What I've Learned from
60,000+ Headlines
This is not a theoretical analysis. Over 10 years, working closely with hundreds of newsrooms, I have analyzed over 60,000 AI-generated headlines and tracked over 130,000 real published headlines. The patterns are clear.
Key Findings
The patterns from analyzing tens of thousands of headlines reveal actionable insights for journalists.
AI vs Real Headlines
Large language models consistently fall into familiar patterns — they repeat the same words, structures, and emotional appeals. Real journalists, by contrast, vary their vocabulary, sentence structure, and approach based on the specific story, audience, and newsroom culture.
I analyzed 60,000+ AI-generated headlines and found they tend to be structurally sound but culturally bland. Real headlines have edge — written by people who understand their audience deeply.
33,000+ Unique Words
Across 60,000+ headlines analyzed, I identified 33,000+ unique words. This diversity is something AI systems struggle to replicate because they rely on probability distributions over familiar vocabulary rather than novel word combinations.
The most effective headlines use specific, unusual words that signal expertise and local knowledge.
Mobile vs Desktop Patterns
When mobile SEO became critical, headlines that worked on desktop often failed on mobile. YESEO helps journalists optimize for the mobile-first reality of search.
A headline can be optimized for discoverability but still fail if it does not work on the mobile screen where most people see it.
The Data-Informed Approach
Rather than letting data drive decisions, YESEO uses data to inform editorial judgment. The tool suggests, the journalist decides — exactly the opposite of "AI-driven" content.
Data should inform journalistic decisions, not make them. That is why YESEO puts suggestions in Slack where reporters actually work.
The Takeaway for Newsrooms
The most important finding from 60,000+ headlines? The best headlines are not generated — they are crafted.
AI can suggest patterns. It can provide data on what works. It can optimize for search discoverability. But the specific words that resonate with a local audience, the unexpected angle that captures attention — these come from understanding the community, speaking its language, and bringing a perspective that only a local journalist can provide.
That is why YESEO exists: to put data-informed suggestions tools in the hands of journalists, in the workflows where they work, so they can make better editorial decisions — not replace them.