More than double.
For a year, Veritone tracked 250 local campaigns on about 100 stations. Spots read live outperformed pre-recorded ones by 109%.
Six studies, with their sources.
Radio comes first because it is closest to a live read. Podcast studies measure host reads recorded ahead of time: related, not the same.
| Year | Study | Finding | Source |
|---|---|---|---|
| 2020 | VeritoneRadio and TV250 local campaigns, about 100 stations, one year | Live reads outperformed pre-recorded spots by 109%. | Katz Radio Group › |
| 2015 | Westwood OneRadioOne advertiser, Blue Cross | Listener responses up 40% over pre-recorded spots. | Westwood One › |
| 2025 | AudacyRadioA listener survey | 2.7 times as many listeners agreed as disagreed that a host’s ads get more of their attention than recorded commercials. | Audacy › |
| 2020 | NielsenPodcasts · recorded host reads260 surveys, 240 brands | Brand recall of 71% after a host-read ad, against 62% for other ads. | Nielsen › |
| 2021 | Stitcher and Signal HillPodcasts · recorded host readsHost reads against announcer reads | Unaided recall 66% higher than announcer reads for established brands, 59% higher for emerging ones. | SiriusXM Media › |
| 2026 | Magellan AIPodcasts · recorded host readsCampaigns from January to March 2026 | Response rate of 2.45% for host-read ads, against 1.93% programmatic and 1.51% produced. | Radio Ink › |
The other side, in full.
Host-read and voice-talent podcast ads had nearly the same effect on recall and purchase intent, across 5,999 listeners.
Spotify Advertising ›With one script, AI copies of hosts’ voices came within 9% of the hosts themselves on paid subscriptions, and beat announcer reads by 26%.
Podnews ›Both measured ads recorded before they ran. A live read is spoken once, during the event, as the crowd hears it.
The stadium is unmeasured.
Every figure above comes from radio, TV or podcasts. No published study has measured a read over a stadium, arena, track or fairground PA, so this page puts no number on one.
- Radio3studies on this page
- Podcasts5studies, two of them against
- Stadiums and arenas—no published study
- Tracks, rodeos and fairs—no published study
The time, then the claims.
Each read keeps the second it aired. Give it an offer, and each claim is tallied from that second on.
Counted
- The exact second on air, and what recorded that second
- Each claim on its offer, minute by minute
- Every airing on its own, a moment read’s included
Never counted
- Everyone who heard it
- What anyone bought afterward
- Who claimed: no name, no email, nothing personal
Who sees the claims
- The business that bought it, on its proof
- The organization that sold it, on its report
- A sponsor at a D1 program or a club with its own sales team, once the organization shares them
How a read is proven: the live read standard ›
Questions about results.
Do live reads outperform recorded ads?
In radio, yes. Across Veritone’s 250 local campaigns, reads done live drew 109% more response than recorded spots, and Westwood One measured 40% more responses for Blue Cross. Podcast studies of host reads mostly agree, though Nielsen, testing for Spotify, found almost no difference.
Has anyone studied reads over a stadium PA?
None has turned up. The research here is from radio, TV and podcasts, and none of it is offered as a number for a PA at a game, a race or a fair.
Do AI voices work as well as a host?
On recorded podcast ads, nearly. A 2026 Blue Wire test for IPVanish put hosts’ AI-cloned voices 9% behind the hosts themselves. A live read isn’t recorded at all: it is spoken during the event and can answer what just happened.
Does Live Reads publish results of its own?
No. Each read’s time and claims go only to its buyer and its seller. The one figure Live Reads publishes is the count of reads it holds on record.
Can a D1 program’s sponsors see offer claims?
Not until the organization chooses to show them. At D1 programs, and pro clubs with their own sales team, claims are hidden from sponsors by default; the organization switches them on in its report settings.