Host-read and programmatic podcast ads occupy the same slot in the same episode, but they are two entirely different businesses. Host-read is a relationship you sell; programmatic is inventory a machine buys, in milliseconds, one impression at a time.
What “programmatic” actually means in podcasting
Programmatic is not an ad type. It is a buying method, and the ad type it delivers is almost always a pre-produced spot — a studio-recorded 15, 30 or 60 seconds, the same file every listener hears.
The pipeline works roughly like this:
- You expose inventory. Your ad slots are made available to a supply-side platform (SSP), which represents you on the sell side.
- A listener requests an episode. Their app asks your host for the audio file.
- A bid request goes out. The SSP tells connected demand-side platforms (DSPs) that an impression is available, along with whatever context is attached to it.
- Buyers bid in real time. Each DSP evaluates the request against its campaigns and returns a price. This is real-time bidding, and it resolves in milliseconds.
- The winner’s audio is inserted. The winning spot is stitched into your marker and the assembled file is delivered.
Step five is the important one for our purposes: programmatic in podcasting is only possible because of dynamic ad insertion. If the ad were welded into the MP3 at edit time, there would be nothing to auction. The IAB and PwC’s U.S. Podcast Advertising Revenue Study found dynamic insertion accounted for more than 90% of podcast ad revenue by 2023 — the plumbing that makes programmatic possible is now how nearly all podcast ads are delivered, host-read spots included.
What host-read actually is
A host-read ad is a spot that you write (or heavily rewrite), voice yourself, and place in your own show. It is usually sold directly: an advertiser or an agency contacts you, or you contact them, and you agree a rate for a number of impressions.
The thing being purchased is not the sixty seconds of audio. It is your credibility, temporarily lent out. That is why advertisers accept a rate that looks indefensible next to a display CPM.
Nielsen’s research on the format, published in 2020 and based on more than 260 brand-lift surveys across 240 brands, found that host-read ads produced an average 50% increase in purchase and recommendation intent compared with non-host-read ads. The premium is not irrational. It is priced against a measurably different outcome.
The comparison
| Host-read | Programmatic | |
|---|---|---|
| Who sells it | You, or a rep you hire | An automated exchange |
| Who voices it | You, in your normal register | A studio voice actor |
| How the price is set | Negotiated, per campaign | Auction, per impression |
| Typical rate | The highest rates in podcasting | Clears near a market floor |
| Effort per deal | High — pitching, contracts, invoicing, reads | Near zero once connected |
| Revenue predictability | Lumpy. Feast, then nothing | Steady, low, roughly proportional to downloads |
| Creative control | Total. You can refuse the copy | Minimal. You can usually block categories, not scripts |
| Fits your show | By construction | By coincidence |
| Minimum audience | Whatever a sponsor will accept | Often a hard download threshold |
| What the advertiser buys | Your endorsement | Your audience’s attention |
Why an unsold programmatic slot is not free
Fill rate is the share of ad requests that come back with an actual ad. It is never 100%. Requests go unfilled when no bidder clears your price floor, when the listener does not match anyone’s targeting, or when demand for audio inventory is simply thin that week.
This matters more than the headline CPM, because it is the multiplier on everything.
Work it through. Say your show does 4,000 downloads an episode and you have one mid-roll:
- Direct-sold host-read at a $25 CPM: 4,000 × $25 ÷ 1,000 = $100 for that episode, assuming you sold it.
- Programmatic at an $8 CPM with a 60% fill rate: 4,000 × 0.6 = 2,400 filled impressions, × $8 ÷ 1,000 = $19.20.
Those numbers are illustrative — see podcast ad rates for how real CPMs are set — but the shape is the point. The gap between the two is not just the CPM gap. It is the CPM gap multiplied by fill.
And the 40% that went unfilled still costs you something. You either ship silence, ship a house promo, or ship a slightly awkward edit point. Listener patience is spent whether or not money arrives.
Running both
Most shows above the threshold should. The two revenue lines are not in competition — they occupy different slots, or the same slot at different times.
The standard arrangement is a waterfall: direct-sold host-read gets first claim on the inventory, and programmatic takes whatever is left. In practice that looks like:
- Mid-roll to your direct sponsor. The most valuable slot, sold at your rate, read in your voice.
- Pre-roll and post-roll to programmatic. Lower-value positions where a produced spot does less damage.
- Programmatic backfill on unsold weeks. When a campaign ends and the next one has not started, the slot earns something rather than nothing.
This is only manageable because insertion is decided at request time. You are not re-cutting episodes when a sponsor’s flight ends — you set the campaign’s end date and the slot reverts. If that mechanism is new to you, baked-in vs. dynamic ads covers what changes when the ad stops being part of the file.
The download threshold problem, honestly
Here is the part most articles skip. Programmatic is largely unavailable to small shows.
Many programmatic marketplaces require a minimum monthly download volume before they will accept a podcast, and those minimums vary enormously between providers and get revised without much notice. Some have opened up to almost any show; others want thousands of downloads a month. There is no single industry number to plan around, so check the current requirement of whoever you are considering rather than trusting a figure you read somewhere.
The consequence is straightforward. If you are a few hundred downloads an episode, the automated money is probably not available to you yet, and chasing it is a distraction.
What is available is the thing that has always worked at small scale: find three or four businesses that would pay real money to reach exactly your listeners, sell them a read directly, and insert it dynamically so you can swap it, extend it, or retire it without touching the episode files.
Which one should you build around?
Answer three questions.
Do you have anything to sell? Not downloads — a specific audience an advertiser can name. “Fantasy football players who spend money on tools” is sellable at 800 downloads. “General interest” is not sellable at 8,000.
Will you actually do the sales work? Host-read revenue is not passive. If you know you will not send the emails, chase the invoices and record the reads, an honest programmatic dollar beats a theoretical host-read ten.
Is your show big enough for automation to be worth the plumbing? Below a few thousand downloads an episode, the answer is usually no, and that is fine.
Most shows land on the same answer: host-read as the business, programmatic as the floor under it, both delivered through the same markers. TurboCast’s ad insertion — markers, campaign start and end dates and promo insertions — is on the Hosting Pro plan at $29/mo billed annually, if you want the direct-sold half of that working before you are big enough for the automated half.