Journalists trust sources that state their limits. This page is the linkable standard that underwrites every report we publish. It does not change per story; individual reports add only the caveats specific to their window.
Every figure comes from the Socialhose Public API, the same aggregated social- and news-listening feed that powers our live boards. We describe the collected, matched feed: posts that our campaigns’ keywords caught, on the platforms we collect from. This is a large, structured sample of public conversation. It is not a census of everything said, and we never write as if it were.
A report is a fixed point. Its numbers are captured for a stated window and do not move after publication, so a citation stays true. When you want the current reading, we send you to the live instrument. The permanent artifact and the live window are deliberately different objects.
Sentiment reflects the emotional register of a post, assigned by automated classification. Conflict, disaster, and health-crisis coverage skew negative by nature. A high negative share is a property of the subject, not an accusation. We report it as tone and resist reading moral judgement into it.
A platform or campaign with little matched volume in a window may be genuinely quiet, or it may be under-collected. We surface these gaps rather than hide them, and we treat an empty cell as a question about our coverage before treating it as a fact about the world.
Likes, reposts, saves, and comments are pooled as “engagement”. They are not comparable across platforms; a save on Instagram and a repost on X are different acts. We use engagement to locate attention, never as a common currency of importance, and we say so wherever platforms are compared.
Keyword matching catches off-topic posts, especially in low-count narratives. Any single post we surface is a lead to be checked against primary reporting before it is cited as fact, by us or by anyone quoting us.
If a figure is wrong, we annotate the report with the correction and the date, rather than silently editing. The record of what we said and when is part of what makes us citable.
Social Knowing is an independently-branded research property. Our underlying data comes from the Socialhose Public API, a professional social- and news-monitoring service, which we credit as our data source. Editorial selection, analysis, framing, and every published conclusion are our own. We link to the raw signal on the live boards precisely so you can check our work against the source.