# Long Form Launch Retro

> Post-launch breakdown (~800-1500 words). What you shipped, the numbers (good or bad), what worked, what didn't, what you'd change next time. Use after every product/launch/release where you have real numbers to share — including (especially) when the numbers are embarrassing.

- Skill: `r-bart/long-form-launch-retro` (Agent Skill)
- Install (CLI): `npx skillmds@latest add r-bart/long-form-launch-retro`
- Raw SKILL.md: https://api.skillmd.com/api/skills/r-bart/long-form-launch-retro/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: r-bart (https://skillmd.com/u/r-bart)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/r-bart/long-form-launch-retro

---


## Metadata

- **inherits**: `../writing-style/SKILL.md`
- **length target**: 800-1500 words
- **cadence**: per launch

# long-form-launch-retro

A retrospective on a launch. Numbers up front (especially when they're bad), what you tried, what worked, what you'd change. Honesty with numbers is the differentiator — most launches get written up as victory laps; this format works because it doesn't.

Voice and tone come from `writing-style/SKILL.md`.

## Structure

```
[Title: states what launched + the headline number. Example shapes (replace with your own):
  - "<product name> launched. €0 in sales. Here's what I learned."
  - "<product name> shipped to 8 free signups and zero paid. Notes."
  Don't bury the lede.]
[Subtitle: 1 sentence — what you're going to walk through.]

[Opener (2-3 sentences): when, what, who it was for.]

## What it is

[2-3 sentences. The product, the problem, the audience. Be concrete. No "all-in-one solution" language.]

## The launch

[What you actually did. List of channels, prep, timing. Bullets work here.]

## The numbers

[Cold, formatted, no padding:
  - Revenue: €X
  - Free signups: X
  - Coupons redeemed: X (per channel)
  - Traffic: X
  - Whatever else you tracked.
If a number is zero, write "Zero." Don't soften.]

## What I expected vs what happened

[1-2 paragraphs. What was your honest expectation going in? What actually happened? Where was the gap?]

## What worked

[2-4 specific things. NOT "the community was supportive" — name the channel, the post, the conversation, the conversion.]

## What didn't

[2-4 specific things. Same rule.]

## What I'd change

[2-4 concrete adjustments for next time. NOT lessons — adjustments. Things you'd actually do differently.]

## Next steps for [product]

[3-5 dated, concrete actions for the product going forward. Or: "I'm shelving it. Here's why."]

[Closer (1-2 sentences): forward-looking, declarative. Naming the next launch with date works well: "Product #2 drops on August 16th. Same approach, different itch."]
```

## Rules specific to this format

- **Numbers are the headline.** Including in the title when possible. Especially if they're embarrassing — that's the differentiation.
- **Expected vs actual is the heart of the post.** Without it, it's just a launch announcement. With it, it's a useful retro.
- **Specifics beat themes.** "The Discord post in #launches got 3 sales" beats "social media drove conversions".
- **No silver lining for the sake of it.** If it bombed, say it bombed. Trust comes from this.
- **Verdict on the product.** Every launch-retro must end with: continue / pivot / shelve. No suspense closures.

## Anti-patterns specific to this format

- "It wasn't about the sales, it was about the learning" — kill on sight. The sales matter AND the learning matters. Both, separately.
- Hiding zeros behind framing. _"Modest reception"_ when it sold zero. No.
- Lessons listed without ties to specific events from the launch.
- Soft closer like _"on to the next one!"_ — replace with a dated commitment.
- Three-adjective stacks describing the product. One adjective, earned.

## Workflow

1. Pull the numbers cold from your tools. Write them down in raw form.
2. Write your honest pre-launch expectation. Be specific (you wrote it down somewhere — find it).
3. Walk through what you did, channel by channel. Note results next to each.
4. Sort outcomes into worked / didn't.
5. Translate "didn't" into concrete adjustments for next time.
6. Decide the verdict: continue, pivot, or shelve.
7. Run `editing-pass` and `ai-slop-check`.