Artificial intelligence for real insight into news
At speed and scale, our language models “understand” beneath the surface metrics
With our understanding of different “tones” of content, we tap into a goldmine of hidden data. Our models capture abstract, human signals like structure, type and tone. Invaluable insight you can’t get through keywords, clicks and shares.
How it works
Qualitative Data
Qualitative data means looking at the abstract, human-understandable qualities of an article or audio clip itself, rather than seeing how many clicks, likes or shares it received to know something about it. Overtone uses the emerging power of language models to “understand” the text.
Today the media industry uses clicks, keywords, shares metrics and get the what…
…but miss the reader experience, personalisation and the why
A.i. as a Solution
Meet Overtone
Overtone looks at the actual text of an article itself like a human would, to provide understanding where metrics like clicks, shares and keywords fail.
The two articles below are from the same outlet, with the same exact keywords, and have roughly the same same length. Most systems in publishing, PR or advertising would treat them the same, but they are radically different editorially.
Overtone scores are based on the editorial differences between articles, differences that most editors and human beings would understand. Because these signals are not based on waiting for people to engage, that means they are also available immediately the moment a piece of text or audio clip exists.
Articles may look the same, but are radically different
