Podcast analytics: matching each count to a decision

What to take away

This article covers podcast-business practice, not individualized tax, legal, copyright, or advertising advice. Duties depend on your content, guests, sponsors, and state. Confirm specifics with a CPA, a media attorney, or the licensing body that governs the use.

The numbers a show actually runs on

A podcast has four measurable layers, and mixing them is the most common analytics mistake. Reach is how many people received the file. Engagement is what they did with it.

Revenue is what the show earned. Cost is what it took to produce.

A show can grow reach while revenue falls, because a new platform pays less per play than the old one.

Downloads are not listeners. One listener can produce several downloads across devices, and a download can come from a bot or a prefetch. Apple and Spotify publish their own definitions of a play or a download, and those definitions have changed over time. Quote the platform's current term when you report a number, not a remembered one.

Know which dashboard each number comes from. Apple's figures come from Apple Podcasts Connect, Spotify's from its own creator dashboard, and YouTube's from YouTube Studio; downloads come from your hosting provider's statistics, such as the dashboards in Buzzsprout, Libsyn or Podbean. Hosting downloads are the count most sponsors ask for, and the industry reference for counting them is the IAB Podcast Measurement Technical Guidelines, which several hosts and measurement services state they follow. When a sponsor asks for downloads, ask whether they mean an IAB-certified count and say which one you are sending.

Retention is the closest thing to a quality signal you get for free. Most hosting dashboards show a consumption curve: where listeners drop off inside an episode. A drop at the two-minute mark usually means the intro ran long. A drop at a sponsor read usually means the read was placed badly or ran too long.

Editing and production time as a denominator

Track hours per episode, not just cost per episode. A weekly interview show often spends more hours in editing than in recording, especially when the guest audio arrives at a different level than the host track. Once you know your hours, revenue per production hour becomes the number that decides whether a format is worth keeping.

The editing suite you choose changes that number. Descript transcribes and lets you cut by editing text, which suits interview and narrative shows.

Adobe Audition and Reaper suit hosts who want manual control over compression, noise reduction, and loudness. Riverside and SquadCast record remote guests locally, which cuts the repair time that bad internet audio creates.

None of these is free of a learning curve, and the curve is a real cost in the first month.

Loudness is a delivery decision, not a preference. Podcast platforms publish target loudness guidance, and a file mastered far from it gets turned down or up by the player, which changes how the show feels next to the one before it. Set your export preset once and stop relitigating it every episode.

Platform economics, read statement by statement

Apple Podcasts does not pay per download. It distributes your RSS feed and reports analytics, and any money comes from your own sponsors or subscriptions.

Spotify runs its own ad program and pays on its own terms, which differ from a host-read deal you sell directly.

YouTube pays through the Partner Program once a channel meets its published thresholds. For a long-form video podcast the relevant route is 1,000 subscribers with 4,000 qualified watch hours in the last 12 months; watch hours from Shorts views in the Shorts Feed do not count towards that 4,000. So the YouTube count that decides whether the video version can earn at all is watch hours, not downloads, and its watch-time model rewards a different edit than an audio-only feed does.

That difference is why the same episode can earn three different amounts. A video version of a show may earn on YouTube while the audio feed earns nothing directly. A host-read sponsor pays a flat rate regardless of platform. A programmatic ad network pays per impression, and impressions are counted by the network, not by you.

Read each statement for the same three things: the unit being paid on, the reporting window, and the payment date. A network that pays sixty days after the month closes is not the same business as one that pays on the fifteenth. Cash-flow planning depends on that gap, not on the headline rate.

Sponsorship rates and the disclosure you owe

Host-read ads are usually priced on a CPM basis, meaning a rate per thousand downloads or impressions. The rate you can charge depends on your audience, your category, and how the sponsor measures the result. Do not copy another show's rate card; it reflects a different audience and a different measurement.

When you sell a read, the disclosure is not optional. The FTC guidance on endorsements, influencers, and reviews explains that a material connection between the endorser and the advertiser must be disclosed clearly.

The rules behind that guidance are the FTC's Guides Concerning the Use of Endorsements and Testimonials in Advertising, 16 CFR Part 255. That means both sides remain responsible for claims that are truthful and not misleading. In a podcast that usually means a spoken disclosure the listener hears before the claim, not a line buried in the show notes.

Music is a separate rights question from sponsorship. The Copyright Office study of copyright and the music marketplace sets out that public performance and other uses can travel through different rights and licensing paths.

If your theme music or beds are commercial recordings, confirm the license with the rights holder or a performance rights organization such as ASCAP or BMI before the episode publishes.

Records, cash flow, and what to keep

Keep records that show income and expenses clearly, and keep the documents behind them. The IRS guidance on records a business should keep describes a system that supports purchases, sales, payroll, and assets.

For a show that means sponsor contracts, platform statements, equipment receipts, and contractor invoices, filed so a CPA can follow them. Whether you file on Schedule C or another form is a question for that CPA.

Cash flow on a small show runs on a lag. Sponsors pay after the campaign reports, platforms pay after their window closes, and editing costs land the week the episode is made.

A simple twelve-week cash forecast, with named variables for expected sponsor payment, platform payout, and weekly production cost, shows the gap before it becomes a problem.

Fill each variable from your own statements and invoices, then compare the forecast with what actually arrived.

Security is part of the same file. Guest releases, sponsor contracts, and unpublished audio are business records, and the NIST small business quick-start guides give a practical starting point for protecting them. Start with access control and backups, which cover most of the risk a one-person studio actually faces.

A worked example: deciding whether to keep a second show

Suppose a host runs a weekly interview show and a monthly solo show. The interview show takes fourteen production hours an episode and earns a host-read sponsor at a flat rate. The solo show takes four hours and earns nothing directly, but it feeds the interview show's guest pipeline.

The decision is not which show has more downloads. Write the arithmetic in named variables: monthly revenue from the interview show, minus monthly production cost for both shows, divided by total production hours.

If the solo show lowers the cost of booking guests on the interview show, that saving belongs in the same equation.

The hours side is already fixed by the example. A month with four interview episodes is 56 hours; a month with five is 70. The solo show adds four, so the denominator is 60 hours in an ordinary month and 74 in a heavy one. Put the month's interview revenue minus both shows' costs over those hours and you have revenue per production hour for the pair. Then run the interview show alone, over 56 or 70 hours: if the pair earns less per hour than the interview show alone, the solo show has to justify its four hours through the guest pipeline, not through its own downloads.

Run the numbers for one ordinary month and one heavy month, then decide whether the second show earns its hours.

Common questions

Why do my download numbers differ between platforms?

Each platform counts a play or a download under its own published definition, and each reports on its own delay. Apple, Spotify, and your hosting dashboard will rarely match exactly. Pick one source as your reporting standard, name it in your notes, and compare that source against itself over time.

Can I use a commercial song in my intro?

Not on the strength of a streaming subscription. The Copyright Office study linked above explains that music uses can involve separate rights and licensing paths. Confirm the specific use with the rights holder or a performance rights organization before the episode goes out.

Do I have to disclose a sponsor read?

Yes. The FTC guidance says a material connection between an endorser and an advertiser needs clear disclosure, and both sides are responsible for truthful claims. A spoken line before the read is the usual practice. Confirm the current requirement with the FTC guidance or a media attorney.

How often should I review these numbers?

Monthly is enough for most independent shows. Review downloads, retention, revenue, and production hours together, since a change in one usually explains a change in another. Reopen the review when you change format, platform, sponsor, or editing workflow, because a silent change in definition makes every earlier comparison useless.