10 ChatGPT prompts that book you on podcasts in 2026

Edison Research's 2025 Infinite Dial puts monthly US podcast listenership at 47% of the 12+ population, and Podtrac's 2025 top-25 ranker shows the leading indie shows pulling 1.5M-12M downloads per month — yet most guest pitches still read like a 2014 PR blast. Each prompt below targets a failure mode Pat Flynn, Jordan Harbinger, and Tim Ferriss have publicly torn apart: generic flattery, no show-specific angle, recycled questions, no audience-overlap thinking. Outputs: fit research, host-pain map, contrarian angle, anti-sludge letter, 1-pager, unasked questions, social-proof shortlist, no-response follow-up, promo cycle, 12-show tour plan. Free Podcast Pitch Generator.

Open the free Podcast Pitch Generator and run your first show through it.

Baseline cold-pitch reply rates to top-100 shows sit around 1-3% based on public teardowns from Pat Flynn (~95% of inbound pitches auto-archived) and Jordan Harbinger (public "why your pitch was rejected" list). Tim Ferriss's "anti-pitch" essays describe the same pattern from the receiving end.

Pitches conditioned on a specific recent episode, a host's stated pain, and a contrarian-but-defensible angle land in the 8-15% band publicly reported by booking agencies like Interview Valet and PodcastGuests. That gap — 1% to 12% — is what "prompts that book you on podcasts" means here. Not a promise of every show saying yes. A defensible move from auto-archive into a real conversation.

Sources: Edison Research Infinite Dial 2025; Podtrac Industry Ranking, May 2025; Pat Flynn SPI blog; Jordan Harbinger pitch teardowns; Tim Ferriss essays; OpenAI prompt engineering guide. Reply ranges are self-reported by agencies; treat as directional.

Each prompt targets a specific failure mode public host teardowns name. Pitch letters fail because they open with "I'm a huge fan" sludge. Questions fail because the host has answered them 40 times. Follow-ups fail because they say "bumping this up." Post-recording fails because guests vanish the day the episode drops. Every prompt below follows the OpenAI guide skeleton: role, context, output format, length cap.

---

Pitching the wrong show is the highest-frequency error per Jordan Harbinger's rejection list — guests pitch a business show with a memoir angle, or a memoir show with a business pitch. Fit research happens before the pitch is written, not after.

``` You are a podcast booking strategist. Given a show, output a fit-fit-fit or pass verdict for me as a potential guest. SHOW: {{podcast_name}}, host {{host_name}}. MY ANGLE: {{1-sentence expertise + the story you want to tell}}. MY AUDIENCE: {{who you're trying to reach}}. Pull from the show's last 10 episode titles + descriptions (pasted below). Output exactly these 6 lines, no preamble: - SHOW TOPIC: 1 sentence, what this show is actually about (not its tagline). - AUDIENCE: 1 sentence — who actually listens. - AUDIENCE OVERLAP with mine: HIGH / MEDIUM / LOW + 1-sentence why. - EPISODE FORMAT: solo / interview / co-host / panel — and length. - RECENT TOPICS THE HOST CARES ABOUT: 3 bullets from the last 10 eps. - VERDICT: PITCH NOW / WAIT / PASS, with 1-sentence reason. Recent episodes: {{paste 10 episode titles + 1-line descriptions}} ```

**Why it works:** the VERDICT line forces a yes/no/wait decision instead of a vague "could be good." Pasting the actual 10 most-recent episodes blocks the model from hallucinating the show's identity from a stale snapshot.

**Sample output:**

> SHOW TOPIC: How operators actually scale revenue ops past Series B. > AUDIENCE: VPs of Sales and Rev Ops at $20M-$200M ARR SaaS. > AUDIENCE OVERLAP with mine: HIGH — I sell to that exact title. > EPISODE FORMAT: 1-on-1 interview, 55-65 minutes. > RECENT TOPICS THE HOST CARES ABOUT: forecast accuracy after pricing changes; comp-plan rebuilds; the death of the SDR role. > VERDICT: PITCH NOW — my forecast-accuracy story maps to ep #142 and #149.

**Do:** paste 10 recent episode titles, not the show's About page. **Don't:** pitch a show whose verdict is WAIT or PASS — that's how guests burn three relationships at once.

---

Hosts repeat themselves. They have 3-5 questions they ask every guest, 2-3 frustrations they raise across multiple episodes, and 1-2 obsessions that shape every conversation. Pitching against those obsessions is the difference between auto-archive and a reply.

``` You are a podcast research analyst. From the 5 most recent episodes of {{podcast_name}} (transcripts or detailed show notes pasted below), extract patterns the host repeats. Output: - 3 RECURRING QUESTIONS the host asks most guests, verbatim if possible. - 2 FRUSTRATIONS the host has raised across episodes — direct quotes, 10-25 words each, with episode reference. - 1 OBSESSION the host clearly returns to, with 2 episode references. - 3 GUEST TYPES the host has explicitly said work / don't work (e.g., "I'm tired of crypto founders"). No guessing. If a category has no evidence in the 5 episodes, say "NO EVIDENCE" — do not invent. Transcripts / show notes: {{paste 5 most-recent transcripts or detailed show notes}} ```

**Why it works:** the "NO EVIDENCE" rule is the entire prompt. It blocks ChatGPT's worst tendency — inventing plausible-sounding patterns. The verbatim-quote requirement ties every claim to a real episode you can cite back.

**Sample output:**

> RECURRING QUESTIONS: "What's the first metric you stopped trusting?" "When did you know it was time to fire that hire?" "What's the contrarian read on this market?" > FRUSTRATIONS: ep #149, host said "I'm so tired of 'data-driven' decisions that mean nothing"; ep #145, "every founder claims AI is core to their stack — most are lying." > OBSESSION: forecast accuracy after pricing changes (#142, #147). > GUEST TYPES: works = operators with named numbers; doesn't work = consultants without scar tissue.

**Do:** paste actual transcripts. **Don't:** ask ChatGPT to summarize from memory — it'll invent.

---

Tim Ferriss has written publicly that the strongest pitches arrive with a thesis the host hasn't heard, defended with data. Angles, not bios, get bookings.

``` You are a podcast pitch strategist. Given a show + my expertise, generate 3 contrarian episode angles I could pitch. SHOW: {{podcast_name}}, recent obsessions: {{paste output from Prompt 2}}. MY EXPERTISE: {{2-3 sentences, with the numbers / scar tissue / data you actually have}}. Each angle must: - Disagree with a widely-held belief in the show's space — name the belief. - Be defensible with one specific data point or story I personally lived. - Map to at least one of the host's recent obsessions. - Be tellable in a 45-60 minute interview. Output per angle: - WORKING TITLE (under 10 words). - THE CONTRARIAN CLAIM (1 sentence, no hedging). - THE EVIDENCE I bring (1 sentence with a number). - WHY {{host_name}} CARES (1 sentence tied to a recent episode). ```

**Why it works:** "disagrees with a widely-held belief" forces ChatGPT past the safe-zone of summarizing your background. Mapping each angle back to a real recent episode proves the pitch is for *this* show, not a generic blast.

**Sample output:**

> ANGLE 1 — "Forecast accuracy isn't a CFO problem." Most ops leaders treat forecast misses as a finance failure; my data from 14 SaaS rebuilds shows 70% live in pipeline stage definitions. Maps to ep #147. > ANGLE 2 — "The SDR role isn't dying. It's bifurcating." Host called it dying in ep #145; I rebuilt SDR teams at 3 companies and the win is a 2-track org, not deletion. > ANGLE 3 — "Pricing changes don't break forecasts. Comp-plan lag does." Lived this at Salesloft post-2024 reprice.

**Do:** tie every angle to a recent episode. **Don't:** pitch a contrarian angle you can't defend in the first 5 minutes of recording.

---

Pat Flynn has said publicly that 95% of guest pitches he receives open with some variant of "I'm a huge fan of the show," and 95% are also auto-archived. The two facts are related. The fix is removing every sentence the host has read 800 times.

``` You are a podcast pitch writer. Write a cold pitch email to {{host_name}} for {{podcast_name}}. DO NOT USE any of these phrases: - "huge fan of the show" / "long-time listener" - "would love to come on" / "think your audience would benefit" - "I'm reaching out because" / "hope this finds you well" - "thought leader" / "share my journey" - any flattery about the host before sentence 4. 5 sentences max: - S1: reference one specific moment from a recent episode (episode number + 5-10 word quote or detail). - S2: state my contrarian angle from Prompt 3 (one sentence, no hedging). - S3: name the data / story I bring (one number, one customer or year). - S4: 2 specific episode titles I'd pitch, one sentence each. - S5: tiny ask — "want me to send a 90-second Loom walking through one?" Under 130 words. Plain text. No emojis. No P.S. for now. ```

**Why it works:** the banned-phrase list does most of the work — it removes the sludge layer mechanically. The "specific moment from a recent episode" rule proves the pitch is for *this* show, not the 40-show blast the host's auto-archive trained on.

**Sample output:**

> Episode #149's line about being "tired of data-driven decisions that mean nothing" — that's the exact frustration I had at Salesloft post-2024 reprice. The contrarian read: forecast misses are mostly pipeline-stage-definition problems, not CFO problems. I rebuilt forecast accuracy at 14 SaaS companies — average lift 64% → 87% in under 8 weeks. Two episode angles: "Why your forecast is lying, and it's not the CFO" or "The 2-track SDR org that replaces the role you keep killing." Want me to send a 90-second Loom walking through one?

**Do:** quote one real episode line. **Don't:** open with anything about yourself.

---

Hosts read a 1-pager. They don't read a 4-page bio doc. Per Pat Flynn's guest-booking workflow, the 1-pager is what gets forwarded to the producer; the bio doc gets lost.

``` You are a podcast producer. From the pitch (paste below), output a clean 1-page guest brief in this exact format: GUEST: name, current role, one-line credibility (number or named org). EPISODE ANGLE: working title + 2-sentence premise. THREE TALKING POINTS: 1. {{point}} — the data / story / moment behind it. 2. {{point}} — the data / story / moment behind it. 3. {{point}} — the data / story / moment behind it. WHAT THE LISTENER LEAVES WITH: 2 concrete bullets. WHY THIS GUEST FOR THIS SHOW: 1 sentence tying to a recent episode. WHERE TO REACH ME: email + LinkedIn only. Total length: under 250 words. No adjectives like "renowned" or "award-winning." Numbers wherever possible. Pitch: {{paste pitch + Prompt 3 angle}} ```

**Why it works:** producers scan for structure. The fixed template means the producer's eye lands on "three talking points" and "what the listener leaves with" in two seconds — which is the entire screening decision.

**Sample output:**

> GUEST: Aisha Okafor — ex-Salesloft revenue lead, rebuilt forecast accuracy at 14 SaaS companies. > ANGLE: "Why your forecast is lying, and it's not the CFO." 70% of forecast misses live in pipeline stage definitions, not finance. > TALKING POINTS: (1) The 64% → 87% accuracy lift — what changed first. (2) Why pricing changes break comp plans before they break forecasts. (3) The 2-track SDR org I built at 3 companies. > LISTENER LEAVES WITH: a 5-question pipeline-stage audit; a comp-plan-lag flag. > WHY: maps to your ep #142 and #147.

**Do:** keep it under 250 words. **Don't:** attach a 4-page bio — it kills momentum.

---

Per Jordan Harbinger's public producer notes, guests who suggest "what's your morning routine" get filed under low-effort. Questions the host hasn't asked yet — generated from gaps in recent episodes — are the entire signal of guest prep.

``` You are a senior podcast producer. Given the recurring questions from Prompt 2 (paste below) and the angle from Prompt 3, generate 7 sample questions the host could ask me that they have NOT already asked generically in the last 10 episodes. For each question: - The question itself (under 25 words). - 1 sentence on WHY this question is new to this show. - 1 sentence on the answer's hook (a number, a counter-intuitive move, a named moment). Rules: - Skip "morning routine," "book recs," "advice to your younger self." - At least 3 questions must involve a specific number I bring. - At least 2 questions must invite me to disagree with the host on air, politely. Recurring questions: {{paste Prompt 2 output}} Angle: {{paste Prompt 3 winning angle}} ```

**Why it works:** the "disagree with the host on air, politely" rule produces the kind of question that makes a real episode. Hosts book guests who tee up tension, not guests who tee up agreement.

**Sample output:**

> Q1: "You called the SDR role dead in #145 — what would change your mind?" Why new: the host has said it dying but never asked what would reverse it. Hook: I'd argue bifurcation, with the 2-track data. > Q2: "You ask every guest about the first metric they stopped trusting — what's the metric you keep trusting that you shouldn't?" Why new: inverts the recurring question. Hook: pipeline coverage ratio. > Q3: "My data says 70% of forecast misses aren't CFO problems. Where do you actually see the breakage?"

**Do:** suggest 7 — the host picks 3. **Don't:** suggest "morning routine."

---

Most pitches dump every credential. The right ones for this show beat the most. Tim Ferriss has written that he ignores generic credit stacks and reads for the one named thing that maps to his audience.

``` You are a podcast pitch editor. From my full credential list (paste below) and this show's audience (from Prompt 1), pick the 3 strongest social-proof items FOR THIS SHOW SPECIFICALLY. Rules: - Each must mention either a name the audience recognizes, a number they care about, or a publication they read. - No "as featured in" lists. - No vanity titles ("thought leader," "top voice," "award-winning"). - If a credential is impressive but irrelevant to this audience, cut it and say why in 1 line. Output: - 3 KEEP: the chosen items, each in under 18 words. - WHAT I CUT: 3 lines, one per cut, with 1 reason each. Full credential list: {{paste full bio / LinkedIn / About page}} This show's audience: {{paste Prompt 1 audience line}} ```

**Why it works:** the cut log is the discipline. Forcing the model to name what got cut and why prevents the default "include everything" output that producers skim past.

**Sample output:**

> KEEP: rebuilt forecast accuracy at Salesloft 64% → 87% in 8 weeks; quoted in SaaStr 2024 on pricing-change comp lag; advised 3 PE-owned SaaS portcos on rev-ops rebuilds. > CUT: Forbes 30 Under 30 (this audience reads ops blogs, not Forbes). CUT: Stanford MBA (operators here distrust pure MBA credentials). CUT: TEDx talk (off-topic, dilutes the rev-ops focus).

**Do:** cut more than you keep. **Don't:** add the cut items to the P.S.

---

"Just bumping this up" is the phrase most negatively correlated with reply by every public host teardown. A follow-up has to introduce something the first email couldn't — usually a freshly-shipped data point, a new episode connection, or a public moment.

``` You are a podcast pitch writer. Write a follow-up to {{host_name}} — no reply to my first pitch 6 days ago. First pitch was about {{angle from Prompt 3}}. DO NOT use: "bumping this up," "checking in," "circling back," "following up," "just making sure you saw," "didn't want this to get lost." 3 sentences max: - S1: introduce ONE new thing — a recent episode they dropped THIS week, a public moment they had, OR a data point I shipped since my first email. - S2: tie that new thing back to my original angle. - S3: same tiny ask as Prompt 4 — the 90-second Loom — or smaller. Under 70 words. No subject-line repeat. ```

**Why it works:** the banned-phrase list strips ChatGPT's reflexive fallbacks. The "ONE new thing" rule forces the model to source something fresh — which is what a manual, thoughtful follow-up does, and an automated one doesn't.

**Sample output:**

> Loved your Monday ep with the Gainsight CRO — her point about "forecast as a coordination ritual" is the exact frame I used at Salesloft. Adds a layer to the angle I sent last week on pipeline-stage breakage. Still happy to send a 90-second Loom if it's easier to evaluate that way.

**Do:** make the new thing genuinely new. **Don't:** invent — host will Google it.

---

Per Pat Flynn, the single highest-ROI guest behavior is post-episode promotion. Hosts re-invite guests who drove downloads. Most guests vanish the day the episode drops.

``` You are a guest-promo strategist. Given an episode (name, host, drop date), build a 4-touch promo cycle. EPISODE: {{title}}, host {{host_name}}, drops {{date}}. MY CHANNELS: {{list — newsletter size, LinkedIn followers, X followers, which Slack / community memberships}}. Output a table with these rows: Day -1, Day 0 (release), Day 3, Day 14. For each row, give: - CHANNEL (pick the right one for the day). - ANGLE (the specific hook for that day). - ASSET (a 280-char post, a 90-sec clip script, a 1-paragraph newsletter blurb, etc.). - THE METRIC THE HOST WILL SEE (downloads, comments tagged, replies). Rules: - Day 0 must include a clip the host can quote-retweet. - Day 14 is the second-wave — sustain past the drop spike. - All copy peer-to-peer, no "check out this episode I'm on!" energy. ```

**Why it works:** the metric column is what re-books you. Hosts can't see your love; they can see the comment thread tagged with the episode hashtag and the spike on day 14 when most guests have gone silent.

**Sample output:**

> Day -1 — newsletter: "This drops tomorrow — the 3 questions I argued with [host] about" + 1-paragraph hook + episode link. > Day 0 — X + LinkedIn: 90-sec clip of the contrarian moment, quote-tag host. Asset: clip with subtitle quoting "forecast as coordination ritual." > Day 3 — LinkedIn long-form: the data behind the strongest moment from the episode, tagged with host. Metric: comments and DMs. > Day 14 — second-wave: "two weeks in, the question that's getting the most reply." Metric: sustained download bump.

**Do:** tag the host on day 0. **Don't:** disappear after day 0 — that's why you don't get re-invited.

---

A tour beats one-offs. Booked correctly, each subsequent show pre-warmed by the last via audience overlap. Booked wrong, each show resets to cold. The sequencing is the entire game.

``` You are a podcast tour planner. Given my goal audience and 12 candidate shows, sequence them in a 12-week tour order that maximizes compounding reach. MY GOAL AUDIENCE: {{1 sentence — who you want to reach and what action you want them to take}}. MY 12 CANDIDATE SHOWS (with monthly downloads + audience description): {{paste list of 12}} Output a 12-row plan: - WEEK - SHOW + HOST - WHY THIS WEEK (audience overlap with the previous week, network effect, topical timing). - THE ANGLE I PITCH FOR THIS SPECIFIC SHOW (from Prompt 3 angles). - THE OVERLAP HOOK with the prior week. Rules: - Open with the most warm / referrable show. - Save the biggest show for week 7-9, after I've recorded 6 reps. - Cluster shows whose hosts cross-promote each other. - Last 2 weeks should target a different sub-audience to broaden reach. ```

**Why it works:** the "overlap hook" column forces the planner to find why show N+1 listeners would have heard about you on show N. That's how a tour compounds.

**Sample output:**

> Week 1: SaaSOps Pod — warmest intro, smallest audience, lets me rep the forecast angle. > Week 2: Rev Ops Weekly — same audience, host follows week 1 host. Overlap hook: "saw this clip from last week." > Week 7: Pavilion — biggest reach, after 6 reps. New angle: 2-track SDR org. Overlap hook: 4 previous hosts have been Pavilion guests. > Week 11-12: shift to CFO Tea + Finance Founders — different sub-audience, opens a second arc post-tour.

**Do:** record the smallest 6 first. **Don't:** lead with your biggest target cold — you'll waste it on rusty reps.

---

Sources: Edison Research Infinite Dial 2025; Podtrac May 2025; Pat Flynn SPI blog; Jordan Harbinger pitch teardowns; Tim Ferriss public essays. Reply-rate multipliers are typical observed ranges from publicly-reported booking-agency data, not guaranteed outcomes.

A 21-day sequence for one target show:

1. **Day -7 (Prompt 1):** fit research. If verdict is PASS or WAIT, stop here. 2. **Day -5 (Prompt 2):** mine 5 recent episodes for host pains. 3. **Day -4 (Prompt 3):** generate 3 contrarian angles, pick one. 4. **Day -3 (Prompt 7):** curate 3 social-proof items for this show. 5. **Day -2 (Prompt 5):** build the 1-pager. 6. **Day 1 (Prompt 4):** send the cold pitch. 7. **Day 7 (Prompt 8):** follow up with one new thing. 8. **Day 14:** booked? Run Prompt 6 for sample questions. 9. **Recording + 14 days (Prompt 9):** execute the 4-touch promo cycle. 10. **Parallel (Prompt 10):** sequence the next 11 shows into a tour.

Run any show through the free Podcast Pitch Generator to assemble these into a sequence without prompt-stitching by hand.

- Podcast Pitch Generator — turn fit research + angle into the pitch letter from Prompt 4. Free.

- Business Email Generator — drafts the follow-up from Prompt 8. Free.

- LinkedIn Post Generator — writes the Day 0 and Day 3 promo assets from Prompt 9. Free.

- Pitch Deck Generator — pairs with the 1-pager from Prompt 5 if a deeper brief is needed. Free.

- Listen Notes — show-search and recent-episode metadata feeding Prompts 1 and 2.

- Podchaser — host credits and cross-show appearance graph, useful for the tour-plan overlap in Prompt 10.

Build your first show pitch in the free Podcast Pitch Generator.