How it works

The panel

We follow a panel of leading podcasts in eight domains, chosen from the Apple Podcasts charts and grown step by step. Every new episode is transcribed, split into chapters, and each person, company, product, place and work mentioned is identified and linked to one record, so “the Fed” and “the Federal Reserve” count as one.

Some things never count: advertisements, a show's own hosts and sponsors, and the names a show repeats in every episode (its own credits, for example).

Share of voice

Raw counts mislead: a busy week makes everything look bigger. We measure share of voice instead: mentions of a name per 1,000 mentions of anything on the same shows in the same window (shown as ‰). No single show can supply more than half of a name's count.

Share of ears

A mention on a show heard by 800,000 people reaches more ears than one on a show heard by 800. Share of ears weights every mention by its show's typical audience per episode: the public view counts of the show's full episodes on YouTube, matched to the show's own episodes so clips and look-alike channels never count. Listening on Apple Podcasts or Spotify is not public, so a show without full episodes on YouTube counts as much as the typical show at its chart position, and a show with a smaller YouTube audience than that counts somewhere in between. Wherever we can, these are measured public numbers rather than modeled listener estimates.

Trending

A name trends when its share of voice and its share of ears rise well above their own recent level (the previous eight weeks for the 7-day list), weighted by how widely it is heard: how many shows talk about it, and how big their audiences are. Names heard on only one show never trend. “Most talked about” ranks by share of voice alone. A breakout is a name first heard in the last four weeks and already on at least two shows.

Tone

For each episode, a model reads the passages around each mention and scores how the speakers talk about the name, from very negative to very positive, quoting the words it relied on. A name's tone is the average over episodes, shown only once at least three episodes have been scored. It describes the conversation, not the facts.

“Why is it trending?”

The summaries are written by AI from the moments where a name came up. Every sentence cites those moments, every quote is checked word for word against the transcript, and every number comes from our own counts. Sentences that fail these checks are removed. Each source links to the episode at the second it was said.

The shows

AI & Tech

Markets & Investing

Business & Startups

Health & Science

Politics & World

Sports

Culture & Entertainment

Self-improvement