A host-read ad is the most valuable thing a podcast produces. Not the episode; the sixty seconds in the middle of it where the host, in their own voice, tells the audience about a product. The industry pays a premium for it over pre-produced spots because it works: three quarters of weekly podcast listeners say they have acted on a podcast ad, and the reason is that the person recommending the thing is the person they chose to spend an hour with. This guide covers the two questions every host faces the first time a sponsor writes in. What do I charge, and how do I deliver this so they come back?
How the Industry Prices a Host Read
Podcast advertising is sold on CPM, cost per thousand downloads, measured over a window (usually thirty days) after an episode is published. Published 2026 benchmarks from the major ad marketplaces cluster host-read mid-rolls between $15 and $30 CPM on the buy side and $25 to $40 on sell-side rate cards, with pre-rolls a little lower and post-rolls lowest of all because listeners drop off. Niche shows with a high-value audience, business and finance especially, command $40 and above. These are real numbers and they are what a network or agency will quote you.
Now do the arithmetic for a show that is not on a network. An episode that reaches 2,000 downloads in the window, which puts it comfortably in the top 10 percent of podcasts on most hosting platforms, earns $50 for a sixty-second host read at $25 CPM. Recording the read, checking it against the script, editing it into the episode, and sending the sponsor a clean file takes the better part of an hour. At 500 downloads it earns $12.50. CPM is a pricing model built for shows with six-figure audiences, and applying it to a small show means working for less than minimum wage on the most valuable minute of your episode.
Why Sponsors Pay More for the Host's Voice
The premium on a host read is not sentiment; it is measured. The Podcast Study 2026, a nationally representative survey of 1,205 US podcast consumers by Point-To-Point Marketing and Strategic Solutions Research, asked listeners what a host endorsement does to them. Fifty-seven percent said they are more likely to remember the brand, 55 percent more likely to notice the ad again, 44 percent more likely to trust it, and 41 percent more likely to consider buying. The same study found that 82 percent say the host is a big part of, or the only reason for, listening to their favourite show, and that 45 percent would at least consider paying for a subscription tier that removed every ad except the host-read ones. Read that last figure twice: listeners will pay to keep your ads and lose everyone else's.
The money side matches. The IAB and PwC's full-year 2025 report put US podcast advertising at $2.862 billion, up 17.6 percent on the year, and noted that dynamic insertion now carries 84 percent of that revenue. Dynamic insertion is often described as the enemy of the host read, and for pre-produced spots it is; but the same report found host-read ads are now mostly served dynamically too. In plain terms: the sponsor still wants your voice, they just want to be able to swap the read in and out of old episodes. That is a technical delivery requirement, and it is why the second half of this article is about the file, not the fee.
There is a catch in the numbers that an independent should not skip past. The $2.86 billion is sold overwhelmingly against downloads, and the download curve is brutally steep: on one major hosting platform the median episode gets 27 downloads in its first week and the top 1 percent gets over 4,500. The CPM market lives above the top 5 percent line. Below it, which is where most shows are, the host read is still the most valuable thing you make; it is just sold differently, on trust and fit rather than on volume, which is the whole point of the next section.
What to Charge Instead
Small and mid-sized shows do better selling direct to sponsors on a flat fee, priced on the value of the audience rather than its size. A local business, a niche software company, a course creator whose customers are exactly your listeners: these sponsors care that the right 800 people hear it, not that 80,000 random ones do. The fee is then a negotiation about what an introduction to your audience is worth to them, and the honest floor is your time plus the trust you are lending.
- Price the read as production work first. A sixty-second host read is a scripted voice performance with a delivery deadline, and voice work has established rates. The free Rate Calculator prices a scripted read by length, usage and term; use it as the floor and add the audience on top.
- Charge for the audience per episode, not per thousand. A flat per-episode fee for a show under 5,000 downloads commonly lands between $100 and $500 depending on how specific the audience is and how well the sponsor fits it. Bundle four or eight episodes for a discount and a sponsor who commits.
- Charge more for usage beyond the episode. If the sponsor wants to use your read in their own ads, on their site, or in paid social, that is a separate licence with a separate fee, exactly as it would be for a voice actor. Say so in the agreement.
- Revisions are billable after the first. One round of changes is reasonable. A sponsor who rewrites the script three times is asking for three recordings. The Revision Checker counts how much of the script actually changed so the fee is fair to both sides.
Say the number in writing before you record
Sponsors are not offended by a rate. They are put off by vagueness. A one-page sheet with your per-episode fee, what it includes (one sixty-second mid-roll, one round of revisions, delivery within five working days), and what costs extra (usage, additional reads, rush) closes more deals than a long email about your audience.
Delivering a Read the Sponsor Approves First Time
The second half of the job is the one that decides whether you get the next booking. Sponsors approve reads on four things, and they reject them on the same four. Get all four right and the approval email arrives in an hour. Miss one and you are re-recording.
1. Every word the script says, and no others
A host read has room for personality but not for paraphrase. Product names, offer codes, URLs, the exact wording of a claim: the sponsor's legal and marketing teams signed those off, and a dropped word or an improvised alternative is the most common reason a read comes back. Read the script as written, then check it. Paste the sponsor's script into Script Accuracy Checker, upload your read, and every word comes back green or red, so you find the swapped word before the sponsor's producer does.
2. On length
A sixty-second read is a sixty-second read. Sponsors buying sixty seconds have usually priced the placement on it, and a read that runs to eighty is either cut by their producer (badly) or sent back. Practise the read against a clock before you record it; if it will not fit at your natural pace, tell the sponsor before you record rather than after. Producers respect a host who says "this is a seventy-five second script" more than one who delivers a rushed sixty.
3. Clean audio, matching the episode
A read recorded on a different day, with a different mic distance, in a different mood, will sound like an insert. Record it in the same session as the episode where you can, at the same distance and gain. Then clean it exactly as you clean the episode: breaths quietened with the pauses kept, any stumble removed, the room taken out if it needs it. Auto Clean Up does the first two in one pass and lists every decision so you can keep a deliberate beat.
4. At the right loudness, in the right format
If the sponsor is inserting the read into their own material, or into other shows, they will ask for a specification: usually a WAV at 48 kHz, -16 LUFS integrated, true peak under -1 dBTP, sometimes -23 LUFS if it is going to broadcast. Ask which, then deliver exactly that. Loudness Normalizer has presets for each and re-measures the finished file, so the number on the delivery note is a measurement, not a guess. Name the file sensibly (sponsor, show, date, version) and send a WAV, not an MP3, unless they ask for one.
↗ Try the tool
Script Accuracy Checker
Script Accuracy Checker compares your read to the sponsor's script word by word. Green is right, red is a re-record, and you know before you send it.
Open Script Accuracy Checker →Three Reads, Three Mistakes
These are composites of the ways a read gets sent back, each of which we have seen more than once. None of them is about talent; all of them are about the file.
- The Monday read. The host records the episode on Thursday, forgets the sponsor, and records the read on Monday morning in a different chair with the mic six inches further away. It is word-perfect and it sounds like a different person walked into the room. The sponsor's producer hears the join before the first sentence ends. Fix: same session, same distance, same energy, or at the very least the same room and a level match before delivery.
- The um read. A relaxed host does a warm, conversational read, which is exactly what the sponsor wanted, but with four ums and a false start in sixty seconds. Charming on the show; unusable as a swappable insert, because the sponsor cannot run a read with a stumble in it for six months. Fix: one clean pass with the fillers and false starts taken out, breaths quietened, and the pauses left alone so it still sounds like a conversation and not an announcer.
- The loud read. Delivered as an MP3 at -12 LUFS with peaks touching zero, because it sounded punchy in the headphones. Every platform normalises it down, the limiter squashes the consonants on the way, and it lands in the episode noticeably duller than the material either side. Fix: -16 LUFS integrated, true peak under -1 dBTP, WAV at 48 kHz, with the spec printed on the rate sheet so it is never a surprise.
The Part That Is Not on the Invoice
A sponsor who books once and books again is worth more than any rate card, and repeat bookings come from the boring things: delivering on the day you said, sending a file that needs no fixing, reporting the download figure honestly, and telling them when something in the script will not work for your audience before you record it rather than reading it flat. The read itself should sound like you believe it. If you do not, the audience will hear that in about four seconds, and it costs you more than the fee.
For the wider picture of where podcast advertising money is going in 2026, and how little of it reaches the median show, see The Podcast Industry in 2026: What the Numbers Say for Independent Creators. The tools mentioned here, in the order an episode uses them, are on the For Podcasters page.
Frequently Asked Questions
My show is small. Can I still sell a host read?
Yes, on a flat fee rather than a CPM, to a sponsor whose customers are your listeners. A show with 400 downloads an episode about, say, home brewing is worth more to a small brewing-supplies shop than a show with 40,000 downloads about everything, and the numbers above on trust and purchase intent apply at any size. Price it on fit and on what the placement includes, put it on one page, and send the page.
Should a host read have ums in it?
Not if the sponsor is going to reuse it. A read that lives in one episode can be as loose as the episode; a read the sponsor inserts into other shows or runs for months needs to be clean, and cleaning it should not turn it into an announcer. Take out the fillers and false starts, quieten the breaths, keep the pauses. The warmth is in the timing, not the stumbles.
What file should I send?
Unless the sponsor specifies otherwise: WAV, 48 kHz, 24-bit if you have it, -16 LUFS integrated, true peaks no higher than -1 dBTP, named with the sponsor, the episode and the date. If they ask for an MP3, 320 kbps and the same loudness. Send the spec with the invoice so the next booking is easier than the first.
Sources and Further Reading
- Point-To-Point Marketing and Strategic Solutions Research, The Podcast Study 2026, 1,205 US podcast consumers, presented September 2026, as released via Podnews: the host endorsement figures (57, 55, 44 and 41 percent), the 82 percent host figure and the ad-free-tier willingness.
- IAB and PwC, Internet Advertising Revenue Report, Full-Year 2025, April 2026, as reported by Radio Ink: podcast revenue of $2,862 million, the 17.6 percent growth and the 84 percent dynamic-insertion share.
- Buzzsprout, Podcast Stats, August 2026: first-week download thresholds by tier.
- Apple, Audio requirements for Apple Podcasts: the loudness and true-peak recommendation used as the delivery spec.
- VoiceEditSuite, The Podcast Industry in 2026 and The Podcast Ad Boom and What It Means for Voice Talent: the companion reports.
Corrections
Survey and revenue figures are as published by The Podcast Study, the IAB and PwC, and Buzzsprout at the time of writing, September 2026, and are revised over time. Spot an error or a newer number? Tell us through the contact page and we will correct it with a note.
