What Makes a Podcast Pitch Land, and How Hosts Answer: Lessons from 639 Pitch Threads

Executive summary

Pitches that name one specific, personal event in the first message got a yes more often than pitches that led with a guest's credentials. Across two mailboxes, one host received 639 pitch-like threads from 458 senders between 2022 and October 2026. Volume went from 70 in 2023 to 213 in 2025 and 242 so far in 2026. Silence stayed near one thread in three through 2025, then fell to about one in ten in 2026, because a clear no replaced it: declines grew from 7% of pitches in 2023 to 28% in 2025 and 53% in 2026. Of the pitches that got any reply, 49% moved to the guest form in 2023 and 2024, against 36% in 2025 and 24% in 2026. Counts come from the wording of the host's replies and are approximate, though a blind second reading found the form and decline labels reliable. In a close read of a subset of the pitches, the ones that advanced nearly all named a concrete event, either in the pitch itself or in a single reply to the host's question.

The rule hardened as the volume grew. In 2022 the host's reply was an invitation. By 2026 it was a written test with a template for the no. In between, the phrases that now define the show arrived one at a time: a guest form (August 2022), "I don't have guests on to share or teach the audience anything" (April 2023), "line in the sand" (September 2024), "fully booked" (February 2025), "very selective" to avoid burnout as a solo creator (August 2025), and a focus on external events such as sudden loss and accidents (July 2026).

Read this as a case study. It follows one host's inbox over four and a half years. It does not measure the industry, and its comparisons show patterns in one mailbox, not causes. What it offers is rarer than breadth: 639 pitch threads, the host's actual replies, and a record of how his rules changed.

Five findings stand out:

1. The moment decides it. A date, an event and a before and after beat any amount of background. The pitches that advanced could be summarized in one sentence about what happened to the guest.
2. Credentials and topic menus lose. Pitches built around a bio, a framework, a book launch or a list of "angles" were declined almost without exception, however polished they were.
3. Burying the lead is costly. In several pitches the one detail that fit the show appeared as the last item in a list, or after an ISBN, a price and a stack of blurbs.
4. Volume tells. Mail-merge slips, a recycled opener, multi-megabyte attachments and four-email follow-up runs marked a pitch as one of many.
5. One reply can rescue a pitch, or end it. Several senders recovered by answering a single clear question. Others answered with a new topic list or pushed back, and the thread ended there.

The practical takeaway is short: say what happened, to whom, and what life looked like before and after, in the first three lines. Everything else can wait for the second email.

About this analysis

This paper rests on a full analysis of two mailboxes from one independent, host-run storytelling podcast: the host's older address, thelifeshiftpodcast@gmail.com, and his current address, matt@thelifeshiftpodcast.com. The older address holds the Guests label from March 2022 to October 2026 (2,051 messages in 722 threads). The current address holds the whole mailbox from January 2024 to October 2026 (6,346 messages in 3,267 threads), including every reply the host sent. A close read of a subset of those threads informs the patterns and quotes in the middle of the paper. The counts and percentages everywhere else come from all 639 flagged threads. It is a qualitative read with rough counts, not a statistical study.

How this was made. I used Claude, an AI assistant from Anthropic, to read the Guests folders in my two show inboxes, sort the threads and count the outcomes. Then I kept going with it. We checked its labels against a blind second reading, corrected what it got wrong, took out claims the mail could not support, and shaped what was left into this case study and the handouts that go with it. The questions, the judgment calls and the rules described here are mine. The sorting and counting were done with Claude, and I have said where it is rough.

What was reviewed. The source was the host's "Guests" mail label, which holds about 1,230 threads in the current address and 722 in the older one. Pitch threads were found with filters and sorted by the wording of the host's replies, and roughly two dozen were read in full. The host's replies show how each pitch was judged.

How outcomes were classified. A pitch counts as moved to the form if the host's reply sent the guest form. It counts as declined if the host passed or said the calendar was full, as asked for the moment if the host asked which single moment, and as no reply if the host sent nothing in the thread. Anything else is an other reply.

Limits to keep in mind.

• The counts are approximate. They come from filters and reply wording, checked against a blind second reader (see the data notes) and by hand.
• One host's taste is the yardstick. The show is built around a single personal moment, so "declined" here means "not this format", not "a bad pitch" in general.
• Mail outside the Guests label was not reviewed.
• Outcomes that happened outside email, such as a call or a social media message, are not visible.
• Names, agencies and personal details are removed throughout. The host's quotes are short and verbatim, pitchers' words are paraphrased or shortened, and none involve personal trauma disclosures.

Who pitches, and how it arrives

Four kinds of sender account for nearly every pitch, and the kind of sender predicted the shape of the pitch better than the guest did.

Two patterns cut across all four.

Contact-form pitches arrive wrapped. Pitches sent through a website contact form land as a notification with the whole message inside quotation marks. Several of those advanced, because they were written by the guest or a close contact and led with the event.

Repeat senders learn. One publicist who placed a first guest after learning the show's format went on to place several more, each with a named moment, and the later guests were booked months out. Relationship plus fit beat any single email.

How the host's answer changed, 2022 to 2026

Pitch volume roughly tripled from 2023 to 2025 and kept rising in 2026, and the host's explicit declines grew with it. Across both mailboxes there are 639 pitch-like threads from 458 senders since 2022. The counts below combine the older mailbox's export and the full export of the current mailbox, with 24 duplicates removed and 34 more threads dropped after a blind check showed they were not pitches. Outcomes are sorted by the wording of the host's replies, so every count is approximate.

Pitch-like threads by year, with outcomes read from the wording of the host's replies:

Moving forward fell as a share of replies. Of the threads the host answered, 3 of 7 moved to the guest form in 2022, 23 of 47 in 2023 and 33 of 68 in 2024. In 2025 it was 51 of 143, and in 2026 it was 51 of 217. The number reaching the form rose from 23 in 2023 to 51 a year in 2025 and 2026, while the pitches around them roughly tripled.

Saying no became the usual answer. Declines or full-calendar replies were 5 of 70 pitch threads in 2023 and 13 of 103 in 2024, then 60 of 213 in 2025 and 129 of 242 in 2026. As a share of the threads the host answered, that is 11%, 19%, 42% and 59%. The question about a pivotal moment appears in his replies from 2022. His own words show the arc:

• 2022: "I'd love to talk more about one of the most pivotal moments in your life, if you'd like to share."
• 2024: "I do not see a strong connection to The Life Shift podcast episode structure and theme."
• 2025: "I ask that the pitches include the specific moment (line in the sand) that changed the potential guest's life forever. The more specific, the better."
• 2026: "The show is built around one deeply personal moment where everything changed for the guest."

Silence held steady, then fell. About one pitch thread in three got no reply from 2023 through 2025 (23 of 70, 35 of 103 and 70 of 213). In 2026 it was 25 of 242, about one in ten, because the host was sending a clear no instead. Some flagged threads were not genuine pitches, so these silence shares are upper bounds.

What the numbers cannot show. Outcomes are read from the wording of the host's replies, not tagged one by one. A blind second reader checked samples of every outcome (see the data notes). The form and decline labels held up, while "asked for the moment" and "other" are noisier. Wording for declines changed over time, so a low early decline count may reflect wording, not behavior. Only about 160 pitch threads moved to a form, against several hundred guests across the show, so most guests arrived another way.

How the guardrails evolved, in the host's own words

The rules that now define the show were not there at the start. They arrived one at a time, usually in response to a kind of pitch the host kept receiving, and each shows up first in a reply he wrote. This table lists the first time each appears in the two mailboxes.

Usage tells the same story. "Line in the sand" appears in 1 reply in 2024, 36 in 2025 and 79 in 2026. "Fully booked" appears in about 63 replies in 2025 and 38 in 2026. "Right fit" appears 6 times before 2026 and 140 times in 2026, once it became the standard way to say no.

2022 to early 2023, an invitation. Replies asked about "one of the most pivotal moments" and sent a form. Almost nothing was declined: 5 of 81 pitches across 2022 and 2023, while 26 of 81 (32%) reached the form.

Mid 2023 to 2024, a format. The host began saying what the show was not, starting with teaching in April 2023. In 2024, 13 of 103 pitches were declined, mostly on fit, and 33 of 103 (32%) reached the form.

2025, a capacity limit. The calendar filled and the reasons for no changed from fit to room. Declines reached 60 of 213 pitches, and 51 (24%) reached the form. The host also began naming his own limits: a full calendar, a solo operation, and burnout.

2026, a written rule. The phrase "line in the sand" became the test, a focus on external events was added, and the no became a template. Declines reached 129 of 242 pitches (53%), 51 (21%) reached the form, and the share with no reply fell to 25 (10%).

The core idea, one pivotal moment, was in the first replies in 2022. What grew around it was the amount the host had to say out loud, and in 2026 a narrowing to external events. Each step followed a rise in volume, and each moved the rule earlier in the thread, where it could do the screening that silence used to do.

Who pitches, and what the numbers say

Pitching to this show is largely an industry. Of 639 pitch threads, 236 (37%) came from 40 organizations that pitched three or more times, and one placement agency alone sent 26. Overall, 268 (42%) came from 87 senders who pitched more than once. Only 103 (16%) came from personal email addresses.

Relationships beat single emails. Repeat senders reached the form 33% of the time, against 20% for one-time senders. A sender who had already learned the format wrote pitches that fit it.

Name the event in the subject line. Subjects with a word like loss, accident, stroke or diagnosis reached the form 43% of the time, against 23% without. This is a keyword match, so it is a rough guide.

An address is not the same as a self-pitch. Pitches from personal addresses reached the form less often (17% against 27%), and the gap holds among one-time senders (14% against 21%) and repeat senders (21% against 35%). But a personal address does not mean the guest wrote it, and a guest with a business domain counts as an organization. In a rough sort by wording on the earlier, broader 673-thread cut, pitches written in the guest's own voice (60) reached the form 23% of the time, against 26% for pitches written by a representative (94), which is no clear difference. The self-written ones went unanswered more often (33% against 18%) and were declined less often (22% against 39%). Fewer than a quarter of pitches could be sorted this way.

Replies were quick, slowed, then sped up. Of the 475 pitch threads where reply timing could be read, the median first reply came after 3.0 days, 25% within a day and 79% within a week. The median was 1.2 days in 2022 and 1.0 in 2023, then 5.4 in 2024, 3.2 in 2025 and 2.4 in 2026.

How to read these. The groups overlap: a repeat sender is often an organization, and organizations use event subjects more. Outcomes come from the wording of the host's replies, and the form and decline labels held up in a blind second reading. The comparisons describe one inbox and do not show that any one habit causes a yes.

Seven patterns in pitches that didn't land

These seven patterns account for most of the declines. The host's words are quoted as written. Pitchers' words are paraphrased or shortened so no one can be identified.

1. The guest is a résumé, not a moment

The first paragraph is patents, medals, bestseller lists and TV appearances. The event that changed the guest's life, if it appears, is one clause.

• One pitch described its guest's story as the most documented and verifiable of its kind. After the host passed, the sender wrote back that they had framed the guest too much around the accomplishments.
• Another sender opened by saying he cared less about a dramatic low point than about the choices that came after.
• A third conceded, after a decline, that they had led with the work and that the show is not about the work.

Why it misses. Accomplishments answer a question the host did not ask. Fix: lead with what happened and when, and put credentials in the last line.

2. A menu of topics instead of a story

• One sender asked the host to pick from a numbered list of options and reply with a quick yes or no.
• Another offered to go practical or personal, depending on what the audience responded to.
• One pitch included 15 suggested interview questions. An agency admitted that its episode angles were only a general starting point.

Why it misses. A menu hands the host the work of choosing and shows the guest has not decided what their story is. Fix: offer one story. If the host asks for options, offer moments, not topics.

3. The lead is buried

• In one pitch, the only event that fit the show, a serious accident, was the last item in a list of six.
• In another, a family medical crisis arrived under an ISBN, a retail price and three endorsements, followed by the line that the guest has more than one moment.
• A subject line led with a book title, a bestselling author's endorsement and an on-sale date. The sender later apologized for burying the lead.

Why it misses. The host decides in the first few lines whether to keep reading. Fix: put the event in the subject line and the first sentence.

4. The pitch is for a different show

One pitch offered "a practical health segment" on dental care. Others offered leadership strategy, a wellness method, a business pivot after a market drop, and a manifestation coaching program built around stage magic. Several opened by praising a specific episode, which showed the sender had listened but not that the guest fit.

Why it misses. Praise for an episode makes the mismatch more visible, not less. Fix: before sending, ask whether the guest's story could run on this show without mentioning their work at all.

5. Template tells

• A greeting that used the show's directory title as if it were the host's name.
• A greeting that ran the host's name into the next word, and the same email sent twice, 21 seconds apart, to two different addresses.
• An opener that joked about fake-fan emails, used by the same sender two months earlier in a near-identical form.
• Three pitches that arrived as emails of 9 MB or more.

Why it misses. It signals mass sending and undercuts any claim of personalization. Fix: one send, one address, plain text under a page, links instead of attachments.

6. Follow-up pile-ons

• One sender wrote four times in about a week, and the third email switched topics to another idea.
• Another wrote four times over about five weeks before the host replied.
• A follow-up asked a question the host had already answered the week before. The sender replied with an apology.

Why it misses. Changing the topic on each follow-up confirms the pitch is about topics. Fix: follow up once, keep the same story, and check the thread first.

7. Re-arguing a decline

After a decline, some senders restated the same guest in the show's vocabulary. One said the guest's life had shifted in a big way, which did not answer the host's question about a single event.

Why it misses. Re-labeling the same story does not change what happened. Fix: respond to the host's actual concern, which is usually "which one moment, and what happened", or thank them and move on.

What the pitches that advanced did differently

The pitches that reached a form or a recording shared six habits. None of them required a clever subject line or a famous guest. Across both mailboxes, about a third of pitches reached the form in 2023 and 2024 and about one in five in 2026, as the host's rules tightened. The habits below describe the ones that made it.

1. They named an external event in the first few lines. The host's reply to one such pitch said the guest's loss was "exactly the kind of moment The Life Shift is built around." The host used nearly the same words for a seizure story and for two sudden-death stories: the event was clear, so the host could say yes to the story before deciding anything else.
2. They answered the one question the host asked. When the host wrote back asking which single moment created the clearest before and after, the strongest replies took a sentence or two: "The clearest before and after in my life came on [date]." Clear answers drew warm, personal replies.
3. They confirmed the format in writing. A publicist wrote that the guest was fine centering the conversation on who she was before the loss, how it changed her, and who she became. The host sent the form within days. That sentence did the work of a pre-call, which the host does not do.
4. They went back to the guest. One publicist said she had asked the guest directly so his own thoughts came through. Another checked with the guest before confirming. Pitches built on the guest's own words were easier to judge than summaries written by someone else, even when the answer was still no.
5. They listened, then asked. The subject line "Just listened to [a past guest]'s episode and I have a quick question?" led to a booking and, later, to more guests from the same publicist. The host noted when a sender had "clearly listened to the show", even in a decline.
6. They took a no gracefully and left a door open. After a decline, one sender replied that they might make a good guest themselves. Several others thanked the host and offered future guests. Not every decline ends the relationship, and the pitcher is sometimes the better guest.

A model from the host's own pitching. When the host pitched a national outlet, the pitch opened with a personal moment and said "honestly the personal part is the pitch." The same instinct works in both directions: lead with the human story, then explain why it matters to the audience.

Listening shows in the fit, not the compliment. About one pitch in thirteen opened with listener language, such as "I listened to your episode with…" or "big fan of the show." Those pitchers were somewhat more likely to name an event as well (34 of 49, against 55% of other pitches), and they got an answer more often (12% went unanswered, against about 25%). But the lift came from the event. Pitches that named one reached the form 31% of the time, against 18% for those that did not, and adding listener language moved that only from 31% to 35%, which is within noise. Praise with no event did worst (2 of 15). Naming one specific episode was no help either: those pitches were declined more often. This is a small sample, so read it as a pattern, not a rate. The pitchers who clearly had learned the show sent a pitch that already followed its format, and sometimes came back with an answer to the host's one question. That is how listening showed up.

The decision rules behind the yes and the no

The host of this show reads every pitch against six questions. Most declines in the sample trace to a "no" on one of them, and the pitches that advanced ended with a "yes" on all six, sometimes after one follow-up question.

Four further observations from the threads sharpen the picture.

The guest's own account has to match the pitch. In one case the pitch headline named one moment, but the guest's form described a different one, a long personal project. The host flagged the mismatch, and the sender went back to the guest, who then named a car accident. The host replied that the accident was "much closer to where the show lives." In another case an external event headed the pitch, but the guest's own description of his single moment was a burnout that built over many months. The host passed, explaining that the moment was "an internal one that built over many months" and that the show focuses on "external events that happened to a guest before they had any say." The pitch is only as strong as the guest's own answer to "which moment".

There are no pre-calls. The host does not do pre-calls, and the conversation unfolds for the first time on the recording. That makes the written confirmation, and the guest form, the only vetting. A pitch that cannot answer "what happened" in writing cannot be vetted at all.

Capacity is a real constraint. The host runs the show alone alongside a full-time job. Replies can take days or weeks, and the calendar has been booked months ahead for most of the period reviewed. In September 2026 the next open recording slots were in December. Several declines praised the story and cited the calendar. A pitch that fits is not guaranteed a slot, which is a reason to pitch early and politely, not repeatedly.

The process is short. Pitch, then a question about the moment, then written confirmation that the guest will center the personal story, then the guest form, then a scheduling link, then the recording. Every step after the first is faster when the first email already contains the moment.

A pitch makeover

The same guest can be pitched two ways, and the second version is faster to say yes or no to. Both examples below are invented composites built from the patterns above. They do not describe a real guest.

Before: the version that gets declined

Subject: Resilience expert and bestselling author for your audience

Hi there,

I represent Dana Reyes, a keynote speaker and author of a bestselling book on
rebuilding after adversity. Dana has appeared on national TV and has helped
more than 10,000 people. Angles for your audience:

1. The 5 stages of rebuilding
2. Why resilience is a skill
3. How to turn pain into purpose

She can go practical or personal depending on your needs. One-pager attached.
Would you consider her for an episode?

After: the version that earns a reply

Subject: Guest idea: the night a car crossed the median and hit Dana's

Hi Matt,

Your show is about the moment a life splits into before and after. Dana's was
March 14, 2011. She was 24, six weeks from starting a nursing job, when a car
crossed the median and hit hers. She spent a year in surgeries and relearning
to walk, and the career she planned never happened.

I asked Dana and she is glad to keep the conversation on that night and who
she was before and after. She would mention her book only at the end, if at all.

If it is a fit, tell me what you need and I will get her to answer in her own
words.

Thanks for reading,
Jordan

Why the second one works

• The subject line is the moment. The host knows the story before opening the email.
• The first paragraph answers the host's usual follow-up. Date, age, event and what changed are already on the page.
• The format is confirmed. One sentence settles the question the host would otherwise have to ask.
• There is nothing to browse. No topics, no attachment, no credentials, one clear next step.
• It is short. The whole pitch is about 120 words.

Recommendations

For anyone pitching a podcast

1. Find the show's one rule before you write. Listen to two or three episodes and ask what every guest has in common. In this sample, the pitches that missed were usually well written for a different show.
2. Put the reason this guest belongs on this show in the subject line. A named event, not a job title or a topic.
3. Pitch one guest with one story. A menu of angles reads as a guest who has not decided what they want to say.
4. Make the first three lines enough. Who, what happened, why it fits. Everything else can wait for the host's reply.
5. Keep it light. Plain text, under about 150 words, links instead of attachments, one address, one send.
6. Make personalization about fit, not flattery. Referencing an episode helps only if the guest's story genuinely belongs beside it.
7. Follow up once, after about a week, and say so. Keep the same story, check the thread first, and tell the host you will not write again if you do not hear back.
8. Answer the host's question in two sentences. If the host asks a clarifying question, a quick, concrete reply is the strongest signal you can send.
9. Take the answer well. Reply once, thank the host, and leave a door open. Courteous replies kept relationships intact, and several senders pitched the show again later.
10. You can pitch yourself. In a rough sort by wording, pitches written in the guest's own voice reached the form about as often as pitches from representatives (23% against 26%). A guest's own words were also easier to judge than a summary written by someone else. If you do pitch yourself, lead with the one moment, and expect a reply to be slow.

For anyone pitching The Life Shift

The show is a deeply human conversation centered on one personal moment: who the guest was before, what happened, and who they became after. The first email should make that visible.

1. Name one moment. Give a date, a place and what happened. Put it in the subject line and the first sentence.
2. Choose an external event. The show currently leans toward sudden loss, accidents, near-death experiences, life-altering medical events and unexpected tragedies. A realization, a career change or a gradual awakening is unlikely to fit right now.
3. Describe the before, the moment and the after. Three or four sentences are enough.
4. Confirm the format. Say that the guest will center the personal story, and that the book, business or method will be mentioned briefly at the end, if at all.
5. Leave out the extras. No topic lists, frameworks, one-pagers, media kits or suggested questions.
6. If you list more than one moment, expect to be asked which one. Choose the one that most clearly divides the guest's life, and say why.
7. Expect a human timeline. The host runs the show alone, replies can take days, the calendar is booked months ahead, and there are no pre-calls. A confirmed story is followed by a short guest form and then a scheduling link.
8. Illustrative subject lines (invented examples).
   • "Guest idea: the night a car crossed the median"
   • "Guest pitch: the call that said her brother was gone"
   • "Guest idea: the seizure at 29 and who he was after"

For hosts and creators who receive pitches

The full export shows one host learning this in public. These steps come from what that host did, and what the mailbox suggests would have saved effort.

1. Write the one-sentence rule before the volume arrives. This host asked for the pivotal moment in 2022 and was declining pitches that did not fit by 2023. A written rule makes the no easy to send.
2. Keep three replies ready. Not the right fit, fully booked, and "tell me the specific moment." The declines in this mailbox take only a few lines each.
3. Say your rule where pitchers will read it. A pitch page that names the one moment, the current focus and what the show is not should screen out pitches before they reach the inbox.
4. Ask once. When a pitch hints at a story, one clear question about the moment is enough. Do not workshop an angle with a sender who sent a topic list.
5. Judge representatives by the story, not the sender. Publicists and experts send a large share of pitches. The same test applies to them as to anyone.
6. Say what silence means. About one pitch thread in three got no reply through 2025. If you will not answer every pitch, tell senders so on your pitch page.
7. Tag every pitch by outcome. Advanced, asked, declined, silent. This host's drift from open invitations to a narrow rule is visible only because the mail was labeled, and tagging lets you steer it on purpose.

Appendix

Follow-up etiquette

The host replied to most pitches within about a week, and some within a day. A few took two to three weeks, often because of a full-time job and a packed recording schedule. Based on that:

• Wait seven to ten days before following up.
• Follow up once, with the same story, and do not introduce a new topic.
• Check the thread first, so the follow-up does not ask a question the host has already answered.
• Say that you will not write again if you do not hear back.
• Accept the answer. A courteous reply to a no is the best possible last email.

One-page checklist before you hit send

• ☐ I have listened to at least two episodes and can say what every guest has in common.
• ☐ The subject line names one event, not a job title or a topic.
• ☐ The first sentence says what happened, to whom, and when.
• ☐ I describe who the guest was before, the moment, and who they became.
• ☐ I have confirmed in writing that the guest will center the personal story.
• ☐ There is one guest, one story and one clear next step.
• ☐ There are no topic lists, suggested questions, frameworks, one-pagers or attachments.
• ☐ The greeting uses the host's name, and the email goes to one address, once.
• ☐ The email is plain text and under about 150 words.
• ☐ If the host asks a question, I will answer it in two sentences.

Data notes

• Source: one host's two mailboxes: thelifeshiftpodcast@gmail.com (Guests label, 2,051 messages in 722 threads, March 2022 to October 2026) and matt@thelifeshiftpodcast.com (full mailbox, 6,346 messages in 3,267 threads, January 2024 to October 2026, of which the Guests label holds about 1,230 threads). A close read of a subset of those threads informs the patterns and examples. All counts and percentages come from the full 639 flagged threads.
• Counts are approximate and based on reply wording, spot checks by hand and about two dozen full reads. They indicate direction, not precise rates.
• Not reviewed: mail outside the Guests label, and any outcome that happened outside email.
• Outbound evidence is now close to complete for pitches. The exports hold the host's replies from both addresses. They hold almost none of the pitches the host sent as a guest to other shows, so the advice for pitchers still leans on inbound patterns.
• Method: a pitch thread is one that began with an incoming message about guesting, in the Guests label of either mailbox, after removing booking notices, newsletters, social messages, reply chains and recording logistics. Senders are identified by address, because the Sent label wrongly marked about 100 incoming pitches as sent in 2025 and 2026. Threads in both mailboxes were matched by sender and subject, and 24 duplicates were removed. Outcomes come from the wording of the host's replies, with quoted text removed. An earlier filter that was too loose counted booking confirmations and other noise, and was dropped. Accuracy was checked in three blind rounds of 60, 50 and 40 threads, drawn across outcomes. A second reader labeled each thread from the host's replies without seeing the automated label. On genuine pitches the form and decline labels agreed in about 25 of 25 cases each, "asked for the moment" agreed in roughly half the cases checked, and "other" was the noisiest. The checks also showed that about 14% of the 639 flagged threads were not genuine pitches (marketing, thank-you notes and booking logistics), mostly in the "other" and "no reply" groups, so roughly 550 are genuine pitches and the silence and "other" shares are overstated. Form and decline counts and the year-by-year pattern are the firmer findings.
• Privacy: names, agencies and identifying details were removed. Personal disclosures made in pitches, including abuse, addiction and bereavement, were not quoted, and are referred to only by general category.
• Next step for rigor: the outcome labels had a blind second reader, but the six-question tags on the close-read subset did not. Have a second reader tag a random 30 of those pitches and compare their calls with these tags.