Editorial Intelligence & Sentiment Analysis Framework

The challenge

Editorial conversations at Alhurra centered on what to cover next, with little grounded insight into what had already worked and why. The team lacked a data-backed way to answer the basic questions that should shape coverage: why did this story resonate and another didn’t, what were people actually saying about it, and how should that inform today’s coverage or a follow-up to yesterday’s story.

The approach

I built a daily analysis process that pulled performance data from Sprout Social and Emplifi, alongside scraped comments and interactions, across a mix of stories covered in the prior 24 hours, and used AI to process that data into an interpretive layer I then reviewed and delivered at the editorial meeting. The deliberate choice was to treat controversial-topic engagement as an expected outlier rather than a signal to chase. Without that guardrail, the data would consistently reward controversy and mislead editorial into thinking provocation was the strategy that worked, when it was simply the format that always spikes.

The execution

1. 01Pulled engagement and audience data daily across platforms (YouTube, Facebook, Instagram, X) from Sprout Social and Emplifi, combined with scraped comment and interaction data, across a mix of stories from the prior 24 hours
2. 02Used AI to analyze that data into a structured, per-story breakdown: engagement pattern, viewer or reader retention, traffic sources, likely drivers of performance, and comment sentiment
3. 03Wrote the interpretive layer myself based on that AI analysis, translating raw sentiment and performance signals into editorial takeaways the team could actually act on
4. 04Delivered the analysis daily at the editorial meeting, structured around what worked, what didn’t, and how that should shape today’s coverage or a follow-up on a story from the day before