The Resume Gate Framework

Nobody tells you why you were rejected. So you guess, and you usually fix the wrong thing.

There are 4 places an application dies. From your side they all look the same: silence. But the fix for each one is completely different.

Below is the one idea that popular resume prompts are missing: a good one asks you questions before it writes anything, and won't fill in what you haven't told it. Try it on a single bullet and you'll see the difference in about three exchanges.

The Gate map

🚪 Gate 0 - Target. Are you aiming at the right role, and is the work that proves it actually on the page?

🚪 Gate 1 - Filter. Do you show up in the search a recruiter runs in the ATS?

🚪 Gate 2 - Skim. Does a human believe in your bullets? Are those bullets "stealable" by your competitors?

🚪 Gate 3 - Defense. Can you back it up out loud in the interview room?

Gate 0 is content, Gate 1 is vocabulary, Gate 2 is whether the writing is believable, and Gate 3 is whether you can defend it out loud.

Find your problematic gate

❌ No callbacks at all → Gate 0 or 1. Your signal doesn't match your substance.

❌ Screened, then silence → Gate 2. They read you and didn't believe it.

❌ Interviews, no offers → Gate 3. The page promised more than you delivered - your storytelling failed.

The trap is treating all these symptoms as if it's a Gate 1 problem - more keywords, tighter formatting, another prompt off LinkedIn. If you're wrong at Gate 0, better keywords only make you fail faster; If you are stuck at Gate 3, you need to fix your stories.

Try it on one bullet

Resume rewrite is a conversation, even with LLM. Most resume prompts work the same way - you paste, it produces, you're done in a single exchange. That single turn is the problem. An LLM model can only rewrite what's on your page, and the thing that makes a bullet believable is in your head - the model has no way to reach. It can't stop and ask you, so it AI-slops back. It doesn't know the difference between a thin bullet VS a thin career.

Mine is built to ask first and rewrite the bullet at the end. Paste any bullet, straight off your current resume.

You are a professional resume writer. Your job right now is NOT to
write anything. It is to interrogate one bullet until the real story is on the
table.

Here is my bullet:
"[PASTE ONE BULLET]"

Collect exactly three things from me:

1. BASELINE — what the number or state was before. The scale it operated at.
2. CONSEQUENCE — why it mattered to the business. What breaks if it doesn't
   happen. Who cared, and what it was worth to them.
3. MECHANISM — what I actually did that caused the change.

How to ask me:

- One question at a time. Wait for my answer. Never batch questions.
- Start by defining any vague word in my bullet. If I wrote "performance,"
  "efficiency," "engagement," or "impact," your first question is what I
  actually mean by it.
- Then work through the three items above in whatever order fits my answers.

Useful probing questions, drawn from how a coach actually drills:
- "Compared to what? What was normal before?"
- "What breaks if this doesn't happen? Who notices?"
- "Was that protecting revenue you already had, or winning revenue you didn't?"
- "Did you decide the approach, or carry out someone else's decision?"
- "How many systems, or people, were involved?"

ALSO ASK FOR THESE TWO - they are what make a bullet impossible to "steal" by my competitors, so
I can stand out from other resumes:

NAME THE SPECIFIC THING. Ask what the system, tool, product, team, or
framework was actually called. Use the real name if it isn't confidential - it
is both un-copyable and a term a recruiter may search for.

WHAT MADE IT HARD. Ask what the constraint, trade-off, or friction was, e.g. a
legacy system nobody had touched in years, two teams that disagreed, a
deadline set before the scope was known. "Delivered on time" is a claim
anyone can make; "delivered on time despite X" is a claim only I can.
Seniority signals reflect what I had to navigate.

THE TEST FOR EVERY ANSWER I GIVE YOU:
Could this statement also be true of a stranger's project? If yes, it is not specific
enough. Ask again - do not accept it and move on.

Say the verdict out loud every time, in a few words, before your next
question: "that could be anyone's - still too general," or "that one's only
yours." If you don't say it, I have no way to know whether you ran it.

A NUMBER IS NOT AUTOMATICALLY AN OUTCOME.
"Ran 200+ standups," "managed 20 campaigns," "filed 3,000 bugs" are counts of
activity, not measures of change. If my number counts things I did rather than
what moved because I did them, ask me what actually changed - revenue, customer satisfaction, man hour saved, foundational setup for future opportunities, etc.

HARD RULES:
- Never invent, assume, or fill in any of the three items. If I haven't told
  you, it stays empty. A plausible guess is a failure, not a help.
- Do NOT rewrite a bullet until you have all three from my own words.
  Not a draft, not a "here's roughly what it might look like."
- If I try to move on early, say what's still missing and ask again.

IF I DON'T HAVE EXACT NUMBERS or objective evidence:

Estimating from memory is fine - I lived it and can back it up. Inventing one
for me is not.

- Ask me to estimate, then ask HOW I got there. "We handled roughly 80-100 a
  week, so a third is about 25-30" is a real answer. "Around 30%" with no basis
  behind it is not - push back and ask what it's based on.
- Write estimates hedged: "~30%", "roughly 25", "about two quarters."
- If I have no number at all, ask for a concrete consequence instead. "The client
  stopped asking for weekly status updates" isn't a metric, but it's specific
  and nobody else can claim it.
- YOU never supply a number - not as a statement, not as a suggestion, and not
  as a range for me to agree to. "Would 15-20% be fair?" is a violation, not a
  question. Simple check: if a figure appears in your message before it
  appeared in mine, you broke this rule. If I can't give a number and can't
  give a consequence either, say the bullet isn't ready and tell me what to go
  find.

THE TEST: if an interviewer asked "how did you get that number?", could I
answer? If yes, use it. If no, it doesn't go on the page.

WHEN YOU HAVE ALL THREE:
Rewrite the bullet. Include nothing I didn't tell you — no invented numbers, no
inflated scope, no added causation.

If I genuinely don't remember something, write the bullet with what I did give
you AND say which is missing: baseline, consequence, or mechanism. Don't hide
the gap.

If it starts writing before it has asked you anything, stop it and say: "Don't write anything yet. Ask me one question at a time until you have all three." That's usually enough to get it back. It's the prompt slipping, not you — and it's worth knowing that this one talks back, so it costs more messages than a paste-and-go prompt does.

If it comes back saying the bullet isn't ready, that's the tool working, not failing. You've just found out which line on your resume can't survive being asked about, which is better learned here than in the room.

See how transformation happens

Before the prompt

"Improved performance by 10%."

🤷 Could be true of others' project

⚠️ A recruiter could not easily remember

After running the prompt

"Cut service response latency 10% to hold new SLA commitments to the existing clients, protecting ~$200K of at-risk revenue."


Same project. Same number. What sits between them is questions being answered — what "performance" actually meant, why 10% mattered, what would have broken if missed it, and what that was worth. None of them were on the resume before, because nobody had ever asked.

What this prompt won't do

It won't hear your hesitation. It won't notice the thing you're steering around. It won't know your industry's tells.

That part still takes a person. I'd rather tell you that up front than let you find out at the wrong moment.

Like the outcome with just one bullet?
Below has the whole method.

  • The target check — if not sure your title matches what you did
  • The keyword gap — the terms a thin posting may have left out
  • The full interrogation — role by role, not line by line
  • The allocation pass — what makes the cut, and what shrinks
  • The room — how to construct impactful stories

The gate map tells you where you're dying. This is the order you fix it in.

The 4 gates above are a diagnosis. They aren't a to-do list, and gate numbers aren't the running order. What follows is.

It mirrors how a professional resume writer actually runs an intake: settle the target first, because everything downstream gets edited against it. Then go role by role and deliberately collect more context. Only then decide what makes the page, and how much room each role gets.

5 steps, but not all the same kind of work.

  1. 1
    The target checkGate 0

    Settle what role you're aiming at, and surface the work that never made it onto the page.

  2. 2
    The interrogationGate 2

    Role by role, get the real story out of your head as bullets.