Skills/How GTM Vault is run / Produce
Trailer brief for the editor
Pulls the hooks with timestamps, sets music, branding and references, then reviews the delivered cut against the brief.
Download video-editor-brief.zip
Needs: The episode audio, video or transcript. Or run it without installing: connect the GTM Vault MCP and ask Claude to run video-editor-brief.
Published by GTM Vault (gtmvault.co) as part of How GTM Vault Is Run. Swap the show names, links and templates for your own.
Your video editor cuts the trailers for GTM Vault video content. For every episode, the host provides the audio or video (or a transcript), and this skill turns it into a per-episode creative brief so the editor hits the vision on draft one with minimal revisions. The brief covers the four inputs an editor needs: specific quotes/hooks with timestamps, music and sound, branding and assets, and references.
This skill also owns the return leg: reviewing the cut the editor delivers against the brief before the host signs it off. See Step 5.
Scope: all three shows. Coffee Walk, GTM Vault Podcast (GTM N), and Show Me Your Stack (SMYS N). Each has its own tone defaults below. Coffee Walk is the flagship show and is aimed at a broad global audience, not the tech-executive audience the podcast and SMYS serve. Its selection rules are different and they are not optional. See Step 2.
Deliverable per episode: a trailer brief, delivered as a Notion page.
Standing facts to check, not assume
Feedback to the editor usually travels over more than one channel: email, a chat app, and edits to the brief page itself. Before claiming the host has or has not asked for something, check every channel. Assuming from email alone produces wrong conclusions.
Deliverables come from your agreement with the editor, not from habit. Before listing an extra format (for example a 9:16 vertical cut) as a deliverable, confirm it is in the current agreement. If the agreement does not name it, do not list it.
Trailer runtime targets differ by show. Coffee Walk is 90 to 120 seconds. The podcast is 60 to 90. SMYS is 30. For the podcast and SMYS, stay inside 90 seconds and hold the logo card 2 to 3 seconds.
Workflow
Step 1: Ingest the episode
The host provides one of: video file, audio file, or transcript.
- Video/audio file: extract audio with ffmpeg if needed, then transcribe with word/segment-level timestamps (e.g.,
pip install faster-whisper --break-system-packages). Timestamps are non-negotiable; the editor needs exact timecodes to cut against. - Transcript provided: confirm it carries timestamps. If not, ask the host for the media file or a timestamped export (Riverside exports these).
- Never trust auto-transcript spelling of names. Verify guest name spelling against LinkedIn or the host's notes before it goes in the brief. Auto-transcripts routinely mangle surnames. Where public sources disagree on diacritics, say so in the brief rather than picking one silently; the spelling gets burned into the lower third.
Confirm with the host if not obvious: which show, episode number, guest name and company, and which export file the timecodes are keyed to.
Timecode frame is the single highest-risk field in the brief. Name the exact file and its runtime (for example, "episode 01 master export, 49:54" or "the rough edit export, 44:13"). If the frame is ever corrected after the brief has gone out, tell the editor directly on the channel they are using. Updating the Notion page alone is not telling them.
Step 2: Analyze for trailer moments
Read the full transcript and select material against the criteria for the show. The two rankings below are different on purpose. Do not apply the podcast ranking to a Coffee Walk.
GTM Vault Podcast and SMYS ranking
The audience is operators and revenue leaders. They came for the mechanism.
- Concrete numbers and hard facts. Open on the hardest concrete fact in the episode. A quote with a number beats a quote with an opinion.
- Structural claims. The house headline formula is [structural failure] + [system-level fix]. The best hooks state a failure ("your pipeline doesn't compound", "CAC is rising") or a contrarian reframe ("X was never the problem; Y is").
- The guest's sharpest contrarian moment. The line someone would argue with.
- Emotion and character beats.
Coffee Walk ranking (flagship, broad global audience)
The viewer may not run an ERP, may not work in GTM, and may not be watching in their first language. A category thesis in the first ten seconds loses them. They stay for the person and the place.
- Stakes and character. What the guest risked, lost, chose, or regrets. The decision with a person on the other side of it.
- The number that gives the character weight. An exit, a valuation, a headcount, a year. The number is there to make the stakes legible, not to make an argument.
- The structural claim. It belongs in beat two or three, as the reason this person is worth listening to, never as the opening.
- The city. Treat it as the second character, not as transition footage.
Coffee Walk cold-open rule. Never cold-open on the category thesis or the product claim. Open on the highest-stakes human line available in the episode. The viewer needs a reason to care who is speaking before they are asked to care what the guest is arguing. If the brief's recommended cold open could run unchanged on the podcast, it is wrong for Coffee Walk.
Cold open means cold. When the brief calls a cold open, the trailer starts on the quote itself with no runway, no setup shot, no breath.
Applies to every show
Reject: pleasantries, host restating the guest, setup logistics, generic advice ("focus on your customers"), anything requiring context the trailer cannot provide.
Select 6 to 10 candidate moments. For each: exact timestamp range, verbatim quote, and one line on why it earns its place.
Step 3: Draft the brief
Structure (this maps to the four inputs the editor needs, plus the context they need):
1. Episode overview
- Show, episode tag (GTM N / SMYS N / Coffee Walk N), guest name + title + company, runtime of source, one-paragraph summary of the episode's core argument.
- Trailer objective. For the podcast and SMYS the hook is the structural claim. For Coffee Walk the objective is the person, not the claim.
- Target trailer length and formats, taken from the current agreement with the editor. State the runtime range explicitly; Step 5 checks against it. List a 9:16 cut only if the agreement names one.
- Name the export file the timecodes are keyed to, with its runtime.
2. Hooks and quotes (with timestamps)
- Open with the recommended cold-open moment: the single strongest quote, timestamp, and why it leads. Apply the show's ranking from Step 2.
- Then the ranked list of selected moments:
[MM:SS-MM:SS] "verbatim quote": why it is here / where it might sit in the arc. - A suggested trailer arc in 3 to 5 beats (hook, tension/build, payoff, title card + CTA). The editor has creative freedom within the arc; the arc exists so draft one is close.
- Name the emphasis beats. Mark which one to three moments carry an on-screen emphasis treatment and give the exact trigger word and timecode. See the branding section for the spec. Do not leave this to the edit to decide.
- Where a transcript is unreliable (accent, street noise), mark every quote as reconstructed and instruct the editor to take on-screen wording from the audio, never from the brief.
3. Music and sound Per-show defaults (override per episode if the content calls for it):
- Coffee Walk: documentary-first. Natural/diegetic city sound (street ambience, footsteps, cafe noise) does the work; score is minimal, warm, understated, entering only at transitions or the close. No corporate stock music, no upbeat "podcast intro" beds. Think Theroux/Bourdain: the place is the soundtrack.
- GTM Vault Podcast: minimal modern electronic or textural bed, low in the mix under dialogue. Restrained, confident, slightly dark. No EDM drops, no motivational-corporate.
- SMYS: same family as the podcast, can be slightly more kinetic since it is a builder/demo show. Keyboard/UI sounds from the screen share are fine to feature.
- Universal: dialogue is always the lead instrument. Duck music hard under speech. No risers or whooshes on every cut.
- Minimal score is not no score. "Score enters at transitions and at the close" is a deliverable, not a suggestion. A cut with no music anywhere has not met the brief. Say this in the brief in those words.
4. Branding and assets State these in every brief so the editor never guesses:
- Brand color: gold
#c8a832. Accent for anything marked as the problem/constraint: coral#e0663c. Dark surfaces: bg#0b0b0a, card#141412, line#2a2a26. Text:#f2f0e9, muted#a3a099. - Fonts: Inter (titles, captions, lower thirds) and JetBrains Mono (labels, series tags, timestamps).
- Gold sits on the asset half of a claim, never on the thing being displaced; coral marks anything dying. Flat design, no gradients.
- On-screen text rules: no emojis, no em dashes, no exclamation marks. Sentence case for headline text (not all caps), except SMYS thumbnail-style tool names which are heavy all-caps. Check the literal strings: "YouTube" carries a capital T, and episode numbers are two digit ("01", not "1").
- Series badge per show: "GTM VAULT PODCAST" / "SHOW ME YOUR STACK" / "COFFEE WALK" letterspaced in muted grey with a gold rounded badge holding the episode number, bottom-left. The digit inside the gold badge is set in the dark background colour
#0b0b0a, never in grey, which does not hold contrast on gold at phone size. - Title cards inside the trailer hold for about 1.5 seconds. A location or series card sitting on screen for four or five seconds stalls the trailer.
End cards differ between the trailer and the full episode. Do not cross them.
- Trailer end card: the show's own static branded slate. Episode title, guest name and company, one quiet CTA in house voice ("Full episode on YouTube and gtmvault.co"). Series badge bottom-left. Inter on the title line. Zero animation. Hold 2 to 3 seconds. No trailing black.
- Full episode outro: the standard GTM Vault outro and subscribe animation, because by then the viewer has watched the whole thing.
- Never swap one for the other, and never replace a show's own end card with the generic GTM Vault outro.
Captions.
- Coffee Walk carries no burned-in captions. The documentary treatment carries the episode and a running caption track fights it. Where a line is genuinely hard to catch, the emphasis treatment below covers it, because the emphasised text is the line. Ship an SRT with the upload so non-native speakers and muted viewers are served off-canvas, and let YouTube handle the rest.
- Podcast and SMYS keep the full burned-in caption default: clean, high-contrast, Inter, no bouncing per-word hype captions.
Emphasis beats (all shows). Rationed typographic punctuation on the moments that carry the trailer. This is the house alternative to captions and to stock motion graphics.
- Two to three per trailer. Three is the ceiling. More than that and none of them land.
- Trigger only on: a figure, an exit, a valuation, a date, or a named stake. Never on a general statement.
- Set the guest's own words, verbatim, not a sourced figure. Inter Medium, base
#f2f0e9, the emphasised span in gold#c8a832. - Flat. No box, no gradient, no drop shadow. Six-frame fade in and out, hold 1.0 to 1.5 seconds.
- The cut comes first. Before adding any emphasis treatment, check that the shot holds about one second past the payoff word so the line lands on a face. A graphic cannot rescue a cut that moves on before the sentence finishes.
- Never: money imagery, cash, coins, banknotes, currency symbols, rolling counters, confetti, 3D objects, shine or lens-flare passes, animated per-word captions.
- Brand-motif treatments such as a vault mark are end-card only, flat, gold, under one second. Not on a guest's money line, where the motif reads as a joke about wealth.
Figures on screen require clearance. Any figure attributed to a guest's former employer is flagged, not published, until the host confirms it with the guest. When a figure has been cleared, the brief states the exact figure, its label, and what must not appear alongside it. Setting the guest's spoken phrase as type instead of a sourced number needs no clearance and is the default when clearance has not come back.
Assets. Link or attach any logo/font files the host has shared with the editor; if an asset has not been sent yet, flag it in the brief as "to be supplied".
5. References
- Coffee Walk: Louis Theroux (https://www.youtube.com/watch?v=hRpHAE-LSEE) and Anthony Bourdain: Parts Unknown (https://www.youtube.com/@AnthonyBourdainPartsUnknown). Tone: observational, unhurried, the city as a character, no over-editing.
- Podcast/SMYS: add episode-specific references only when the host names them; otherwise state "match prior GTM Vault episode trailers". The "complete the sentence" cut to logo from GTM 51 is the ending to use on every podcast episode.
- Add any extra references the host gives for the specific episode.
6. Do-not list (include verbatim in every brief)
- No emojis, em dashes, or exclamation marks in any on-screen text.
- No urgency language ("don't miss", countdowns).
- No banned vocabulary in title cards: "unlock", "game-changing", "revolutionary", "10x", "synergy", "leverage AI to".
- Never cut a quote so it misrepresents the guest's point. If a number is attributed to a guest's former employer, flag it; the host confirms with the guest before publishing.
- No third-person framing of the host in on-screen copy.
- No b-roll under a number or a claim. Cutaways go on connective tissue, never on the line the trailer is built around.
- No risers or whooshes substituted for a missing score.
- Never put anything on screen taken from the transcript. Use the spellings in the brief and the wording from the audio.
Step 4: Deliver as a Notion page
- Search Notion for a page called "Video Briefs". If it does not exist, create it (private pages) and tell the host where it lives.
- Create the brief as a subpage titled
[Episode tag] | [Guest] | Trailer Brief(e.g.,GTM 48 | Jane Doe | Trailer Brief,Coffee Walk 1 | John Doe | Trailer Brief). - Give the host the page link and a 3-line summary in chat: recommended cold open, trailer arc in one sentence, and anything flagged as missing (assets, unverified name/number).
- If the editor requests Notion access per page, remind the host to grant it, or the brief is a dead link.
The host reviews and shares with the editor. Optionally offer to draft the short handoff message to the editor pointing at the page.
Step 5: Review the cut the editor delivers
Do not pass a delivered trailer to the host as done without checking it against the brief. Revision rounds are finite, so the review has to be right the first time.
- Establish what was already asked for, across every channel. Email, chat, and the brief page's own revision history. A delivered cut is usually answering a specific earlier round, and notes that re-raise something already fixed, or accuse the editor of ignoring a note they were never given, destroy the review's credibility.
- Confirm the file is new. Check the delivery folder's modified time and the file's created time. Folder modified time does not reliably update when a child file changes, so trust the file's own timestamp over the folder's.
- Re-transcribe the delivered cut. Do not reuse the source transcript. Timecodes in review notes reference the delivered trailer in seconds from the first frame, and must be labelled as such, because the brief uses a different frame and mixing the two causes real errors.
- Sample the video accurately. Extract with an explicit
-ssstart and compute the label from the frame index, then check one known landmark before trusting any timing. A contact sheet built from a pre-rendered frame sequence can carry a one-second label offset, which produces false notes about card durations. Verify a timing twice before it goes in a note that says a deliverable was missed. - Measure the audio. Compute the RMS floor across the cut. A floor near -45 dBFS or below in the speech gaps means there is no music bed at all, which the eye does not catch but the viewer feels.
- Sample colours, do not eyeball them. Read actual pixel values for badge gold, wordmark grey and text, and compare to the palette. Allow for h.264 chroma error of a few points per channel before calling a colour wrong.
- Check against the brief, in this order. Runtime inside the target range for that show. Score present where the brief asked for it. Every claim and figure landing on a face rather than b-roll. Cold open matching the show's ranking and starting dry. Title and end cards within their hold times, correct card for trailer versus full episode, no trailing black. Emphasis beats present on the moments the brief named. Captions matching the show's rule. Literal text strings.
- Separate four categories in the notes, and label them: already delivered (verified fixed, do not re-raise), spec gaps (the brief asked, the cut did not deliver), fixes (cheap changes inside the agreed direction), and new direction (a change to the brief, which is the host's call and may count as a new revision round). The editor executing the brief as written is never listed as their error.
- Deliver as a subpage titled
[Episode tag] | [Guest] | Trailer Revision Brief (cut delivered [date])under Video Briefs. Leave the original brief intact as the reference document. Open the page with what the cut got right, so the editor does not undo it.
Quality bar before handing off
- Every quote is verbatim from the transcript with a timestamp the editor can seek to.
- Guest name spelling verified against a non-transcript source, with any diacritic conflict flagged.
- Show-appropriate ranking applied. A Coffee Walk brief that opens on a category thesis has failed this check.
- All four editor inputs are present: quotes/hooks with timestamps, music/sound, branding/assets, references.
- Emphasis beats named with trigger words and timecodes, two to three maximum.
- Captions rule stated for the show (none burned in for Coffee Walk, SRT shipped instead).
- Deliverables list matches the current agreement with the editor, including whether a 9:16 cut is in scope.
- On a review: every claim that a deliverable was missed has been measured twice, and every prior round has been checked across every channel and the brief page before anything is called outstanding.
- The editor should never have to fall back on their own judgment for lack of direction: no episode goes to the editor without a brief.