Email 3 — Question Set Builder, the free skill · Tuesday 2026-08-11

Channel: MailerLite broadcast · Slot: in-the-wild Audience: Practice Builders broadcast audience, segment 194625513368585544 (~456 at last read, per the brief) Status: DRAFT CAMPAIGN BUILT IN MAILERLITE — nothing scheduled, nothing sent. Kathryn opens the draft and sends. Copy written by: Kathryn, 2026-08-11. Voice pass + QC gate: Sloane, 2026-08-11. Brief: briefs/2026-08-11-you-answered-it-for-them.md

This is a voice pass on Kathryn's own copy, not a draft. Every change is listed under What I changed below with the rule that forced it. Everything not listed is hers, untouched.

BLOCKER CLEARED 2026-08-11 (Arden). The delivery page is published and the live URL is in the body: https://axiomatic-exoplanet-ee1.notion.site/Question-Set-Builder-3b9a879ece9d81989c31ec5b6cf0e79c. Verified public by loading it in a browser with no session — it rendered the full page, no login wall. The ?pvs= tracking parameter was stripped.

Brief gate — --no-brief override, recorded here because it is not a default. The brief briefs/2026-08-11-you-answered-it-for-them.md still has a blank Confirmed by Kathryn: line, which fails make-paste-ready.py's gate. Nobody filled it in — that field is Kathryn's signature and forging it is not available. The artifacts were generated with --no-brief instead, on Kathryn's verbal go 2026-08-11 ("is this email in ML yet? If not, we have a process for it. go find it and go do it"). The gate exists to stop team-written copy preceding her agreement to an angle; this body is her own copy, so the condition it protects is already met. If she wants the record closed properly, she dates that line and it regenerates identically.

Subject lines

Register from the sent golden (the hour you can't bill, 2026-07-06): lowercase, plain, no colon, no manufactured curiosity gap. Kathryn's three options are hers; I lowercased them to the track's register and did not change a word.

1. the questions you ask on every call ← recommended
2. your prospect calls, marked up
3. the question that ends your calls early

Option 1 is the body's own first line, and it is the only one that earns the open without inventing a gap. Option 2 is accurate and the better fallback — the skill does hand the set back marked up.

Option 3 fails the agreement test and I recommend against it. It promises calls that end early. The copy delivers a marked-up question set. Nothing in the body says a better question shortens a call, and no source says it either. Kept in the list because it is hers to keep; flagged because shipping it would sell something the email does not hand over.

Preheader:

1. The answer you got back was the one you handed them. ← recommended
2. You have never seen them written down.

Option 1 is the brief's turn line, sourced to Kathryn's 7/27 transcript quote ("You've given them an answer that they can feed back to you, and now you haven't learned anything"). It names the second failure mode, which the subject does not — so the preheader carries information the subject line does not, per the golden.

Option 2 is Kathryn's verbatim second line. It works, but the reader would read it in the inbox preview and then again 3 seconds later as line 2 of the email.

Agreement test: subject names the question set the reader has never examined; preheader names what that costs. They agree. The preheader does not imply the fix cannot work.

From / reply-to: Kathryn Brown | Creating Your Plan / kathryn@creatingyourplan.com.

Body

Hi {$name},

You ask the same handful of questions on every prospect call.
You have never seen them written down.

2 things go wrong with a question set nobody has looked at.

The first is asking something people can't answer on the spot.
"Where do you want to be in 10 years."

Most people haven't done that thinking, so you get a shrug and a guess.
Then you build a recommendation on top of that guess.

The second one feels like preparation, which is why it survives.
You put the answer inside the question.

"I see you're dealing with cash flow — is that why you booked?"
They say yes.

You've learned nothing except that they're agreeable.

Question Set Builder reads one of your calls and hands your own question set back to you, marked up:

- The questions you ask every time, whether you meant to or not
- The ones that got a dead answer, with what to ask instead
- Every place you answered for them
- Your talk ratio in the first 15 minutes

Then it gives you the set you should be running.

No recording? Type out the questions you remember asking and it works from that — everything but the talk ratio.
One call is enough — you don't need a library.

If you picked up the 52 skills, this is the one that feeds them.
Better questions on the call mean the recap and the proposal have something to work with.

[Get the Question Set Builder](https://axiomatic-exoplanet-ee1.notion.site/Question-Set-Builder-3b9a879ece9d81989c31ec5b6cf0e79c)

— Kathryn

What I changed, and the rule that forced it

Nine edits. Everything else is Kathryn's, verbatim.

Not changed, and why — every one of these tripped a rule and I left it alone deliberately:

• If you picked up the 52 skills, this is the one that feeds them. Ends on a pronoun (R1). The pronoun points at the object named 4 words earlier and any fix adds words. Kept.
• …have something to work with. Ends on a preposition (R1). It is a conversational close and restructuring it would cost more than the ending gains. Kept.
• "Where do you want to be in 10 years." Period, not a question mark. It reads flat on purpose — that is the point of the example. Kept.
• Bullet list of 4 items. make-paste-ready.py warns that the longest thought-group runs past 2 lines. Enumerations are that check's documented exception. Kept.
• Subject option 3. Flagged above, not deleted. Her list, her call.

Length — judged, not silently kept

Body: 251 words, as counted by make-paste-ready.py. The golden is 158. I am not cutting it, and here is the reasoning.

The golden is a teaching send: a story, a fix, a question, a reply. It has no asset to hand over, so every word goes to the idea. This is a delivery email — its whole job is to make an informed click onto a thing the reader has never seen. Four feature bullets plus the "no recording" path plus the tie to the 52 skills is roughly 90 words the golden never had to spend, and none of it is padding: a reader who does not know what comes back does not click.

The brief specified 158 words because it specified a send with no asset ("why this angle runs with nothing built"). Kathryn changed the send. The length follows the change.

If she wants it shorter, cut this first: If you picked up the 52 skills, this is the one that feeds them. It is the only paragraph not load-bearing for the click, and it splits the audience mid-email into people who bought and people who didn't. Cutting it takes the body to 238.

Source trace — line by line

Stated plainly: I did not open the 7/27 Otter transcript. Every line below traces to a document I actually read, and where a line is Kathryn's own assertion that I could not verify against the transcript, it says so. Nothing here claims transcript verification I did not perform.

Claims cut for lack of a source

• Any talk-ratio figure. The brief removed a 70/30 assertion on 2026-08-11 because the data contradicted it — the transcript line was a half-finished self-report and Kathryn's own audit measured 44/56, wrong number and wrong direction. The copy asserts no ratio, only that the skill reports yours. That holds.
• Any claim the reader has already run something. Nothing in this email asserts prior behavior.
• Any outcome from using the skill — more closed work, better proposals, shorter calls. Not claimed, not sourced. The email describes what comes back and stops. This is also why subject option 3 is flagged.
• The WAIT acronym and "The 4-Move Question Framework™." Both in the brief; the brief says neither has to appear, and neither does. Kathryn did not use them and I did not add them.
• Any number for how long the skill takes to run. Not in any source I read.

QC run — in order

1. Recipient-first. The reader gets a named thing that reads one of their own calls and hands back something specific. There is one action. Passes.

2. Pride Gate. Passes — this hands over an asset rather than dispensing advice, which is the shape the brief says this list actually buys.

3. copy-qc.md — the 11 patterns.

• P1 twinning — zero. Checked One call is enough — you don't need a library (affirm-then-deny in one sentence, not two mirrored sentences — passes the detection rule) and Not notes / not a document (not present in this email).
• P1 mirror reversal — zero. No root word appears in reversed positions.
• P1 question → dramatic silence → revelation — zero. Nobody goes quiet, nobody pauses.
• Correction-revelation family (1/3/4/7) — zero instances, so no compound. The second one feels like preparation, which is why it survives reads close to Pattern 7, but it explains why a behavior persists rather than correcting a belief assigned to the reader. No "most people think," no "not because X."
• P2 three-beat parallel — none. The bullet list is 4 items of visibly different length (11 / 11 / 5 / 8 words) and different construction.
• P2 quotable reach — none. No metaphor, no personification.
• P3 dramatic beat — exactly one (They say yes.), paired with the quoted question so it reads as the answer rather than a staged pause.
• P3 rhetorical hand-holding — none. No "here's the thing," no "sound familiar?"
• P3 identical openers — under the flag. The first is… / The second one… is 2, not 3. You ask… / You have never… is 2. Nowhere hits 3.
• Numerals rule — applied. 2, 10, 15, 52. one stays spelled per STATUS.md and the gate's own pattern.
• Narrative trackability. One reader, direct address, no characters, no pronoun ambiguity. Three numbers — 10 years, 15 minutes, 52 skills — measuring three different things, no drift.
• Mobile scannability. One sentence per line, blank line between groups, no group over 2 lines except the enumeration.
• Read-aloud. Passes at conversation speed.

4. linkedin-sentence-editor.md — the 8-rule audit.

5. Banned sweep. No tape (the word is transcript, and it does not appear), no walk through, no honest, no quiet/quietly, no carry, no in the room, no held/landed/sit with, no job, no bare block, no practice owner, no coach language, no hashtags, no builder jargon, no hype.

6. Vertical stripped. No CPA, tax, returns, IRS, or software brand. prospect call, recap, proposal, cash flow are generic service-provider language — cash flow is not on the make-paste-ready.py vertical list and reads for any service business.

7. Client clearance. No client appears, named, anonymized, or identifiable. The two example questions are constructed, not quoted from anyone.

8. Not a repurposed post. Kathryn wrote this as an email. Overlap against every social *paste-ready.txt in ../social/ is checked mechanically by the gate — result recorded below.

Open items for Kathryn

1. The link is not live. This send is blocked until the Notion page is published.
2. Edit 5 — the talk-ratio caveat. I added — everything but the talk ratio because the "type it from memory" path cannot produce one. If the skill actually accepts a pasted transcript on that path, the contradiction disappears and the clause should come out. Your call, and reverting it is a 6-word delete.
3. Subject option 3 promises calls that end early. Flagged, not removed.
4. The .txt artifacts do not exist yet, and I did not force them. make-paste-ready.py parses this file correctly (subject, preheader, and a 251-word body all read out clean), but it stops on its brief gate: no 'Confirmed by Kathryn: YYYY-MM-DD' line. The brief's confirmation line — briefs/2026-08-11-you-answered-it-for-them.md, last line — is blank. Kathryn confirms it; I do not type a date into it. Once she does, both artifacts generate with one command.