Real CPM ranges by show size, why transcript-matching beats title search, how to pitch host vs network, and the 4-8 week soak period founders never plan for.
Podcasts deliver the highest trust per impression of any creator channel — and the slowest ROI curve of any creator channel. If you can't tolerate a 4-8 week soak period before you see the data, sponsor something else.
For founders who can, podcasts are quietly one of the highest-LTV acquisition channels in B2B SaaS, dev tools, and AI infrastructure. Customers acquired from a podcast a founder actually listens to weekly retain at roughly 2-3x the rate of customers acquired from paid social. The unit economics aren't comparable.
This is the playbook. Real CPM ranges, the discovery insight that beats searching titles, how the host-vs-network dynamic actually works, and what to expect once your spot ships.
Three structural reasons the channel hits harder than its raw numbers suggest.
1. The host's voice is the ad. A YouTube viewer sees a sponsorship break and starts mentally tuning out. A podcast listener has trained themselves to hear the host's voice as the show. When the host reads a host-read ad, the listener's filter doesn't engage the same way. That's not a small effect — it's the entire reason the channel works.
2. Listeners are doing one thing at a time. A podcast listener is driving, walking, doing dishes, lifting weights. Their hands are busy, but their attention is undivided in a way it never is on YouTube or X. 60 seconds of host-read ad inside that attention window is a different media unit than 60 seconds of pre-roll on a video someone is half-watching.
3. Episodes don't expire. A YouTube sponsor mention compounds for maybe 30 days before the algorithm moves on. A podcast episode keeps getting downloaded for 6-18 months. Half the conversions from a good podcast spot land after week 4.
The tradeoff: there's no Friday-deadline version of this channel. Anyone telling you they need 200 trial signups from a podcast spot by next Friday has misunderstood the medium.
Podcast pricing is usually quoted as CPM (cost per 1,000 downloads). The number that matters is listened CPM — based on the show's actual median downloads per episode in the first 30 days, not the show's all-time biggest episode.
For a 60-second host-read mid-roll in 2026:
| Weekly downloads / episode | Host-read CPM (B2B / dev) | Host-read CPM (consumer / lifestyle) | Pre-roll / produced CPM | |---|---|---|---| | Under 5K | $20 – $40 | $15 – $30 | $10 – $20 | | 5K – 20K | $25 – $55 | $18 – $35 | $12 – $25 | | 20K – 50K | $30 – $70 | $20 – $40 | $15 – $30 | | 50K – 150K | $35 – $90 | $22 – $45 | $18 – $35 | | 150K – 500K | $40 – $120 | $25 – $55 | $22 – $45 | | 500K+ | Custom (often $60-$150+) | Custom | $30 – $80 |
A few things this table won't tell you on its own:
Host-read sells at a 2-3x premium over pre-roll for a reason — it's roughly 2-3x as effective. Founders who try to save money by buying produced pre-rolls in B2B niches almost always regret it. The trust transfer happens in the host's voice; without it, you're buying audio impressions, not endorsements.
Dev / B2B / fintech CPMs run high. $25-$60 host-read CPM is normal in these niches. If a niche dev show with 8K weekly downloads quotes $400 for a host-read, that's $50 CPM and entirely defensible. The same CPM in a consumer comedy show would be a rip-off.
Top-tier shows are priced like media businesses, not creators. Once a show crosses ~250K weekly downloads, you're negotiating with an ad ops person against a rate card. The flexibility you get with mid-tier creator-run shows mostly evaporates.
What founders actually pay tends to land 15-30% below the top of the range after one negotiation round, especially on multi-episode commitments.
For comparison against other channels: see YouTube sponsorship rates (where CPMs run higher but reach is shorter-lived) and Substack sponsorship rates (where dwell time per impression is dramatically higher than either).
This is the single biggest discovery mistake founders make on podcasts: they search show titles.
Show titles are marketing artifacts. A podcast titled "The Future of Marketing" might never discuss your category in a year of episodes. A podcast titled "ACQ²" or "Latent Space" or "Software Engineering Daily" might discuss your exact category every single week. You will never figure that out from the title.
The right primitive is transcript-level search. Pull transcripts (most podcast platforms expose them now, and Apple, Spotify, and YouTube all auto-generate them). Search for the words your buyer uses — the actual technical terms, competitor names, problem descriptions. The shows that come up are the ones whose audience is already discussing your category, which means the host already has the vocabulary to talk about your product without making it sound forced.
Concretely, what to search for:
A 6K-download show whose host has discussed your category 11 times this year is worth more to you than a 60K-download show that's never touched it. The ratio is not subtle — it's often 10-20x on conversion.
This is also where most off-the-shelf podcast advertising platforms quietly fail founders. They sell you on reach by category tag, not by what the host actually talks about. Tags lie. Transcripts don't.
Two completely different humans, two completely different rules.
Most podcasts under ~50K weekly downloads are still founder-run or host-run. You're emailing the same person who edits the audio. The rules are creator-rules:
Reply rates we see on this kind of outreach run 20-35% for relevant pitches. Generic templates run 1-3%.
The leverage move at this tier: offer to be a guest and a sponsor in the same conversation, with the sponsorship in a different episode. Done well, this is one of the highest-converting moves in B2B podcasting because the listener now associates the host's recommendation with a face and a story they've heard.
Once a show is on a network — Acast, Libsyn, Megaphone, Wondery, iHeart, Vox Media — you're emailing an ad ops person who has 40 deals open this quarter. The rules invert:
The right move with networks: ignore the package pitch, hold to the specific show, and ask for a multi-episode test (3-4 episodes) at a discount in exchange for an option to scale into a quarterly buy if performance hits. Networks will say yes to this far more often than founders expect, because their internal metric is filling inventory.
If your target show is on a network, do not pitch the host directly first. The host can't sell you the spot, and going around the ad ops team poisons the relationship before it starts. Pitch the network, mention the host by name, and let them set up the conversation.
Podcasts have a soak period. This is the part founders most consistently underestimate, which is why most "podcasts didn't work for us" postmortems are actually "we measured at the wrong week."
Here's the realistic timeline for a single host-read sponsorship in a B2B / dev / SaaS show:
| Window | What's happening | What to measure | |---|---|---| | Week 1 (launch week) | First 30-50% of total downloads land. UTM clicks spike, but conversions lag. | Click volume, brand search lift, direct traffic spikes on episode-drop day | | Weeks 2-4 | Long-tail downloads continue. Listeners who heard you week 1 start converting. Brand-search and direct-traffic conversions begin showing up. | Trial signups, demo bookings, direct-traffic conversions tagged to time-of-episode | | Weeks 4-8 | The "soak" finishes. Listeners who needed 2-3 mentions to act now act. Network effects kick in if you've sponsored adjacent shows. | Trial-to-paid conversion, LTV cohort tagging, attribution surveys | | Week 8+ | Compounding tail. Episode keeps getting downloaded. Conversions continue at 5-15% of week-1 rate, sometimes for 12+ months. | LTV, retention, post-attribution survey responses ("how did you hear about us") |
The mistake is measuring at the end of week 1 and concluding the spot didn't work. In B2B podcast sponsorships, week 1 conversions are typically 25-45% of total 90-day conversions. If you turn off the spend before week 4 because week 1 was quiet, you're killing the channel before it shows up.
The other reason to plan for the soak: podcast attribution is structurally bad. Listeners hear the URL, often don't click, search your brand later, and arrive via direct or organic. UTM-tracked conversions usually represent 30-60% of true conversions from a podcast spot. The rest show up in brand search lift and direct traffic — which means you need to baseline brand search before the episode drops.
This isn't a Friday-deadline channel. If you need pipeline by Friday, run paid creator content on YouTube instead, where the attribution loop closes in 2-3 weeks. Podcasts are a Q+1 channel: you spend in Q1, you measure in Q2.
Here's a generic-but-useful pattern we see repeatedly with B2B SaaS and dev tool placements. Treat the numbers as a representative shape, not a specific deal.
Setup. A Series A dev tool company decides to test podcasts. They pick three independent shows in their niche, each with 8K-25K weekly downloads, each with hosts who have organically discussed their category in transcripts. Total spend: ~$9,000 across three host-read mid-roll spots, run 2-3 weeks apart.
Week 1 across all three. ~3,200 episode downloads day-of, climbing to ~9,000 by end of week. Tracked-link clicks: 280. Trial signups: 41. The founder is mildly disappointed.
Weeks 2-4. Downloads keep accumulating, now ~22,000 across the three episodes combined. Tracked clicks reach 510. Trial signups reach 92. Direct-traffic spikes appear on the days adjacent to each episode drop, suggesting another 30-40% in unattributed signups.
Weeks 4-8. The compounding kicks in. Brand search volume is up 60-80% over baseline. Customer onboarding survey ("where did you hear about us") starts surfacing the show names. Trial-to-paid conversion on podcast-attributed signups runs 28%, vs 14% on paid social over the same window.
Week 12. Total tracked attribution: ~140 trials, ~38 paying customers. Estimated unattributed contribution from brand-search and direct: another 15-25 customers. Blended CAC lands at roughly 60-70% of paid social CAC for the same period. LTV cohort tracks 2.1x higher.
The pattern. The two shows where the host had used the product before recording outperformed the third by 3x. The founder commits to a quarterly buy on the two winners and drops the third. By month 6, the two winning shows are running monthly and combined make up the cheapest acquisition channel in the company.
The lessons that generalize:
A defensible podcast test for a post-PMF startup looks like this:
Anything smaller than this won't generate enough data to find the pattern. Anything bigger before you've found the pattern is buying inventory, not running a test.
This sits inside the broader influencer marketing playbook for startups — podcasts are one of four channels worth running, and the principles (5-8 deals minimum, per-creator measurement, scale into the pattern) apply identically here.
Podcasts are the channel where the founders who win at creator marketing eventually do their biggest spend, and the channel where founders who quit creator marketing usually quit at week 3.
The medium rewards patience and punishes Friday deadlines. If you can sit through a 4-8 week soak, treat the first 3-5 spots as exploratory, and resist the temptation to declare victory or defeat at week 1, podcasts will quietly become one of your highest-LTV channels.
Pick shows by transcript match, not by show title. Pay for host-read, not pre-roll. Measure at week 8, not week 1. Scale into the pattern when it shows up.
That's the entire game.
GrowthHunt's podcast discovery surfaces shows by transcript-level topical match, not category tag — so the shortlist is ranked by what hosts actually talk about, not what the show description claims. Same engine across YouTube, Substack, podcasts, and X. See it in action →
Related: The Complete Guide to Influencer Marketing for Startups · YouTube Sponsorship Rates · Substack Newsletter Sponsorships
Join 1,847 founders who have already reserved their seat.
Join the waitlist →