Important non‑news landing pages often shape a visitor’s first impression, yet they rarely get the same attention as high‑visibility content. At Princeton University, more than 100 strategic pages needed regular evaluation, but limited staffing made traditional audits impossible.
To address this, I built an automated content auditing system that:
- Processes Drupal node JSON and GA4 metrics through Python‑based analysis
- Uses an LLM to interpret page purpose, message alignment, risks and recommended actions
- Generates a complete .docx audit report in minutes
The result is a repeatable audit pipeline that editors can actually use.
This session walks through the end‑to‑end workflow: data ingestion, metrics processing, Drupal content extraction, prompt design, report generation, limitations and the realities of using AI for editorial decisions.
