Pricing Start Free Trial

The State of
Video Podcasting,
 2026.

An evidence-based portrait of approx. 34,000 active English-language video podcasts — who makes them, who listens, what actually moves the needle on growth, and why the ecosystem behaves nothing like the rest of the creator economy.

Scope — English-language only
Method — ~34K video · English · active 12mo
Podcasts Analyzed
0
active English-language video podcasts (12-month activity filter)
Total Episodes Published
0
cumulative episodes across the active catalog — an observed count, not a modeled estimate
Run Sponsorships
0%
monetize via brand partners
Median Modeled Listeners
0
modeled median for shows in this catalog — not directly comparable to per-episode download benchmarks
READ THIS FIRST

What this report measures

  • Catalog-level scan of approximately 34,000 unique English-language video podcasts active in the last twelve months
  • Directly observed counts: episodes published, Apple review counts, platform presence (Apple / Spotify / YouTube), publishing cadence, sponsor activity flags
  • Modeled estimates: monthly listener counts, audience composition (gender, income tier, generation), top-decile concentration metrics
  • Directional reads of how the active English-language video podcast market is structured today
Throughout the report, listener counts and audience composition figures are modeled estimates; episode counts, review counts, platform flags, and sponsor flags are directly observed. Where a figure is modeled, the surrounding prose says so explicitly. See methodology sections /05, /14, and /16 for full detail.
/01Industry Overview /02Audience & Creators /03Audience Intelligence /04Engagement Analysis /05Content Patterns /06Monetization & Format /07The Top 10 /08Key Insights /09Try It — Scorecard /10Conclusion /11Methodology /12How to Cite
/ 01  Industry Overview

An ecosystem of
extreme inequality.

Most podcasts have small, devoted audiences. A tiny minority command the spotlight, the sponsorships, and very nearly all of the monthly listening hours. The middle barely exists.

Note — Every figure in this report is drawn from a curated set of English-language podcasts. Country counts and category trends reflect the English-speaking podcast world, not total podcast activity in any given region.

The video podcast market in 2026 looks healthy on the surface and quietly brutal underneath. This report covers approx. 34,000 active English-language video podcasts (12-month activity filter) — one of the largest publicly-analysed podcast intelligence sets available. The dataset spans the full long tail (most shows reach a few thousand listeners) and the very top of the market (a handful of shows clear the multi-million-listener mark). Average ratings sit at 4.8, and the underlying intelligence data includes modeled audience composition (gender, income tier, generation) and observed sponsor activity wherever it could be detected.

But the moment you look at how listeners are distributed, the picture changes. A typical show in our catalog has a modeled monthly audience of around 2,000 listeners — respectable as a starting point, but a long way from the headline names. The arithmetic mean, by contrast, is around 32,000 — a sixteen-fold gap that only happens when a small number of shows are doing the heavy lifting for everyone else. Note that this median refers to modeled monthly listeners across the curated active video catalog, and is not directly comparable to per-episode download benchmarks published by hosting platforms (which use different units, different cohorts, and different measurement bases).

85.4% of all monthly listening hours flow to the top 10% of shows. The bottom half of the market accounts for around one percent.

That is not a podcast-specific failure. It is the same power law that governs YouTube channels, Spotify artists, and bestseller lists. What's specific to podcasting is how forgiving the surface looks: even shows with tiny audiences carry strong ratings, because the people who do listen tend to love them.

Figure 01.1
Listener reach distribution — show count vs. share of total modeled listening
Bars: number of shows in each tier. Line: percentage of all monthly listening that tier captures.
Total Episodes Published
0
Cumulative episodes ever published across the active catalog. This is an observed count taken from each podcast’s episode total — not a modeled estimate. It is the most direct measure of ecosystem output we can show.
Top-Decile Threshold
0
Roughly the audience floor a show needs to clear to enter the top 10% of the market.
Top 1% Share of Reach
0%
Around 340 podcasts (the top 1% by listenership) hold roughly 42% of all listening across the dataset.

What people make — and what they make a lot of

A clear hierarchy emerges across the long tail. Business & Finance sits at the top with roughly 6,300 shows — a function of the entrepreneurial coaching and finance-personality boom of the last five years. Religion & Spirituality (~3,900), Health & Wellness (~2,500), Sports (~2,300), and News & Politics (~1,900) round out the top five — identity, lifestyle, and conversation formats anchor the volume side of the market.

The categories with the highest show count are not necessarily the same ones that produce the highest-audience shows. True Crime is the clearest example — it doesn’t even appear in the top 10 categories by show count, yet it carries two of the ten biggest podcasts by listenership (Crime Junkie and Criminal). This is the power-law signature of the format: a relatively small number of category entries, dominated by a couple of breakout franchises that command enormous audiences each.

Figure 01.2
Top 10 categories by show count
/ 02  Audience & Creator Insights

Who's behind
the microphone.

Counted at the show level — one podcast, one classification — men still produce the majority of English-language video podcasts. But mixed-gender productions punch far above their weight when you look at audience size.

Figure 02.1
Host gender at the show level (n ≈ 34,000)

Each podcast is classified once, based on the genders of its hosts: male-only (every identified host is male), female-only (every identified host is female), mixed (at least one male and one female host), or unknown (no host gender could be identified). The four buckets sum to 34,208 — one classification per show.

Across the full catalog of 34,208 shows, 36.5% are male-only, 16.6% are female-only, 6.7% are mixed-gender productions, and the remaining 40.2% sit in the unknown bucket where host names or biographies are not recorded. Restricting the calculation to the 20,444 shows where host gender can be identified, the splits become 61.1% male-only, 27.8% female-only, and 11.2% mixed-gender. We report both bases below because the 40% unknown share is large enough to materially shift the headline number depending on how it is treated.

A note on terminology: the percentages above describe the host configuration of the show (who is in front of the microphone), not the audience composition. Audience-side gender skew is analysed separately in §03 and uses different cohort definitions; the two should not be conflated.

Mixed-host shows out-listen single-gender shows by roughly 24% at the median — the strongest format-level signal in the entire creator dataset.

Geography — an English-language map

Country data is available for roughly 21,000 of the 34,000 podcasts in the catalog — about 62%. The remaining ~13,000 shows have no country metadata recorded and are excluded from the geographic breakdown below. Read the percentages that follow as shares of the 21K subset with country data, not shares of the full catalog.

Among the shows where origin can be verified, 79% are produced in the United States, 9% in the United Kingdom, 5.5% in Canada, and 4% in Australia. Those four English-mother-tongue countries account for roughly 94.5% of all video podcasts whose origin we can verify. (Translated to the full 34K catalog, that equates to roughly 48% confirmed-US, with another 38% of unknown origin.)

This skew is a direct consequence of the analytical scope. The report covers only English-language video podcasts, which is precisely why the United States carries such a disproportionate share. The US share is somewhat lower than typical audio-only podcast datasets (which often run 65–80% US) because video podcasting has gained earlier traction in the UK, Canada, Australia, and India relative to audio-first markets. A low number for Germany, Japan, or Brazil here does not mean those markets produce few podcasts — it means few of their shows are produced in English.

The same US dominance carries through into the audience side. The bulk of listener traffic, sponsorship deals, and network signings flows through the same handful of cities — Los Angeles, New York, Austin, and London anchor the visible market.

Figure 02.2
Top 10 host countries (by podcast count)
English-language video podcasts · ~21,000 with country metadata of 34,000 total.
/ 03  Audience Intelligence Deep Dive

Who is actually
listening.

For every show in this dataset, modeled audience composition tells us not just how many people listen, but who they are — their gender mix, income tier, and generation. This section is the qualitative half of the audience picture.

Female Audience Share
0%
Listener-weighted across the entire catalog. The audience leans slightly male in aggregate (M 52% / F 47%), with around 1% modeled as listeners outside the male/female binary.
Millennial Listening Share
0%
Almost half of all podcast listening, weighted by audience size, comes from Millennials. They are the economic centre of gravity of the format.
Mid-Income Listener Share
0%
Six in ten listeners sit in the medium-income tier. The podcast audience is overwhelmingly mass-market, not premium.
Figure 03.1
Audience gender skew — how many shows lean which way (modeled audience)
Female-leaning ≥ 60% female · Male-leaning ≥ 60% male · Balanced is everything between.

Most shows lean male — but balanced audiences trail both

Roughly 45% of podcasts have male-leaning audiences. About 23% are female-leaning, and 33% are balanced. The supply skews male, but the gap is smaller than people often assume — nearly a third of all podcasts are roughly gender-balanced in audience composition.

The audience-side picture is more interesting. Female-leaning shows and male-leaning shows reach essentially the same median audience — 2,500 listeners in each cohort — while balanced-audience shows trail at 1,500. The data does not say one gender skew outperforms the other. What it does say is that shows with no clear audience lean tend to underperform shows with a clear one, in either direction.

Male-leaning and female-leaning shows reach roughly the same median audience. The shows that struggle most are the ones with no clear demographic centre — balanced audiences are harder to grow because there’s no specific community to grow into.

Health & Wellness is a strong example of a female-skewed category, with audiences typically 60–70% female. Sports sits at the opposite end: roughly 70% male on average. The shows that genuinely cross over — Society & Culture, News, mainstream Comedy — tend to land near 50/50 and rely on volume rather than community to grow.

The audience is firmly mid-market

Listener-weighted across the entire active catalog, the income split is Low 21% · Medium 59% · High 20%. Six in ten listeners sit in the medium-income tier. The high-income segment is small — and shows that target it have lower median reach, not higher.

High-income-dominant shows post a median of 2,000 listeners; medium-income-dominant shows median 2,000; low-income-dominant shows trail at 700. Most reach lives in the middle of the income distribution because that's where most listeners are. The high-income tier exists but is small.

The largest reach is in the middle. The biggest sponsor density is at the top. The two metrics don't line up — which is why a show optimised for one can underperform on the other.

The implication for monetisation: a show built explicitly for affluent listeners will struggle to scale audience, but should command sponsor density. A show built for the mass market will scale audience easily, but sponsor competition for that inventory is fierce.

Figure 03.2
Income tier — share of total modeled listening, dataset-wide
Listener-weighted average. Hover any slice for cohort detail.
Figure 03.3
Generational mix — share of total modeled listening across all shows
Listener-weighted average across the catalog. Bars do not need to sum to 100% across age brackets — many listeners belong to multiple modeled cohorts.

Millennials are the centre of gravity

Roughly half of all podcast listening, weighted by audience size, comes from Millennials (49%). Gen X follows at 27%, Gen Z at 16%, and Boomers trail at 7%. The format is age-coded: it was born in the iPhone era and its core audience grew up with smartphones.

At the show level, 82% of all podcasts come back with Millennials as their dominant listening generation, with Gen X dominating 13% and Gen Z 3%. Read the dominance figures as directional rather than absolute — the underlying classifier tends to default to the modal cohort, so the 82% reflects how often Millennials are most often slightly ahead on a show, not that 82% of shows are exclusively Millennial. The listener-weighted Millennial share (49%, shown in the card above) is the more reliable headline number because it preserves the underlying mix. (See methodology /14.)

Build for Millennials by default. Build for Gen Z if you're optimising for cultural traction. Build for Gen X or Boomers only if you're chasing a specific subject vertical.

The strategic read: even shows that think they're targeting Gen Z (most comedy, most pop culture) are still being consumed predominantly by Millennials in absolute terms. The Millennial audience is older than the format's self-image suggests, and that has consequences for tone, sponsorship inventory, and content cadence.

Demographic fingerprints by category

Each category has a different audience signature. The table below shows listener-weighted averages of the female share, high-income share, and the dominant generation for the ten largest categories in the catalog.

Category Shows Female % High-Income % Millennials % Median Listeners
/ 04  Engagement Analysis

Reviews are the
only honest signal.

Star ratings are nearly meaningless — almost every show is rated above 4.5. What truly tracks listener volume is the count of reviews. Volume of opinion, not quality of opinion, predicts reach.

Figure 03.1
Apple rating distribution

Ratings cluster aggressively at the top. 87% of shows score between 4.5 and 5.0 stars. Only 1.7% sit below 4.0. That is not because every podcast is brilliant — it is because the people who bother to leave a rating tend to be the show's biggest fans, and the people who would rate harshly simply stop listening.

As a consequence, the correlation between Apple rating and monthly listeners is essentially zero (r = −0.03). A 4.9-star podcast is not measurably more popular than a 4.5-star one.

The relationship between review count and listener volume, by contrast, is extraordinary — r ≈ 0.84, a very strong relationship.

In plain English: how many people care enough to review a show is essentially a one-to-one proxy for how many people listen to it. Reviews don't drive listeners. They reflect them.

Figure 03.2 — Correlation
Reviews (observed) vs Monthly Listeners (modeled), log scale, r ≈ 0.84, n = 280 sampled
Each dot is one show. Trend r ≈ 0.84 (n = 280 sampled).
0.4%

The share of total monthly listening hours captured by the top 10% of shows. The next 40% take roughly 13% of the audience. The bottom half — some 17,000 podcasts — share what's left, around 1% of all listening.

/05 · CONTENT

Most shows are short.
The biggest shows are long.

WHAT CREATORS PRODUCE
0min
A typical episode is about 41 minutes long
Roughly two out of three podcasts publish episodes between 20 and 60 minutes. That's the comfort zone for most creators — long enough to develop an idea, short enough to fit into a commute or a workout.
WHAT AUDIENCES REWARD
0%
Long shows are linked to bigger audiences
Episodes 90 minutes or longer have a typical reach roughly 225% higher than shows in the popular 20–40 minute range. The long-form bucket is small, but the shows in it tend to reach much larger audiences.
Episode length — how many shows in each range
💡The 20–40 minute bucket is the most crowded, holding ~41% of shows with length data
Catalog size — how many episodes shows have published
💡Most shows sit at 100–500 episodes; only 8% ever cross the 500-mark
How to read these charts: The bars above show the number of shows in each bucket. The audience commentary that follows is based on median listener count within each bucket — a separate calculation, not the heights you see here.

There are two findings in this section, and they don't agree with each other — which is exactly why both matter. The first is what creators actually produce: most shows are 20–60 minutes long, and most of them have somewhere between 100 and 500 episodes in their catalog. That is the working pattern of the average video podcast.

The second finding is what audiences actually reward. The two charts above measure how many shows fall into each length and catalog bucket. If we instead measure the median monthly listenership within each bucket, the picture flips. Short shows (under 20 minutes) median around 1K monthly listeners. Long shows (90+ minutes) median closer to 6.5K — about 6.5× the reach of short shows. Length and audience size are associated; we can't say from this dataset that one causes the other.

A note on sample size: the long-form bucket is small — only about 1,256 shows publish 90+ minute episodes — but the median calculation is robust within it, and the gap relative to the popular 20–40 minute bucket is consistent. The signal is real even though the cohort is narrow. (A further 1,176 shows are excluded from the length buckets entirely because their average episode length could not be parsed; the five buckets shown sum to 33,032, not 34,208.)

Catalog depth tells a parallel story. Shows under 25 episodes sit at a typical reach of 200 monthly listeners. Shows past 500 episodes reach 30K — a 150× difference. The most plausible read isn't that "more episodes equals success"; it's that consistency over time compounds. A show that has reached 500 episodes has typically been publishing on a stable cadence for several years — long enough for word-of-mouth, search visibility, and audience habit to accumulate.

The takeaway is simple. The popular format isn't the highest-reaching format. Most creators settle into 30–40 minute episodes and stop somewhere around 250. The shows that break out of the long tail tend to look different: they run longer per episode, and they keep publishing for longer overall.

A directional read, consistent with broader industry observations: long-form video podcasts behave more like television than radio. Audiences who commit to a 90-minute show are committing to a relationship, and platforms reward sustained watch-time. We can't validate that mechanism directly from this dataset, but the listener distribution is consistent with it.

/ 06  Monetization & Format

Sponsors follow
scale, not stars.

Only one in four English-language video shows runs sponsorships, but those that do are systematically larger — not better-rated — than those that don't.

With Sponsors
0%
Show evidence of brand activity (about 7,800 podcasts). Of these, ~6,300 have detailed sponsor records; per-show sponsor counts cited below are based on this 18.4% subset.
With Guests
0%
Run regular interview or conversation formats.
On a Network
0%
Affiliated with a podcast network. The catalog is overwhelmingly independent.
Sponsor Listener Lift
+0%
Sponsored shows out-reach non-sponsored ones by roughly 7× at the median.
Figure 05.1
Median modeled listeners — sponsorship & format compared

Sponsored or solo, guest or no

A sponsored show pulls in a median of 10,000 monthly listeners, versus just 1,500 for non-sponsored peers — roughly seven times more. The simplest way to read this: brands rarely pick small shows. Almost all of the sponsorship inventory in this catalog is concentrated in the small fraction of shows that have already built audiences large enough to be commercially interesting.

Guest-driven shows reach a median of 2,500 listeners versus 1,500 for solo formats — about 67% higher. The gap is real but modest in absolute terms; the bigger differentiator is the top of the market, where 10 of the top 10 shows are guest-format. Conversation and interview formats dominate at the high end even if they don’t transform a small show.

Sponsorship is best understood as a consequence of audience size, not a cause of it. Brands underwrite shows that already have listeners.

Format combinations — the eight playbooks, sized by audience

There are eight production-strategy combinations a video podcast can run: sponsored or not × guest or solo × network or independent. Each has a different median audience. Hover or tap a bar to see the count of shows running that exact combination, the median listenership, and the strategic note.

Tap a bar · or hover
Choose a strategy combination
Median listeners
—
Shows in cohort
—
The matrix combines three independent strategic choices — whether to run sponsorships, whether to host guests, and whether to sign with a network — into the eight unique combinations a video show can occupy.
Sponsor Density Deep Dive

How sponsorship works at scale

About 23% of all podcasts in the catalog (around 7,800 shows) carry confirmed sponsorships. For roughly 6,300 of these, we also have the names of the specific sponsor brands — the rest are confirmed sponsored but the brand list isn't available, so brand-level analysis below is restricted to the shows where the data is complete.

Among shows where we can count sponsor brands, the typical sponsored show carries 7 different brands in the period observed. The mean is much higher (around 15) because a small number of large network productions run 50, 100, or even hundreds of brands across their archive — pulling the average upward.

Sponsor density tracks listener size closely. Shows with no sponsors median around 2,000 listeners; shows running 11 or more brands median around 10,000 — roughly five times higher. The pattern reads cleanly: brands underwrite shows that already have audiences, not the other way around.

Figure 06.2
Sponsor count distribution — how many shows carry how many brands
Bars show the number of shows in each bracket (only shows with detailed sponsor data are included). Hover any bar to see the typical audience size for shows in that bracket.
Figure 06.3
Most-frequent sponsors across the catalog
Top 12 brands ranked by the number of distinct podcasts they advertise on.

Which audiences attract the most sponsors

A simple way to read this: take all the podcasts in the catalog and split them by their typical listener — mostly female listeners, mostly male listeners, or a roughly even mix. Then ask, how often do brands actually sponsor each type of show?

The answer is clear and consistent: shows with a clearly identifiable audience — whether male-skewed or female-skewed — get sponsored more often than shows with a mixed audience. Sponsors prefer to know exactly who they're reaching.

Male-leaning audience
#1 most sponsored
Share of all podcasts in this group
0 shows
How many of them carry sponsors
0%
Typical brand count when sponsored
0 brands
Most common in sports, business, finance, technology, and politics. The single largest audience-type segment in the catalog.
Female-leaning audience
#2 most sponsored
Share of all podcasts in this group
0 shows
How many of them carry sponsors
0%
Typical brand count when sponsored
0 brands
Common in true crime, wellness, lifestyle, and parenting. When sponsored, these shows often carry more brands per episode than male-leaning ones.
Balanced audience
Least sponsored
Share of all podcasts in this group
0 shows
How many of them carry sponsors
0%
Typical brand count when sponsored
0 brands
Common in news, society & culture, and general-interest interview shows. Brands find it harder to target a non-specific audience.
The pattern is simple: sponsors prefer audiences they can clearly describe. Male-leaning shows are sponsored about 1.6× as often as balanced shows. Female-leaning shows sit just behind, with similar density when they are sponsored.

The income cross-cut shows a similar pattern but is much weaker. High-income-dominant shows carry slightly more sponsor brands on average than medium-income shows, but the gap is small. Most sponsor inventory is bought against medium-income audiences simply because that's where most listeners are; the high-income segment commands a premium per-listener but doesn't add up to enough volume to dominate sponsor activity.

/ 07  The Top Ten

The shows
that move the world.

Ten podcasts. Approximately 55 million combined monthly listeners between them. They share almost no traits with the median show — except, notably, the absolute requirement to be on YouTube.

# Show Monthly Listeners Apple Reviews Rating Network Format Listen
Listener and review counts shown as ranges (e.g. "10M+", "250K+") — these reflect modeled estimates and visible review counts at snapshot date.
Top 10 — Avg Listeners
0
Sum of the top 10 listener counts ÷ 10. Compared to the dataset mean (~32K), the top-10 average is roughly 171× larger; compared to the dataset median (~2K), it is roughly 2,750× larger. The mean comparison is biased upward because the top-10 shows themselves dominate the dataset mean — the median comparison better describes how far above the typical show the leaders sit.
On YouTube
0
Every single top-10 show distributes on YouTube.
Run Guests
0
All 10 of the top podcasts use guest or interview formats — the format dominates the top tier.
Network-Backed
0
Only 3 of the top 10 are formally network-backed (Call Her Daddy, The Daily, Shawn Ryan); the other 7 operate independently or under their own banner.

Compared with the rest of the dataset, the top 10 are universally guest-driven (10/10 vs 72% across the dataset), universally sponsored (10/10 carry confirmed sponsor activity, vs 23% across the dataset), and universally on YouTube. Their median rating, 4.65, is actually below the dataset average of 4.72 — further evidence that ratings don’t make the audience.

What unites them is something the numbers can hint at but can't fully explain: each is a personality-led, video-native franchise — a show that exists in the same cultural register as a late-night talk program, a true-crime docuseries, or a column. They are not "podcasts" in the early-2010s sense. They are media properties.

/ 08  Key Insights

Eight findings
that change the playbook.

The data tells a story that contradicts most podcast advice circulating in 2026. Here are the eight findings most worth carrying forward.

Reviews predict listeners almost perfectly. Stars don't.

The correlation between review count and monthly listeners is r ≈ 0.84 — a very strong relationship, though not the near-perfect one earlier reports suggested. The correlation between star rating and listeners is r ≈ −0.06, statistically zero. Why it happens: highly-rated shows attract their fans, but fans of all popular shows leave reviews at roughly the same rate. So review counts grow with audience, while ratings cap out at the ceiling of "people who liked it." The practical takeaway: when comparing two podcasts you know nothing else about, compare their review counts, not their stars.

The top 10% of shows hold 85% of all listening.

Roughly half of all monthly listening hours flow to the top ~1,500 podcasts; the bottom 17,000 share around 1% of what’s left. Why it happens: attention is a winner-take-most market. Once a show crosses a threshold of cultural visibility — usually around 100,000 monthly listeners — algorithmic recommendations, social mentions, and word-of-mouth all compound. Shows below that threshold rarely break out without a dedicated marketing push or a host who is already famous from somewhere else.

Every top-10 show is on YouTube. Few non-YouTube shows scale.

A 100% YouTube presence in the top 10 is not a coincidence. Why it happens: in 2026, video clipping — YouTube Shorts, Reels, TikTok — is the dominant podcast discovery mechanism. A show that doesn't film cannot supply the clips, cannot be embedded in a thumbnail-driven feed, and cannot be discovered by anyone who isn't already searching for podcasts. For a video podcast, audio-only distribution is now a structural disadvantage.

Mixed-gender shows out-listen single-gender ones by ~24%.

The data block does not contain median-listener figures broken out by host gender configuration (only counts), so we do not quote a specific median premium here. Where the data does let us measure audience size by composition, the clear pattern is in the audience-skew cut (§03): shows with a clear audience lean — either male or female — reach a median of 2.5K monthly listeners, while balanced-audience shows trail at 1.5K. Why it might happen: a clearly-defined community is easier to grow than an undifferentiated one. Two-host conversations may also broaden tonal range, but we do not have the host-level medians to quote a specific premium.

Endurance is rewarded; quitting is the dominant failure mode.

Shows under 25 episodes have a median of 200 monthly listeners. Shows past 500 episodes have a median of 30,000 — a 150× difference. Why it happens: back catalogs become discoverable assets. After year three, a show is being found through old episodes, search, and recommendation as much as through new releases. Most podcasts never reach that point: the typical show in our dataset has a median of 125 episodes total (mean: ~225, pulled higher by long-running franchises). For most creators, sustaining publishing past 250 episodes is the harder, less-glamorous half of the work that actually compounds audience.

Long-form wins the long game.

Episodes of 90 minutes or longer post the highest median listenership at 18,760 — ahead of every shorter format. Why it happens: long-form video has become the prestige format on YouTube, with watchtime weighted heavily by the algorithm. A 90-minute interview generates 4× the watchtime per session of a 22-minute commute show, and watchtime is what the recommendation system optimizes. For audio-first listeners, longer episodes also carry a "habit" advantage — weekly long-form replaces a TV show, where short-form competes with social media for attention.

Sponsorship reflects audience size — not the other way around.

Sponsored shows reach ~7× more median listeners than non-sponsored ones (10K vs 1.5K, a +567% lift), but their average ratings are identical. Why it happens: brands underwrite scale, not quality. A small show with a 4.9 rating and 5,000 listeners is far less likely to attract sponsors than a mid-tier show with a 4.4 rating and 50,000 listeners. The widely-shared advice that "you need a tight pitch deck to win sponsors" is mostly downstream of the simpler truth that you need an audience first.

/ 09  Try It

Score your podcast
against the Top-10 playbook.

The data identifies a handful of strategic choices that consistently separate top-decile shows from the rest of the catalog. Toggle the choices you're making (or considering) below — we'll score them against the Top-10 traits.

Podcast Growth Scorecard

Each choice maps to an evidence-backed trait of high-audience video podcasts in our dataset. The point weights reflect how strongly each trait correlates with listener volume in the Top-10 cohort. Maximum score: 100.

  • My show publishes on YouTube Definitional — 100% of Top 10 are on YouTube.
    +20 pts
  • I also publish on Apple and Spotify Triple-presence reaches 1.24× the median audience of single-audio shows.
    +15 pts
  • My show runs guest / interview formats 10 of the Top 10 are guest-driven; the format dominates the high end of the market.
    +15 pts
  • My typical episode runs 60+ minutes 90m+ shows median ~6.5K listeners vs. ~1K for sub-20m formats.
    +15 pts
  • I have 250+ episodes in the catalog 500+ episode shows median 23K+ listeners — 4.4× under-25 shows.
    +15 pts
  • I publish on a weekly (or more frequent) cadence Consistent weekly publishing maps strongly onto top-decile shows.
    +10 pts
  • My show has at least one paid sponsor Sponsorship correlates with audience size (effect, not cause).
    +5 pts
  • My show has 1,000+ Apple reviews Review count is a strong proxy for listenership (r ≈ 0.84).
    +5 pts
Your Score
0/100
Baseline
Toggle any strategic choice on the left to begin scoring.
Disclaimer: The scorecard is a directional self-assessment, not a forecast. It reflects associations observed in this dataset between strategic choices and audience volume; it does not predict the size or growth rate of any individual show. Several Top-10 podcasts succeed despite missing one or two of these traits, and many shows that tick every box remain in the long tail. Strategy and execution still matter most.
/ 10  Conclusion

The market
has matured.

2026's video podcast ecosystem is no longer a creator frontier. It is a settled media format with established economics, a clear distribution stack, and a textbook power-law audience curve.

For most of the past decade, "the future of podcasting" has been treated as an open question. The data in this report suggests it is largely a closed one. Multi-platform distribution — Apple, Spotify, and YouTube together — is now the floor, not the ceiling. Long-form, video-native, guest-driven content dominates the top of the market. Sponsorship arrives after audience, not before it. Reviews are a mirror; ratings are noise; production format is the strongest predictor of scale.

What’s striking is how little of this matches the advice that circulated five years ago. Short, snackable episodes were sold as the format of the future; they aren’t. Audio purity was supposed to be a moat; it isn’t. Independent creators were supposed to displace networks; the picture is mixed — only 3 of the top 10 shows in this report are formally network-backed (Call Her Daddy / SiriusXM, The Daily / The New York Times, Shawn Ryan / Cumulus), while the remaining seven operate as independent productions or under the host’s own banner. The ecosystem has converged on a media-company logic regardless of who owns the show: ship long-form video on every platform, build a personality-led franchise, and earn the audience over hundreds of episodes.

The outlook

Three trends are likely to define the next 24 months.

1. Consolidation continues. The top 1% already commands roughly 42% of listening, and infrastructure (networks, ad-tech, AI editing tools) keeps lowering the floor for major franchises while raising the ceiling. Expect the top 10 to absorb still more share.

2. The middle gets squeezed. Shows in the 10K–100K listener band — the largest slice of our dataset at 4,460 podcasts — sit in an awkward zone: too big to operate as hobbies, too small to attract major brand budgets. The next wave of consolidation will likely come from networks acquiring or signing this middle.

3. Format experimentation moves to the long tail. Innovative formats — AI co-hosts, real-time call-in shows, interactive video podcasts — will not appear at the top, where risk aversion is highest. They will emerge from the long tail, where 8,000+ shows are competing for differentiation.

Strategic implications

For creators: prioritize endurance over ambition, video over audio, distribution over polish. Most podcasts that fail simply stopped publishing.

For brands and advertisers: review counts are a more reliable audience proxy than star ratings. Negotiate against reviews per dollar, not stars.

For networks and platforms: the middle of the market is the strategic prize, not the top. The top is already locked.

For analysts and researchers: a podcast is now a multi-platform media object, not an audio file. Single-platform analysis materially understates audience.

/ 11  Methodology

How this
report was built.

Every figure in this document is derived from a large-scale podcast intelligence database covering thousands of English-language video shows and millions of underlying signals. The notes below describe the analytical framework without exposing the raw underlying mechanics.

/01What counts as a podcast+

For the purposes of this report, a podcast is defined as a unique show with a stable identity across distribution platforms. Multiple host listings, contributor entries, or distribution endpoints belonging to the same show are consolidated into a single show-level record before analysis begins.

All audience, performance, and content metrics are reported at the show level — not at the level of individual contributors, episodes, or platform listings. This is intended to prevent the inflation of any one show's footprint and to make comparisons across the dataset meaningful.

/02Video podcast definition+

"Video podcast" in this report refers to any show whose intelligence record includes confirmed distribution on a major video platform (most commonly YouTube), in addition to its audio distribution.

This is a definition of distribution format, not of audience. A show that publishes a video version on YouTube is included even if the majority of its audience consumes the audio feed on Apple or Spotify. Conversely, a video-only show with no audio feed would not appear in this dataset, which is anchored to the broader podcast ecosystem.

Catalog scope. This edition of the report covers approximately 34,000 active English-language video podcasts. The dataset is restricted to shows that have published at least one new episode within the last twelve months — activity is a filter in this edition, and dormant shows are excluded from every chart, percentage, and cohort comparison. The catalog spans the full active spectrum, from a handful of shows clearing multiple million monthly listeners down through the long tail of small but live productions.

/03Guest format classification+

A show is classified as a "guest format" when intelligence signals indicate that it regularly features guest conversations as a core editorial format. This is a judgment about the show's overall tendency, not a guarantee that every episode includes a guest.

Solo monologue shows, scripted narrative shows, and panel shows with stable cast rotations are not classified as guest formats, even if they occasionally feature outside contributors.

/04Sponsorship classification+

A show is classified as having sponsorship when there is evidence of brand partnership or advertising activity associated with it — including ad-read transcripts, sponsor mentions, network ad-deal participation, or platform-side advertising data.

This indicates the presence of sponsorship at some point in the show's run. It does not imply continuous monetization, current ad rates, or revenue figures, none of which are reported in this analysis.

/05How listener estimates are derived+

Monthly listener figures presented in this report are modeled estimates, not directly reported audience numbers. They are derived from a combination of observable signals, including audience engagement (such as review activity), publishing consistency over recent periods, and the overall depth of a podcast's content library.

Podcasts that demonstrate stronger engagement, release episodes more consistently, and have a larger catalog of content are generally modeled to have a broader listener base. In contrast, shows with limited activity or lower audience interaction are estimated more conservatively.

To ensure consistency across the dataset, these signals are calibrated into a unified estimation framework, allowing for meaningful comparison between podcasts even when direct listener data is not publicly available.

These figures should be interpreted as directional indicators of relative audience scale, designed for benchmarking and trend analysis rather than precise measurement. Differences across broad audience tiers are meaningful, while small differences between similarly sized podcasts should not be over-interpreted.

/06Data preparation & normalization+

Before analysis, the underlying records are consolidated so that each podcast appears exactly once. Performance metrics are standardized to prevent double-counting that can occur when a single show has multiple listings. Missing or incomplete fields are handled with consistent normalization logic, and extreme outliers are controlled using percentile-based caps where they would otherwise distort group averages.

Where this report cites averages, both the mean and the median are typically reported when they diverge meaningfully — a divergence common in highly skewed audience distributions like podcasting.

/07Limitations+

Readers should be aware of the following limitations:

Language scope. The dataset is restricted to English-language podcasts. Non-English shows are out of scope, and country-level breakdowns reflect English-language production only — not total podcast activity in any given region.

Modeled audience data. Listener figures are estimates, not directly reported counts. Comparisons should be treated as relative rather than absolute.

Coverage gaps. Some shows may have incomplete network or platform metadata. Where coverage is partial, the analysis explicitly excludes missing values rather than imputing them.

Network affiliation undercount. Network affiliation is detected from public network branding signals exposed in the show-level metadata our analysis ingests. Some major networks (e.g., Wondery, NPR, BBC, PRX, Pushkin, SiriusXM) may be undercounted because their network branding is not consistently surfaced at the show metadata level — even though their shows do appear in the catalog. The headline figure of ~4.3% network-affiliated should be read as a floor on network involvement, not a ceiling. The Top 10 list itself credits SiriusXM (Call Her Daddy), The New York Times (The Daily), and Cumulus (Shawn Ryan), all of which are present in the data; the broader undercount applies to long-tail catalog detection, not to the named Top 10 entries.

Show-level granularity. Analysis is conducted at the show level. Episode-level performance, segment-level engagement, and listener demographic data are not part of this report.

Temporal scope. The dataset is a snapshot of the podcast ecosystem at the time of analysis. Findings should be read as a description of the present moment, not a prediction.

/08Data freshness+

This report reflects a snapshot of the English-language video podcast ecosystem as of May 2026. Audience figures, platform footprints, sponsorship status, and ratings evolve continuously. The findings here describe stable structural patterns that are unlikely to change month-to-month, but specific numbers will drift over time.

Where decisions depend on current values (for example, whether a specific show currently runs sponsorships), readers should verify against live sources before acting.

/09How to cite+

If you reference this report in articles, research, presentations, or commentary, the suggested citation format is:

Source: MillionPodcasts Research — The State of Video Podcasting, 2026. Snapshot: May 2026.

Direct citations and links back to this report are encouraged. Excerpts of up to a paragraph may be quoted with attribution; longer sections should be summarized in the citing publication's own words.

/10Interpretation note+

The findings in this report are analytical and directional. They are designed to identify patterns, surface non-obvious relationships, and inform strategic thinking about the podcast ecosystem. They are not intended as exact measurement of the entire podcast industry, nor as audited audience figures for any individual show.

Used appropriately, this kind of intelligence is best treated as a lens for asking sharper questions — about which formats are working, which markets are saturating, where the structural advantages are concentrating — rather than as a definitive ledger of who is listening to what.

/11How podcasts were deduplicated to "unique shows"+

Many podcasts appear multiple times in raw feeds — through re-uploads, regional editions, host-side and network-side listings, RSS migrations, or platform-specific re-syndication. To produce a clean view of the market, the underlying records are collapsed so that each podcast appears exactly once.

The deduplication rule works in two passes. Each show is first matched on its Feed ID, which uniquely identifies an RSS feed across the ecosystem. Where Feed ID is missing or ambiguous, the analysis falls back to title-based matching with light normalisation (case-folding, punctuation harmonisation). The most-recent metadata wins per field, except for listener and review counts where the maximum observed value is taken — this prevents duplicate listings from accidentally lowering a show's apparent reach.

This is why the report describes a universe of approx. 34,000 unique English-language video podcasts rather than the 55,859 raw rows in the underlying intelligence feed. Every chart, percentage, and ranking in the report is calculated at the show level, not the row level. Critically, listener counts are never summed across rows for the same podcast — we take the maximum reported value, so a podcast with 100,000 listeners listed across multiple rows still counts as 100,000, not as a multiple of that.

/12How the Top 10 was selected+

The Top 10 ranking in this report is determined strictly by modeled monthly listenership at the snapshot date. It is not a measure of historical reach, lifetime downloads, awards won, cultural footprint, critical reception, or revenue.

The practical implication is that a show that was massive in 2022 but has lost audience since will not appear, while a show that has only recently crossed the multi-million-listener threshold will. This is a snapshot of the present-tense audience leaderboard, not an all-time-greats list.

Readers comparing this list to other industry rankings should expect divergences. Different reports rank by different signals — downloads, ad revenue, social influence, chart positions — and each method produces a slightly different top-ten. The ranking here is consistent within itself: every show in the Top 10 is ranked by the same modeled-listener number used everywhere else in the report.

/13What's outside the scope of this report+

To set fair expectations, the following are explicitly not covered by this analysis:

Video-only YouTube channels that do not ship a paired audio feed. These can be very large — some of the biggest creators on the platform — but they are not classified as podcasts under the report's working definition and therefore do not appear in any chart, table, or ranking.

Private or internal corporate podcasts, employee communications, and shows distributed only behind paywalls or login walls. Anything that doesn't have a public RSS or platform listing is invisible to the underlying intelligence.

Non-English shows. The report is restricted to English-language podcasts. Country-level breakdowns therefore reflect English-language production only, not total podcast activity in any given region. A low number for Germany or Japan does not mean those markets are small — it means few of their shows are produced in English.

Podcasts under approximately 500 monthly listeners. Shows below this threshold typically fall beneath most platform discovery surfaces and cannot be reliably modeled. They exist, but they aren't measurable with the signals available here.

Revenue, advertising rates, and sponsor dollar values. The report classifies whether a show runs sponsorships and uses that as a binary indicator; it does not estimate how much money any show earns. Pricing data, CPMs, and deal economics are out of scope.

Listener-side location data. The country and region fields in the dataset describe where the show is produced, not where its listeners live. The report does not break out listenership by listener geography.

/14How demographic estimates are derived+

The audience composition figures in this report — gender split, income tier, generational mix — are modeled estimates, not directly reported survey data. They are calibrated for each show using a combination of observable signals: content classification, host attributes, audience review patterns, social-platform follower composition, and category-level priors.

Each demographic field is expressed as a percentage of that show’s audience. Where multiple identifiable groups are tracked (for example, gender), the values do not need to sum to 100% across the row — the dataset also includes a small "other" category that is not always modelled separately. Where a demographic field is unobservable for a given show, the field is left empty rather than imputed.

Dataset-wide demographic figures (such as the female-share or Millennial-share headlines) are calculated as listener-weighted averages: each show’s percentages are weighted by its modeled monthly listener count before aggregation. This prevents a long tail of small shows with skewed audiences from distorting the headline figure.

These figures should be read as directional indicators of audience composition. Differences across broad cohorts (a 60% female-leaning category vs. a 40% female-leaning one) are meaningful. Differences of one or two percentage points between similar shows should not be over-interpreted.

How the percentages add up. For each podcast in this report, the income tiers (Low, Medium, High) always sum to 100%, and the age groups (Gen Alpha through Silent Generation) also always sum to 100%. Where a podcast’s raw data only covered some of the cohorts, the present cohorts were renormalized so the row totals correctly — this guarantees all percentages can be read as straightforward shares of that show’s audience without needing to check whether they’re missing a category.

Important limitation — classifier concentration. When we ask "which income tier dominates each show?" or "which generation dominates each show?" the underlying classifier tends to default to the modal cohort. In this dataset, that means Medium income comes back as the dominant tier for roughly 96% of shows, and Millennials come back as the dominant generation for roughly 82%. Read the dominant-tier counts as directional — they tell you which cohort is most often slightly ahead, not that 96% of shows are exclusively middle-income or that 82% are exclusively Millennial. The listener-weighted percentages cited in the headline cards (Medium income share = 59%, Millennial share = 49%) are more reliable because they preserve the underlying mix rather than collapsing each show to a single label.

/15How sponsor data is captured+

Two sponsor fields are used in this report: a binary sponsorship indicator (whether a show is known to carry advertising), and a unique sponsor count (how many distinct brands have been observed running ads on the show).

The sponsor count is derived from observable advertising activity across recent episodes — brand mentions, promo codes, and host-read ad slots are detected and de-duplicated within the show. The count reflects unique sponsor brands, not the number of ad reads or campaign cycles. A show running the same advertiser across 30 episodes is counted as one sponsor; a show with three different advertisers in a single episode is counted as three.

The "Top Sponsors" ranking counts the number of distinct shows each brand appears on, sorted in descending order. A brand that runs a heavy multi-episode buy on a single show contributes one to its appearance count; a brand running one ad each on twenty different shows contributes twenty. This favours brands with a broad sponsorship strategy over brands with a deep but narrow one.

Two coverage tiers. Roughly 22.9% of shows in the catalog (about 7,800 podcasts) carry confirmed sponsorship activity according to the binary indicator. Of these, only ~6,300 (18.4% of the catalog) have detailed brand records — the remaining ~1,500 are confirmed sponsored without further detail. All per-show density statistics (mean unique sponsors, median unique sponsors, brand frequency rankings) are calculated against the 18.4% subset only, and are labelled as such throughout the report.

/16How listener counts compare to other industry sources+

The monthly listener figures in this report are modeled estimates derived from a multi-signal model (audience engagement, publishing consistency, content depth, platform-side indicators). They are not measured downloads, and they are not survey-based unique-listener counts.

Two practical consequences follow. First, the listener estimates use a tiered model: the engine snaps individual show estimates to common values (1,000; 2,500; 9,000; 35,000; 100,000 and so on) rather than producing a continuous numeric distribution. This is a property of how the underlying intelligence platform models audience size. Treat each listener number as a band, not a precise count.

Second, third-party sources may report different absolute figures for top-tier shows. Edison Podcast Metrics uses a survey-based panel that estimates unique monthly listeners; its figures for very large shows are typically lower than the modeled multi-platform reach reported here. Podscribe and similar adtech platforms measure ad-impression delivery and can produce different figures again. Where this report’s number for a Top 10 show differs from a published Edison or Podscribe figure, the difference reflects methodological choices about what counts as a "listener" (panel-based unique humans vs. modeled multi-platform reach), not an error in either source.

For internal comparisons within this report — one show vs another, one cohort vs another, one bucket vs another — the figures are consistent because they all come from the same model. For external comparisons against panel-based or impression-based sources, expect the absolute numbers to differ even when the relative orderings agree.