acr-tracker
ACR Tracker — Livostyle AI Citation Rate Monitor
Daily public measurement of AI Citation Rate (ACR) for Livostyle.com across Perplexity, using a standardized 50-query panel. Companion to our case study paper.
📊 Live dashboard: https://arturayupov.github.io/acr-tracker/
What this measures
Two metrics, daily:
acr_cited_rate— fraction of queries where at least one Livostyle canonical domain is in Perplexity’s citations.acr_brand_rate— fraction of queries where “Livostyle” / “Arcada LLC” appears in the answer text.
The panel is 50 fashion-vertical queries across 5 categories: direct brand, generic vertical, comparison, occasion/use-case, AI-tool-specific.
Files
| Path | What |
|---|---|
queries/panel.json |
50 standardized queries + canonical domains to track |
scripts/monitor.py |
Daily measurement script (uses Perplexity API) |
data/daily/{YYYY-MM-DD}.json |
Raw per-day snapshot with all 50 queries + citations |
data/timeseries.json |
Daily aggregate time-series (for dashboard) |
dashboard/index.html |
Static dashboard (chart.js + fetch from timeseries.json) |
.github/workflows/daily-monitor.yml |
Daily cron at 07:00 UTC |
How to reproduce
git clone https://github.com/arturayupov/acr-tracker
cd acr-tracker
pip install requests # no other deps
export PPLX_API_KEY=pplx-...
python scripts/monitor.py
Why this dashboard exists
For our DTC GEO case study we needed an ongoing measurement framework, not a one-shot. This dashboard is that framework. It’s also part of the strategy itself: AI engines preferentially cite sources that publish ongoing measurements, not just claims.
Ethics note
This tracker only queries the API legitimately and observes what Perplexity returns. It does not attempt to manipulate model behavior, push spam content, or violate any provider’s ToS. Public publication of measurements is the entire point.
License
MIT — see LICENSE.