All projects
CP
Marketing Intelligence2026

Crisis Pulse

Consumer signal dashboard for MENA markets, refreshed daily at zero infra cost.

Built for: WPP Media MENA · crisis reporting

Open the live app Source
loading demo…

Live: fetches real daily Wikipedia pageviews for the same signal topics the production dashboard uses, then normalises them the way the collector does. Production adds five more sources and per-market Google Trends.

The problem

During a disruption, brands need to know within days whether people are retreating into entertainment, comparison shopping, or doom-scrolling news. Paid research takes weeks. Free public data updates daily.

How the real system works

  1. 01A Python collector runs on GitHub Actions every morning and pulls Wikipedia pageviews, Google Trends RSS, GDELT article volume and tone, Google News RSS, ACLED conflict events and Twitch viewership.
  2. 02Every signal is normalised to a 0 to 100 interest index so unrelated sources can sit on one chart.
  3. 03The collector commits a static JSON file. Vercel redeploys. The dashboard is a pure frontend that reads that file, so there is no server and no running cost.
  4. 04Signals are grouped under hypotheses (stress drives escapism, financial stress changes purchase intent) so the chart answers a question instead of showing a number.

What you take away

How to turn free, noisy public data into a comparable index, and why normalisation matters more than the raw count.