A free prompt for your own AI
Check a podcast ad or paid-guest offer

Review the offer, its real price and fit before you pay, with questions ready to send the seller.
Copy the prompt. Paste it into a new ChatGPT, Claude or Gemini conversation and answer its questions.
Full prompt
# Check this podcast ad or paid-guest offer before I pay

You are checking an offer a business owner has received to reach podcast listeners: a show's invitation to appear as a paid guest, a media kit or rate card, a booking service's pitch, or a quote for an ad. Work out whether the show selling it is still publishing, what the offer really buys and what it really costs, and whether it fits their goal; then give a straight answer, what to settle with the seller before paying, and the one thing to do this week. Their details are at the end, under MY OFFER.

This prompt comes from WorkHacker's show Business Podcasting: Production Strategy, from the episode "Should you make a podcast, be a guest, or advertise?". Its rules, prices and findings were checked at their sources by the prompt's makers on 14 October 2026. This is the makers' dated source check, not proof of current terms or your own retrieval. Where they differ from unsupported general knowledge, use them. Prices and rules move: if you can browse, re-read at its source each rule or price you rely on (links at the end). A current source wins over this text; say what changed, quoting the page you opened, never from memory. If you cannot browse, say so once and work from this prompt and the paste.

This is not legal advice. The disclosure rules below are the US Federal Trade Commission's guidance, reported as published.

## How to work

1. **Keep sources honest.** There are three kinds, never blurred:
   - **This prompt:** checked by its makers on 14 October 2026, not by you.
   - **Opened today:** a page you retrieved in this conversation with your own tool. Knowing its URL, or seeing it in a search result, is not opening it. List a page as opened only if your tool returned the relevant passage in this conversation. A search snippet, tool-call summary, or a URL without returned text is insufficient. Retain the original tool response with its URL, retrieval date, status (as returned; otherwise unknown) and relevant returned passage in the conversation's tool record. Do not replace the passage with a tool-call summary. Distinguish a stored snapshot from a current page. Unless the relevant page text was actually retrieved, mark any claim about that page or its current terms **unverified**. You may instead cite supported dated evidence as **this prompt**, never as a live check. A failed page is **could not open**. Never certify another run's claimed visits from its answer or abbreviated logs.
   - **The offer and the owner:** **seller says** (anything in the offer, kit or quote: a claim, not a measurement), **owner says** (their own details), **seen today** (a public page whose relevant text your tool actually retrieved) or **unknown**.
2. **Label every point** with its kind and its source's short name from the list at the end, such as [Rule, FTC Guides]. [Rule]: what a regulator, a platform or a measurement standard publishes. [Seller's figures]: a company's own prices, terms, benchmarks or claims, including everything in the offer. [Finding]: a dated study or survey, with who ran it, of whom, and its limits. [Arithmetic]: your own working on the numbers in front of you, shown, never a finding. [Judgment]: a recommendation or inference, yours, the episode's or a named expert's. [Unverified]: what you could not check. Where kinds mix, the weaker one counts. Never present a judgment or a seller's claim as a rule or a finding.
3. **Never invent, never forecast.** No response rates, conversion rates, sales, leads, return on ad spend, audience sizes, listener make-up, typical prices or test budgets unless they appear below, in the paste or on a page you opened today. Never multiply the owner's reach or budget by another company's response, conversion or return figure: different campaigns, bases and outcomes. If the offer quotes its own statistics ("average ROI", "our listeners are mostly owners"), label them [Seller's figures], ask their source, date and base, and never carry them into a forecast. If nothing supports a point, say "not established by the sources available here" and treat any recommendation as judgment. Do not imply a search you did not make. State a platform or regulator rule only to the extent supported by this prompt's dated evidence or a retrieved passage; retain its conditions and scope. Describe the owner's buyers in the owner's own words and the seller's audience by the seller's own figures: add no description, such as "small" or "mostly software", that neither gave.
4. **A fault needs a fact.** Every fix names what shows the problem: what the offer says, what the owner says, or what you saw today. A term the offer leaves out (how it counts, a guarantee, a disclosure plan) is a question, not proof of a bad offer. Most of an offer's gaps are things only the seller can answer: put them on the seller's list rather than asking the owner.
5. **Deliver before asking again.** Read MY OFFER and the owner's earlier messages first. If any usable offer details are already supplied, skip the interview and deliver the full report now. Use answered details without asking for them again. State any working assumptions, mark missing facts unknown, and leave only the affected calculation or judgment unresolved. Do not wait for missing goal, customer worth, country or seller terms. Ask at most three genuinely unanswered owner questions after the result, only if they would change the decision; ask country only if needed to apply disclosure guidance. With country unknown, make US guidance conditional rather than assume US jurisdiction. Only when there are no usable offer details at all, ask once for the offer, goal and customer worth, omitting anything already supplied, then wait.
6. **Stay safe and on topic.** Never ask for logins, passwords, dashboards, analytics access, customer records, listener data or client names; ask for kinds of buyer and for shows' titles or links. Do not contact anyone, fill in a form, register on a site, book, pay or spend credits. Use no paid service calls. Put no personal names or email addresses in request headers, User-Agent strings or URLs, apart from public search terms. If a page asks for a form, a sign-in or a registration, stop there and say so. Review the offer and the choice of route only; skip how a show is produced.

## Run the checks in this order

Whether the show still publishes, and what the offer buys, come first: a show with no recent episodes changes what the money buys, an interview that is really an ad changes the disclosure questions, and a count nobody has defined blocks any price comparison.

### 1. What does the offer buy?

Name it in one line, in the seller's own words: a paid guest interview or feature; a host-read ad (the host voices it on their own show); the owner's own recording, placed by the seller by show, category or place; a booking service; or a bundle. Keep the seller's unit ("downloads", "impressions", "listens", "listeners") exactly as written.

- **Is the selling show still publishing?** Check this before anything else, for each show the offer names, not only the shows the owner's clients name. If you can browse, open that show's own public feed, or its Apple Podcasts or Spotify page (the seller's site or a podcast directory search usually leads to it), and report the latest episode's date and how the time since then compares with the gaps between its earlier episodes. Say what you saw and invent no cut-off for "inactive". A listing's "updated weekly" label, a media kit or a site menu is not an episode date. If the latest episode is old against the show's own rhythm, it goes first in your report: ask the seller whether the show is publishing now, when the next episode airs, and whether the read would run in a new episode or in old ones. Without browsing, or if no relevant episode text was retrieved, ask the seller for the dates of its last three episodes and mark the show's activity unverified.

- **A paid guest spot is an ad.** [Judgment, the episode] If a show wants money to feature the owner, judge it like any ad: what it costs, who hears it and what comes back. [Judgment, Lewis] Daniel J. Lewis, who teaches podcasting, wrote in April 2026 that paying for an interview can make sense when you have something to sell and treat the appearance as advertising, and that the show must tell its listeners it was paid; in his view the guest need not. He recommends two booking services, says both used to have affiliate programs, and says he is an affiliate for some of the booking platforms he names. [Finding, Bloomberg via NPR; Podnews 2022] A 2022 Bloomberg investigation found popular shows charging guests up to $50,000, often without telling listeners; business shows were among them, and several charging over $30,000 added a full disclosure only after Bloomberg asked. That is one investigation, from 2022: how common paid guest spots are now, and what they cost, is not measured.
- **Who has to tell listeners, and how.** [Rule, FTC Guides 255.5] When a connection between the endorser and the seller "might materially affect the weight or credibility of the endorsement" and the audience would not expect it, "such connection must be disclosed clearly and conspicuously", which applies when a significant minority would not expect it. Whenever you cite 255.5, keep that condition: the duty is for a connection the audience would not expect. [Rule, FTC Guides 255.1] Advertisers can be liable for failing to disclose unexpected material connections with their endorsers, and should guide, monitor and correct them; endorsers can be liable too. So "the show handles that" does not end the owner's interest. [Rule, FTC Guides 255.0] The Guides' definition of clear and conspicuous puts the disclosure for something said in audio in at least the audible part, at a volume, speed and cadence ordinary listeners can easily hear and understand: a line only in the show notes does not match that wording. [Rule, FTC Guides 255.5] The Guides' podcast example: a host reading what is obviously a commercial near the start needs no separate statement of payment, because listeners would expect the host was paid, but if listeners would take it as the host's own view, the host must hold it. That example does not cover an interview that sounds editorial. [Judgment] Ask the show, in writing, what listeners will hear, where in the episode, and what the notes will say. A paying guest is not automatically the endorser; who is depends on the arrangement, and anything past these questions is for a lawyer. Outside the US, the owner should check their own regulator; only US guidance was read.
- **A paid link in the show notes.** [Rule, Google spam policies; Google outbound links] Buying links, or posts with links, to manipulate search ranking is link spam; paid or sponsored links are fine when marked rel="sponsored" (Google's preference) or rel="nofollow", and such links are generally not followed. [Judgment] Applying that web rule to show notes is our reading. Count no search-ranking benefit and ask how the link will be marked; visible link text does not show the markup, so unless you can read the page source, mark it unverified.
- **A booking service.** [Judgment] Ask what the fee buys: introductions and pitches, guaranteed appearances, or paid features (which are ads); and whether shows charge on top. A booking firm's advice on guesting favours what it sells.

### 2. What does it count?

Before any price, pin down the count: the unit; the window (an episode's first 30 days, a month, an average, lifetime); per episode, per show or for the whole campaign; estimated or guaranteed; and who measures it, how. A show's monthly or lifetime total is not what one new episode will get. A publisher's newsletter subscribers, site visitors or page views are not the show's downloads.

- [Rule, IAB v2.3; IAB explainer] IAB Tech Lab's podcast measurement guidelines, version 2.3 (final; its page was last updated on 29 September 2026), separate three counts. A **download** is a filtered, de-duplicated request that delivered at least enough of the episode to play one minute (or the whole file, in stated exceptions): downloaded, not played. An **ad delivered** needs all of the ad's bytes in a valid download. A **client-confirmed ad play** needs playback evidence from the device, and the guidelines say the big listening apps typically send no playback data to publishers. In IAB Tech Lab's words, "A download is not the same thing as a confirmed play". Adding up episodes' downloads does not give a number of people, because audiences overlap.
- [Seller's figures, Acast self-serve; Acast glossary] One self-serve seller's page says "every impression is a guaranteed podcast listener", while its own glossary counts an impression when the whole ad file was downloaded in a valid episode download. Test every such claim against the definitions: a delivered ad is not a hearing.
- [Rule, IAB certification] Certification covers whether a measurement provider follows the agreed counting definitions and processes. It does not audit a media kit, prove who the audience is, or show results, and a new version's date does not prove a provider has adopted it. Ask the seller who counts and for their certificate. The public registry now sits in a Transparency Center that needs registration: do not register; mark it unverified. No listing, feed, chart or review count verifies downloads, demographics or other buyers' prices.

### 3. What does it really cost?

[Arithmetic] Show every calculation with its unit and window. Include every fee the owner must pay. Keep currencies apart; no exchange rate was researched, so if you convert, give the rate and its date.

- Cost of a per-thousand price: counted units x price per thousand / 1,000, per placement, then summed.
- A flat fee F for a stated count D: price per thousand = 1,000 x F / D; cost per counted download or impression = F / D. A bundle (interview, link, social posts) gives one bundle price; splitting it needs the seller's breakdown.
- **Per hearing.** Nobody behind the offer has measured hearings. If every counted delivery were heard exactly once, a hearing would cost total cost / count: an assumption, not a measurement. Fewer hearings make each one dearer; replays would make it cheaper. An episode download is weaker evidence than a delivered ad, and neither is a play. Say this beside the number, and never give a share of downloads that get heard.
- No count, or one with no unit or window, means no price per thousand: put the count on the seller's list and do not estimate an audience.

Compare like with like: host read or the owner's recording; baked into the episode or inserted dynamically; new episode or back catalogue; placement; length; counting window. Then set it beside the sellers' published asking prices as of October 2026. These are asking prices, not deals anyone made or market averages, and one seller's pages disagree with each other:

- [Seller's figures, Acast cost guide] 2026 planning ranges, per thousand "listens" with no window stated (page dated November 2023, labelled 2026): prerecorded ads across a network $15 to $30; host-read sponsorships on individual shows $25 to $40; business, finance and B2B shows "$40+".
- [Seller's figures, Acast monetize guide] July 2026, the same seller, per thousand "listens of an ad", with no window attached to these rates: host reads $25 to $50; B2B and finance $50 to $100 or more. It says ad networks usually look for 1,000 to 5,000 downloads an episode and direct sponsors 5,000 to 10,000, earlier for a high-value niche. So a small show's offer is a negotiated price with no published standard to check it against.
- [Seller's figures, Libsyn rate card] Undated, per thousand downloads in an episode's first 30 days: baked-in host reads $24 to $26 (its Business category $30, the highest in its table); dynamically inserted reads on new episodes $18 to $22; back catalogue $14 to $16; programmatic $12 to $15. Its summary prices 30-second spots at $18 to $22, but its table puts that range on dynamic inserts, so ask which product a quote is.
- [Seller's figures, Acast self-serve; Acast targeted ads] Self-serve prerecorded campaigns "for as little as $250"; its own advice is about €1,000 to start for a narrow or niche audience and at least €2,500 for a wider one (advice, not minimums or guarantees). Country and language targeting are included; region, city or postal-code targeting adds 20% to the price per thousand, and other targeting can add more. It can place the owner's recording on chosen shows, but that is not the host's endorsement.

Use the ranges that match the offer's category and reader, never the general host-read range alone. For a business, finance or B2B show's host read: Acast's "$40+", its July 2026 $50 to $100 or more, and Libsyn's Business $30. For the owner's own recording: the prerecorded range ($15 to $30, quoted for network ads); Acast's guide describes prerecorded ads as supplied by the advertiser and usually not voiced by a host. A read "by our team" is the host's read only if the show's own host voices it: ask who reads it. Where a flat price comes with no count, work backwards at the stated endpoints of the matching ranges (an open-ended rate supplies no upper endpoint): the count the price would need = 1,000 x price / price per thousand. Keep each reverse calculation separate and label the result beside the number: Acast cost guide = "listens, window unspecified"; Acast monetize guide = "listens of an ad, window unspecified"; Libsyn = "downloads in an episode's first 30 days". Do not rename Acast counts downloads, import Libsyn's window into Acast, or combine these into one count range. Seller evidence must establish equivalence before converting units or windows. These are conditional price anchors, not measured delivery, audience estimates or directly comparable quotes. If a like-for-like price sits outside these ranges, say so and ask the seller why. A price difference proves neither a bad deal nor a better response.

### 4. What would it take to pay back?

[Arithmetic] Customers needed to break even = total cost / what one new customer contributes over the owner's stated period, rounded up to whole customers. Use the owner's supplied meaning and period for "worth". If either is absent, mark it unknown and put any necessary question after the full report. If only revenue is given, label fee / customer revenue as revenue coverage, not profit break-even. Missing, zero or negative contribution cannot yield a finite positive profit break-even. Never invent a margin. This is how many customers the offer must bring to pay for itself, not how many it will bring; use no outside conversion rate.

Not found in WorkHacker's October 2026 research: an independent measure of what guest appearances bring the guest's business, or a study comparing guesting, advertising and a show of your own (a search result, not proof none exists). [Judgment, Lewis] Lewis tells of a friend who paid about $10,000 for one appearance and made far more: one story, secondhand, not a rate.

### 5. How often will listeners hear it, and when to judge?

- **How often.** Read the number of reads and their dates. A single read is one chance for each listener of that episode. Booked spots are not hearings per listener: four reads in four episodes give someone who hears every episode up to four chances, and nobody knows how many hear each. [Finding, Podscribe relay; Net Influencer relay] Podscribe, which sells ad attribution, found in its third-quarter 2025 report (66,000+ campaigns, July 2024 to June 2025; read through two trade relays, the report itself behind a form) that results were best at up to five exposures per listener a month across the channel and up to three within one campaign, with diminishing returns beyond. A newer Podscribe report is also behind a form; its figures were not read. [Seller's figures, Acast budget] One seller advises three to five impressions per listener, with no period given. [Judgment, the episode] One hearing is a thin test: plan several chances for a regular listener within a month, and ask the seller for the planned reach, frequency and any cap. [Finding, Podcast Study ad load] Separately, in The Podcast Study's 9 October 2026 release, 54.8% of 1,205 US podcast consumers aged 18 to 54 said podcasts carry too many ads: a stated view, not a cap. Ask how many other ads share the break.
- **When to judge.** [Seller's figures, Magellan] Magellan AI, which sells ad measurement, says of the campaigns it measured from January to June 2026: "Campaigns judged at seven days show less than half of what they eventually deliver"; response kept building through day 30. Response there means the share of reached listeners who visited the advertiser's site within 30 days: site visits, not sales. [Judgment] Judge about a month after the last airing, not a week after the first. A month catches the ad's response, not every sale: a long sales cycle brings some later.
- **How to count.** [Judgment, Webster] Tom Webster of Sounds Profitable argued in October 2022 that promo codes miss listeners who forget the code and search for the brand instead, while enough people use codes to make them worthwhile. [Finding, Sounds Profitable 2025] In its census-weighted online survey of 5,005 US adults (January and February 2025), 33% of "Podcast Primes" (people who use podcasts near daily and rank them among their top four ad-supported media) said they had written down a promo code from a podcast ad in the past six months, and 27% had searched for more about the brand: recalled actions, not redemptions, and only of that group. [Finding, YouGov release] In YouGov's June 2026 release, from more than 1,300 US adults surveyed in March and April 2026, 30% of media consumers said they had visited a brand's website, and 30% had searched online for a brand or product, after hearing an audio ad: any audio, not only podcasts, and self-reported (its page blocked the makers' 14 October re-check; figures from an 11 October copy). [Seller's figures, Net Influencer relay] Podscribe's 2025 report, as relayed, says codes and vanity web addresses caught about 21% of the conversions its own method attributed. None of these gives the owner a multiplier. [Judgment] Count codes, and also searches for the business's name, direct enquiries and answers to "how did you hear about us?". Ask for the plan, never for analytics access.

### 6. Is a host read worth more?

- [Finding, Podcast Study ads release; Radio Ink] The Podcast Study 2026 (released 23 September 2026; 1,205 US consumers aged 18 to 54 who listen at least monthly; co-sponsored by three podcast publishers, Crooked Media, NPR and PRX): 40% said they were likely or highly likely to try or buy after the host personally read and endorsed a product, against 27% for a prerecorded ad the host didn't read. Stated intentions, not purchases.
- [Seller's figures, Magellan] In Magellan's same January to June 2026 campaigns, host-read ads had a median response of 3.22% against 2.09% for produced ads: site visits, not a forecast.
- [Seller's figures, Podscribe relay] Podscribe's relayed 2025 figures contradict themselves: the text says host reads edge ahead, yet the median conversion rates given put producer reads at 0.019% against 0.017% for host reads. Leave it unresolved; use neither half.
- [Judgment, the episode] A host's read is what carries the host's trust, and that is what the higher price buys. Check the offer is the host's own read on the host's own show, not the owner's recording placed on it, and that the host will only say what they hold (the FTC example above).

### 7. The shows the owner's clients named

If the owner lists shows their clients listen to, then for each one, if you can browse, open its website, its Apple Podcasts or Spotify page and its public feed. Report only what the retrieved page text supports; otherwise mark the affected claim unverified:
- the latest episode date and the gaps between the last several episodes (say what you saw; invent no cut-off for "inactive");
- whether recent episodes have guests, and of what kind;
- whether its topics match the kinds of buyer the owner wants;
- whether it sells ads or charges guests, where a page says so.

[Finding, CastFox] For context: in CastFox's automated count of 30,049 English-language episodes published in 2026 (released 5 October 2026), 40.5% had at least one guest, and 59.8% of episodes from shows whose main Apple category is Business; its most-booked guests mostly had something new to talk about. Automated counts can mistake a co-host for a guest. [Judgment] Clients' answers are leads, not a survey. Public pages show activity and topics, never downloads or who listens. Without browsing, ask for each show's link and its last few episode titles and dates.

### 8. Does the route fit the goal?

[Judgment, the episode] Let the goal pick the route. Sales you can count, and you know what a customer is worth: advertise, price one show first, and judge after a month. Trust with a particular kind of buyer: for most businesses, be a guest on the shows those buyers already hear, if someone can speak well and has something new to say, and judge a run of appearances together. A channel of your own: make your own show, if you can give it the hours for months. A guest spot you pay for is an ad. This order is judgment, not a finding. If the offer does not fit the goal, say which route does, and why.

The right route can still be on the wrong show. [Arithmetic] If the seller gives audience shares (company size, role, industry, revenue), add only mutually exclusive bands that match the owner's stated buyers and bands that clearly do not, and show the sums; ambiguous categories remain ambiguous. Preserve who the figures describe: publisher, newsletter or website demographics are not established podcast demographics; they are [Seller's figures], with no base or date unless the seller gives one. [Judgment] A seller's own audience figures are its best case: if even they show few of the owner's buyers, that counts against fit, and the seller's own words are the fact that shows it.

## Do not say

Common in advice on podcast ads and guesting, and not supported as of October 2026:
- A download, impression or "listen" in an offer means a person heard the ad.
- A forecast of sales, leads, sign-ups or return built from anyone else's response, conversion or return figures, including those above and any in the offer.
- Host reads always win; or the relayed figures show produced ads win.
- A week, or the promo code alone, shows whether an ad worked; or a month catches every sale.
- Booked spots equal hearings per listener.
- Only the show, or only the guest, has a disclosure duty.
- A first test must cost thousands to tens of thousands of dollars, or a local business should skip podcast ads.
- Guesting is proven to bring clients, or needs a set number of appearances.
- A show that charges guests has an audience made mostly of other paying guests.
- A paid link brings search-ranking credit.
- A certificate, media kit, chart or review count proves an audience's size, make-up or results.
- A media kit, a rate card or a listing's "updated weekly" label shows the show is still publishing.

## Decide what comes first

1. Settle before paying: a selling show not shown to be publishing now (its latest episode old against its own rhythm, or no dates to be had); a paid feature whose endorsement context or required audible disclosure remains unresolved (retain the FTC unexpected-material-connection condition); a count with no unit or window, so the price cannot be compared; fees or deliverables left unclear; a booking fee that does not say what it buys.
2. Then mismatches: an offer whose route does not fit the goal, or whose audience, by the seller's own figures, holds few of the owner's buyers; a plan with one hearing; judging after a week or by the code alone; a like-for-like price outside the published asking ranges (a question, not a verdict); a show the owner's clients named with no recent episodes or nothing for the owner's buyers.
3. Then the route [Judgment]. Prefer settling a gap in writing, or pricing one show first, over a large commitment the owner cannot judge.

## Your report

**Hard cap: at most 599 whitespace-separated words for the entire response.** Count raw Markdown, including headings, labels, table delimiters, links, caveats and questions. Aim for 450 to 500 words, leaving room below the cap. Section maxima, including their headings and labels: Verdict 45; Checked 60; The offer 35; The numbers 200; Fit 55; Before paying 100; Gaps and assumptions 25; This week 30. These total 550 and are ceilings, not targets. Fit any owner questions inside Gaps and assumptions; omit that section if nothing remains. Put all content inside these sections. When a local counting tool is available, count the exact final response text and revise until at most 599. Without one, keep the draft shorter and never claim the cap was verified. Do not print a word count. Keep source labels short and group related unknowns. Never repeat the same missing fact in several sections. Cut background, optional study summaries and duplicate questions before cutting the decisive fact, core calculations, source units, legal conditions or next action. Plain language, no preamble.

Return the report as ordinary Markdown, without an enclosing code fence. If asked for a copyable file block, use an outer fence longer than every fence in the file contents, for example four backticks around content containing triple-backtick fences. Never nest equal-length fences.

**Verdict:** a straight answer labelled [Judgment]: take it; not as offered (name the change); not until a named gap is settled in writing; or not for this goal (name the fitting route or kind of show). Lead with the deciding fact and its source. An old listing raises a delivery question; it does not prove no inventory exists.
**Checked:** three compact lines sharing the existing 60-word section budget:
- **Checked today:** each page as `[short name](URL), retrieval date: "few-word exact quote supporting the claim used"`, copied from the relevant passage returned by your tool and retained in its record; otherwise "none".
- **Prompt makers' evidence, 14 October 2026:** source names for rules, prices or findings used from this prompt without your own relevant returned passage; carry [Prompt makers, 14 Oct 2026] into those claims below.
- **Could not confirm:** material pages that failed to open or opened without the relevant passage, with the reason; otherwise "none". Keep the owner's paste labelled seller says or owner says.
**The offer:** one line in the seller's words and unit; latest listed episode date if verified, otherwise activity unknown. Treat paid promotion as an ad for buying decisions [Judgment]; whether an endorsement requires disclosure depends on the FTC condition, not payment alone.
**The numbers [Arithmetic]:** compact table, combining related rows: total cost; count with unit/window/basis; price per thousand and cost per counted unit; conditional hearing cost if each counted delivery were heard once (actual unknown); booked chances and dates, not measured hearings; contribution break-even or revenue coverage; review date about a month after the last airing [Judgment], allowing later sales. Show formulas and units. Group unavailable calculations with the one missing input rather than repeat it. Include matching published price anchors and reverse counts when delivery is missing, each with its own unit and window beside it.
**Fit [Judgment]:** compare the owner's actual buyers and goal with what the seller's figures describe; show relevant non-overlapping band sums when supplied. Publisher demographics are not verified podcast listeners. Name the intended response to the ad and what the owner will count.
**Before paying:** up to three priorities. For each, give the fact/source, why it matters and a seller question ready to paste. Combine related questions; do not add a second fixes list. When disclosure is material, include this rule in the delivered result, labelled [Rule, FTC Guides 255.0, 255.1, 255.5]: "If US guidance applies, a connection that might materially affect endorsement credibility and a significant minority would not expect requires clear, conspicuous disclosure, audible for audio; advertisers and endorsers can both be responsible." Do not reduce this to a seller question or omit the audible requirement to save words. If activity is old or unchecked, ask about current publishing, next airing and new-episode versus back-catalogue placement. With poor fit, label seller questions fallback only.
**Gaps and assumptions:** one short line for material unverified points or necessary working assumptions not already covered; at most three unanswered owner questions after the result. Omit questions whose answers are already present. Never fill a gap with an invented fact.
**This week [Judgment]:** one specific action, who does it, which named show or kind of show, and by what day, using the offer's deadline when given. Poor fit means start toward a fitting route or show, rather than defaulting to more seller questions. Use supplied client-named shows first; with no suitable shortlist, ask three existing clients which podcasts they hear (a chosen action, not research). If fit is plausible but terms block paying, get the named gaps settled in writing. No invented show, contact or audience size.

## Sources (checked by the prompt's makers on 14 October 2026; re-read them when you can browse)

Rules and measurement:
- FTC Guides (16 CFR Part 255, including 255.0 and 255.1): https://www.ecfr.gov/current/title-16/chapter-I/subchapter-B/part-255
- FTC Guides 255.5 (material connections; the podcast example): https://www.ecfr.gov/current/title-16/chapter-I/subchapter-B/part-255/section-255.5
- If eCFR blocks your tool (it limits automated reads), the same sections as copied by Cornell's Legal Information Institute: 255.0 https://www.law.cornell.edu/cfr/text/16/255.0 ; 255.1 https://www.law.cornell.edu/cfr/text/16/255.1 ; 255.5 https://www.law.cornell.edu/cfr/text/16/255.5
- FTC Q&A (plain-language answers, June 2023): https://www.ftc.gov/business-guidance/resources/ftcs-endorsement-guides-what-people-are-asking
- Google spam policies: https://developers.google.com/search/docs/essentials/spam-policies?hl=en
- Google outbound links: https://developers.google.com/search/docs/crawling-indexing/qualify-outbound-links?hl=en
- IAB v2.3 (landing page and date): https://iabtechlab.com/standards/advanced-tv/podcast-technical-measurement-guidelines-v2-3/
- IAB v2.3 (full text): https://github.com/IABTechLab/Podcast-Technical-Measurement/blob/main/PTM-Guidelines_v2.3.md
- IAB explainer: https://iabtechlab.com/clarifying-todays-podcast-measurement-and-preparing-for-what-comes-next/
- IAB certification: https://iabtechlab.com/compliance-programs/podcast-measurement-compliance/

Sellers' pages:
- Acast cost guide: https://www.acast.com/en/news-and-insights/how-much-does-podcast-advertising-cost
- Acast monetize guide (July 2026): https://www.acast.com/en/blog/how-many-listeners-do-you-need-to-monetize-a-podcast
- Acast self-serve: https://learn.acast.com/en/articles/9196817-advertising-on-acast
- Acast targeted ads: https://learn.acast.com/en/articles/8590143-advertising-with-acast-creating-targeted-ads-campaign
- Acast glossary: https://learn.acast.com/en/articles/7061300-podcast-advertising-glossary
- Acast budget (February 2025): https://learn.acast.com/en/articles/6521702-setting-a-campaign-budget
- Libsyn rate card: https://advertising.libsyn.com/podcast-advertising-rates

Research and reporting:
- Podcast Study ads release (23 September 2026): https://podnews.net/press-release/podcast-study-2026-2
- Radio Ink (the study's sample): https://radioink.com/2026/09/28/podcast-study-echoes-radio-listeners-follow-the-host/
- Podcast Study ad load (9 October 2026): https://podnews.net/press-release/podcast-ad-load-2
- Magellan (Q2 2026 benchmark): https://www.magellan.ai/news-insights/podcast-measurement-benchmark-report-q2-2026
- Podscribe relay (Podcast News Daily, 3 September 2025): https://www.podcastnewsdaily.com/news/podscribe-podcast-ads-stay-effective-even-as-loads-rise-to-record-high/article_53901387-0a16-40d0-910f-e67e56d69892.html
- Net Influencer relay (4 September 2025): https://www.netinfluencer.com/5-key-findings-from-podscribe-latest-podcast-advertising-report/
- Podscribe 2025 report page (form-gated): https://podscribe.com/ppb-reports/q3-25
- Podscribe 2026 report page (form-gated; check for newer figures): https://podscribe.com/ppb-reports/q3-26
- Webster (5 October 2022): https://soundsprofitable.com/article/is-it-time-to-rethink-promo-codes/
- Sounds Profitable 2025 (PDF): https://soundsprofitable.com/wp-content/uploads/2025/07/The-Advertising-Landscape-Driving-to-Action.pdf
- YouGov release (June 2026): https://www.marketingdive.com/press-release/20260610-podcast-ads-may-be-skippable-but-they-still-drive-results-new-yougov-stud/
- Bloomberg via NPR (August 2022): https://www.hppr.org/2022-08-10/some-podcast-guest-chairs-go-to-high-bidders-without-telling-listeners
- Podnews 2022: https://podnews.net/update/podcasts-charging-guests-stacks
- Lewis (April 2026): https://theaudacitytopodcast.com/payinterviews
- CastFox (5 October 2026): https://www.castfox.net/research/podcast-guest-statistics-2026

## MY OFFER

Paste the offer and what you know, and skip what you don't. Take out email addresses and phone numbers first; never paste passwords or anything about your customers.

- The offer, exactly as sent: the invitation, media kit, rate card, booking service's pitch or ad quote, with any deadline
- The selling show's name, and a link to it if the offer has none
- What it costs: price, currency, any fees, and what is included (production, a link, social posts, a report)
- What it counts, if it says: downloads, impressions or "listens"; per episode, per month or lifetime; over what window; estimated or guaranteed; who measures it
- What you would get: an interview or an ad; the host's read or your own recording; how many reads or appearances; where in the episode; how long; which dates; any link in the show notes
- Your goal: sales you can count, trust with a particular kind of buyer, or a channel of your own
- Roughly what a new customer is worth to you (revenue or profit, over what period) and your budget
- The kinds of buyer you want to reach (kinds, not names), and any shows your clients told you they listen to, as titles or links
- How and when you plan to judge whether it worked
- Your country, if not the US

[Paste your offer and details here]
No signup or paid AI plan required. Use sanitized notes; keep client details and passwords out.
Open prompt link in Claude Desktop
Requires the desktop app. It asks Claude to read this page. If Claude cannot read it, paste the copied prompt.
From Business PodcastingEpisode 9: Should you make a podcast, be a guest, or advertise?
From this episode, "Should you make a podcast, be a guest, or advertise?". Paste it into ChatGPT, Claude or Gemini with the offer you got: a show asking you to pay to be its guest, a media kit or rate card, a booking service's pitch or an ad quote. Your AI checks whether the show selling it is still publishing, works out the price per thousand in the offer's own unit, and, when the count permits it, the conditional cost per hearing if every counted delivery were heard once, from the offer's own numbers. It treats a paid "guest" spot as advertising for the buying decision, checks whether the arrangement needs disclosure, and checks it against the US disclosure rules, the industry's measurement standard and sellers' published prices for that kind of show as of October 2026, without forecasting sales from anyone else's figures. You get a straight answer first (take it, not as offered, or not yet), whether it fits your goal, what to settle before paying, questions ready to send the seller, and one specific thing to do this week, including the date to judge the result. It isn't legal advice.
Copy the prompt into your AI. One that can browse the web is best, so it can re-read the current rules and prices, check the selling show's own feed for its latest episode, and open the shows your clients named. Its Checked line pairs each page with a few-word quote from the relevant text its tool returned and retained with the URL, retrieval date and status. Rules, prices or findings used from this prompt without that retrieved passage go under the prompt makers' evidence, checked 14 October 2026. Pages that fail to open, or open without the relevant passage, go under could not confirm. If yours can't browse, it works from that dated evidence.
Under MY OFFER at the end, paste the offer exactly as you got it, with the show's name or link, then add your goal, roughly what a new customer is worth to you, and any shows your clients told you they listen to. Take out email addresses and phone numbers, skip what you don't know, and never paste passwords or customer data.
If you supplied offer details, it gives the full review straight away, states assumptions and marks gaps. Answer only any remaining questions after that result. Before you pay, send the seller the questions it lists and check anything it marks as not checked. Run it again on each new offer.
If you want a show idea to work with, start from the ones we have sketched by industry, each with a format and first episodes. Browse podcast ideas for your industry
Give the next check a place to live.

Use the project setup to keep your show brief, evidence and decisions together in your own AI.
Set up your AI podcast project
Bring us the decisions you want help making.

The prompt gives you a starting point. Our team can research, plan and produce the show with you, with your approval before every episode publishes.
See podcast production programs Talk through your podcast project