A host-read ad is voiced by the show’s own host, usually from a set of talking points, in the host’s words and at the pace of the show. A produced ad is a finished spot recorded before the campaign, either voiced by an announcer or actor (announcer-read) or made by the brand itself (brand-produced), and played in the ad breaks of many shows at once. Host-reads usually cost more per thousand downloads and tend to do better on recall and response, while produced spots cost less, reach more shows faster and let the brand approve every word. We build most plans around host-reads and use produced spots where reach, speed or fixed wording matters more.
What each format is
IAB’s Podcasting Buyer-Seller Checklist (October 2023) distinguishes three kinds of podcast creative: spots read by the host, spots read by an announcer, and spots the brand produces itself. Spotify draws the same line in its guide Who should voice your podcast ads? (2021), between host-read ads voiced by the host in the show’s style and voice talent ads pre-recorded by an actor or producer.
In practice a host-read is a small piece of the episode. The host has usually been sent the product, has read a brief, and talks about it in their own voice for anywhere from 30 seconds to a couple of minutes, often with a remark about the episode they are in. A produced spot is closer to a radio commercial, written, cast, recorded and mixed before the campaign starts and then delivered as a file that every show in the plan plays unchanged.
What the research says
Most of the published research favors host-reads, although a Nielsen study commissioned by Spotify found the two formats about equal on recall and purchase intent.
Nielsen reported in October 2020 that host-read ads produced, on average, a 50% increase in purchase and recommendation intent compared with ads that were not host-read, across more than 260 surveys for 240 brands. A year later, in September 2021, it reported a 71% brand recall rate for host-read ads and found that ads running 60 seconds or more lifted purchase and recommendation intent by about 40% more than shorter ads did. Both studies are now several years old.
More recent figures point the same way. Magellan AI’s Q2 2026 measurement benchmark (September 2026) put the median response rate for host-read ads at 3.22% against 2.09% for produced ads in the first half of 2026, where response rate means the share of reached households that visited the advertiser’s site within 30 days. Magellan sells attribution, so this is a vendor’s figure, and we read it as one.
The Spotify study narrows the gap. Its Nielsen fieldwork ran in November and December 2020 among 5,999 US listeners, and Spotify reported in 2021 that host-reads did best when they ran between 30 and 60 seconds and voice talent ads when they ran under 30, and that on recall and purchase intent the two came out about level. The IAB checklist (October 2023) points the same way, suggesting that an announcer spot cut to suit a show’s tone will frequently do as well as a host-read at a lower CPM.
Read together, the studies suggest that a produced spot which sounds as if it belongs in the show can close much of the gap, and that a host-read which sounds like a host reading someone else’s copy loses much of its advantage. The format matters less than how well the ad fits the show it plays in.
What each one costs
Host-reads cost more per thousand downloads (CPM) on the published rate cards. Acast’s pricing guide (2026 pricing, as of October 2026) puts pre-recorded announcer ads at about $15 to $30 CPM and asks $25 to $40 for a host-read sponsorship, rising to $40 or more on niche or high-value shows such as business, finance and B2B. The Libsyn Ads rate card (2026 edition) prices a baked-in host-read at $18 to $22 CPM for a 30-second read and $24 to $26 for a 60-second one, compared with $12 to $15 for programmatic placements. Both are sellers’ own published prices.
A produced spot also carries a production cost that a host-read does not, since it has to be written, cast and recorded before it runs. What the brand gets in return is control, because once a produced spot is approved every listener hears the same words, while a host-read differs from show to show and has to be checked each time it airs.
Where host-reads are stronger
A host-read works because listeners have spent a long time with the host and have a sense of when the host means something. The read carries some of that, so a host who has used the product and says what they noticed will usually do better than any copy we could write for them. Host-reads also suit products that take a minute to explain, since the host can take the time and answer the obvious objection in passing, and they tend to improve across a run of episodes as the host settles into the product.
They ask more of the advertiser in return. Every host reads differently, some will drift from the brief, and claims made in a host’s own words still have to be claims the brand can stand behind. That is why we check each read once it airs, and why the brief gives the host reasons and a few exact lines in place of a page of copy.
Where produced spots make sense
Produced spots are the better choice in four situations.
- The plan needs reach across many shows quickly. A finished file can run across a network or a programmatic buy within days, which suits a launch date or a seasonal offer.
- The wording has to be fixed. In categories where every word needs approval, such as some financial and health products, a produced spot removes the risk of a host improvising a claim.
- The test is about which shows work. Running one spot across several shows keeps the creative constant, so the show becomes the thing being tested.
- A short reminder will do. The Spotify and Nielsen study found voice talent ads did best under 30 seconds, so a short reminder inside an otherwise host-read flight can be produced.
Disclosure
The FTC’s Endorsement Guides ask for no extra disclosure of payment on a host-read that is plainly a commercial, and a produced spot announces itself as an ad. The Guides, as revised in June 2023, illustrate this with a host who reads an unmistakable commercial early in a podcast, a case in which listeners would assume the host had been paid (16 CFR 255.5, Example 12). Disclosure needs care when a host-read drifts into the conversation and starts to sound like editorial, which is a reason to keep reads clearly framed as sponsorship.
How we decide
For a brand selling something considered, such as a financial product, a course, a piece of software or a well-made object, we start with host-reads on a short list of shows whose hosts suit the product, since the format’s advantages are largest there. We add produced spots when the plan needs reach the host-read list cannot give, when wording has to be fixed, or when a short reminder will do.
We also use a hybrid, a host-read built around two or three mandatory lines that the host says exactly, with the rest in their own words, which keeps the legal language intact and leaves the read sounding like the show. Whichever format we choose, we measure each show separately so the plan can move money toward what is working.