIT