How the VoiceRank score works
One number, 0 to 100, for how strong a podcast booking someone is. It combines four signals, each measured against every other rated voice in the network rather than against an absolute target — so a score of 80 means “ahead of 80% of rated voices,” not “80 out of a possible 100 points.”
The four signals
Presence
30% of the scoreHow consistently they are actually being booked
Guest appearances on shows that aren't their own over the last 12 months, weighted so recent bookings count more than old ones (an appearance from 90 days ago counts half), then damped by how many different shows they reach — so a standing weekly slot on one show doesn't read as an omnipresent guest.
Reach
25% of the scoreHow many people actually hear them
Total views across the episodes they guest on, on a logarithmic scale — the difference between 10k and 100k views matters far more than the difference between 3M and 3.1M.
Pull
25% of the scoreWhether the audience shows up because of them
View lift: for each show, the average views of episodes they appear on divided by that show's own average, weighted by how many episodes they did there. A 2.0× means their episodes pull double the show's norm. Their own shows are excluded — the question is what they do for someone else's audience.
Authority
20% of the scoreHow central they are to the conversation
Their position in the guest network — who they appear alongside, and how connected those people are in turn. Computed weekly across the whole graph.
Why some people are “Unrated”
A score built on too little evidence is worse than no score, so people below the signal floor are shown as Unrated rather than given a low number. To be rated, someone needs at least three guest appearances across at least two different shows they don’t host, and enough corroborating signal that the score isn’t resting on a single pillar. Hosts who never guest elsewhere are not rated at all — the score measures guesting, and ranking a host low for staying home would be meaningless.
When someone genuinely has no data for one signal — no view counts on their episodes, for instance — that signal is left out and its weight is redistributed across the others, rather than counted as a zero they didn’t earn.
How often it updates
Scores recompute every night as new episodes are ingested. The change shown next to a score (▲3, ▼2) is week-over-week, not day-over-day — a number meant to be quoted shouldn’t twitch daily. Because the score is relative to everyone else, it can move when the field moves, not only when a person does.
What it doesn’t measure
- · Audio-only listens. View counts come from YouTube. Shows that publish audio-only, or whose audience is mostly on Spotify or Apple, are undercounted on Reach and Pull.
- · Everything ever said. The corpus is large but not exhaustive, and skews English-language and video-first.
- · Whether they’re any good. VoiceRank measures reach, pull and centrality — not craft, and not whether a particular guest is right for a particular room. That judgment stays yours.
Questions about a specific score? hello@voicerank.app. Or browse the rankings.