# Daniel Nogueira, voice file

**Built from:** last 30 LinkedIn posts, Aug 7 to Oct 8, 2026, pulled Oct 8, 2026 from his own activity page in Claude's browser. 25 regular posts, 5 lead magnet posts.
**Refresh:** once a month.

## Three things the count found that he wouldn't have said

1. **Em dashes crept in.** 6 em dashes, all in 3 posts published between Aug 24 and Aug 27. None in the other 27. That's a drift to fix, not a habit to copy.
2. **His shortest posts are his best.** Regular posts under 300 characters: median 12 reactions and 477 impressions (mean 18 and 2,011). Everything longer: median 9 and 301 (mean 13 and 701). 7 of 25 regular posts are that short.
3. **Proof beats personal.** Openings that lead with proof (years, a count, a client result): 5 posts, median 15 reactions and 794 impressions. Personal openings: 5 posts, median 6 reactions and 226 impressions. The two all-caps openings got 0 comments between them.

## Who's writing

A LinkedIn ghostwriter, nearly 7 years in, writing for founders who want booked calls, not applause. Uses Claude every day and argues about it in public: AI is fine, a generic thought is the problem. AI, Claude or slop comes up in 17 of the 30 posts.

## Rules you can check

| Rule | What the 30 posts show |
|---|---|
| Em dashes | 6, in 3 posts, all Aug 24-27. Write with none. |
| Fake bold letters | 0 in these 30. The two times he used them, earlier in August, both flopped. Never. |
| Opening line is a question | 0 of 30 |
| Regular post ends on a question | 0 of 25 |
| Opening line length | median 49 characters, 16 of 30 at 50 or under |
| Opening line in all caps | 3 of 30 ("THE ALGORITHM DOESNT HATE YOU!!!", "STOP:") |
| A word in caps for emphasis | 16 of 30 posts (SUCKS, NEVER, CLEAR, WITHOUT, DOESN'T GET) |
| P.S. | 4 of 5 lead magnet posts, 0 of 25 regular posts |
| "lol" | 5 regular posts, 0 lead magnet posts |
| Emoji | none in 21 of 30 posts. In lead magnet posts, 👇 at the ask |
| Lists in regular posts | `1)` (4 posts), `*` (3), `•` (3) |
| Lists in lead magnet posts | `→` in 5 of 5 |
| One line per paragraph | 228 of 257 paragraphs are a single line |
| Length | regular posts median 601 characters, lead magnet posts 1,138 |

## How his posts open

| Type | Posts | Median reactions | Median comments | Median impressions | His words |
|---|---|---|---|---|---|
| Flat claim, no setup | 6 | 8 | 3.5 | 640 | "Adapt the message to the medium." |
| Tool, news or R.I.P., then the fix | 6 | 16.5 | 39.5 | 1,011 | "Claude SUCKS at writing your Linkedin posts." |
| Proof first (years, counts, a client result) | 5 | 15 | 5 | 794 | "I've been ghostwriting for nearly 7 years." |
| Personal or story | 5 | 6 | 1 | 226 | "I'll NEVER use a voice dictation app." |
| Meme skeleton (pov, the flip, 99%, STOP/START) | 4 | 12 | 3 | 476 | "You'll see this and think:" |
| Whole line in caps | 2 | 7 | 0 | 48 | "THE ALGORITHM DOESNT HATE YOU!!!" |
| Unpopular opinion | 1 | 17 | 6 | 397 | "unpopular opinion: the anti-ai-slop button is just the shiny new thing..." |
| Rebuild scenario | 1 | 17 | 1 | 171 | "If I lost my LinkedIn account tomorrow, here's my next 30 days:" |

Flat claims have a median of 8 but a mean of 24.5, because two of them ("Adapt the message to the medium." and "Claude doesn't make your posts sound like AI. You do.") are his two best posts. When a flat claim lands it lands hugely, and most of the time it doesn't. 5 of the 6 tool and R.I.P. openings are lead magnet posts, which is why their comments are high.

## Lines only he would write

- "Adapt the message to the medium. / That's it." (Aug 10)
- "Claude doesn't make your posts sound like AI. You do." (Aug 22)
- "it's a tool. the thought still has to be yours." (Aug 22)
- "From my POV, I love it lol." (Aug 14)
- "Half of your competition is posting AI-slop, the other half isn't even showing up." (Aug 14)
- "I need to constantly reach 1,000 impressions per post if I want to be an account worth having 10,000 followers. / We're not the same." (Aug 18)
- "A good copywriter will 10x their productivity with AI. / A bad copywriter will produce 10x more AI slop." (Aug 19)
- "Social media is the only casino where the stakes are low, but there's no cap on the upside reward." (Aug 26)
- "KEEP POSTING FRIENDS!" (Aug 26)
- "Respectfully, LinkedIn, get over your superiority complex about not using AI to write." (Sep 2)
- "It's not even close." (Aug 24)
- "don't overcomplicate it." (Aug 18)

He leaves typos in ("I've wrote", "founder who know", "thougth"). Don't add new ones, and don't fix the ones he quotes.

## How his posts end

- **A flat statement that lands the point** (20 of 25 regular posts): "We're not the same." / "That's it." / "It's not even close." / "Funny how that works."
- **An order** (4): "don't overcomplicate it." / "PLEASE keep that in mind." / "KEEP POSTING FRIENDS!"
- **A sign-off** (1): "Cheers 🥂"
- **A question** (0)
- **Lead magnet posts** end on the ask, then a P.S. in 4 of 5.

## Regular posts vs lead magnet posts

| | Regular | Lead magnet |
|---|---|---|
| Median length | 601 characters | 1,138 characters |
| Lists | `1)`, `*`, `•` | `→` arrows |
| Emoji | none in most | 👇 at the ask |
| P.S. | 0 of 25 | 4 of 5 |
| "lol" and slang | yes | never |
| Median comments | 3 | 48 |

## Register

Blunt, operator to operator, competitive out loud. Drops register when the line earns it: "lol" in 5 posts, "af" in 1, all regular posts. A word in caps for emphasis in about half the posts. Warm rarely, and when it is, it signs off.

## What's working

| Post | Reactions | Comments | Impressions | Characters |
|---|---|---|---|---|
| "Claude doesn't make your posts sound like AI. You do." (Aug 22) | 61 | 26 | 6,913 | 519 |
| "Adapt the message to the medium." (Aug 10) | 59 | 11 | 6,667 | 44 |
| "This post made Jonathan over $240,000 in under a month..." (Sep 2) | 15 | 19 | 5,133 | 176 |
| "I've been ghostwriting for nearly 7 years." (Aug 14) | 21 | 7 | 794 | 805 |
| "You'll see this and think:" (Aug 18) | 19 | 7 | 1,000 | 262 |

At the bottom: "Just got LinkedIn premium." (2 reactions), "Copy a post and you get witch-hunted." (4), and the two all-caps openings (6 and 8 reactions, 0 comments each).

What the top five share: one claim someone could argue with, said in the first line, and nothing that needs explaining.

## Ban list

| Pattern | Line he rejected | Why it's not him | Added |
|---|---|---|---|
| "a room full of" | "That's not saturation. That's a room full of people not trying." | A picture used as a verdict. He never writes in images like this. | Aug 21 |
| /^That's not .+\. That's .+\.$/ | same line | His contrasts only work when both halves are literal. | Aug 21 |
| "that's the whole" | "That's the whole strategy." | Reads as AI to him, in every version: the whole trick, the whole point, the whole lesson. | earlier |
| "the gap" | any line built on "the gap is..." | A word he doesn't want in his posts. | earlier |

Not checkable by script, still banned: paragraphs that are exactly two sentences, lists of exactly three things, and names made of two jargon words the reader has to decode.

## Machine-readable rules

```rules
posts: 30
median_chars_regular: 601
median_hook_chars: 49
em_dash_posts: 3
fake_bold_posts: 0
question_hooks: 0
ps_on_regular: 0
no_emoji_posts: 21
```