Creative Patterns®
Connect Reddit Ads

001—Specimen / Home

Your ads in.
GPT prompts out.

You run Reddit Ads. We run the math you'd hire a data scientist for — and ship it as prompts your LLM already speaks.

Connect Reddit Ads

Decision Brief · Pattern Dossier · Signal Spotlight · Runbookreads in ChatGPT, Claude, Cursor

002Compatibility

Ships plain Markdown into the LLM you already trust.

No proprietary plugin, no model lock-in, no copy-paste rituals — every output is a paste-ready brief your assistant already speaks.

  • OpenAIChatGPT
  • AnthropicClaude
  • CursorCursor
  • GeminiGemini
  • GitHub CopilotCopilot
  • PerplexityPerplexity

06platformsone universal payload

003Method

You run the ads. We run the data science.

No analyst, no spreadsheet wrangling, no SQL. Connect Reddit once and every campaign feeds the pattern engine — out come killer, evidence-backed recommendations you can ship the same day in any LLM.

  1. 1You

    Run Reddit Ads

    Keep doing what you already do — launch campaigns and pay for clicks like normal. You bring the spend, the offer, and the audience.

  2. 2You

    Connect your account

    One-click Reddit Ads API auth (or drop in a CSV). We read your campaigns — no rebuilds, no manual exports, no copy-pasting ad names.

  3. 3Engine

    Ships killer recommendations

    The engine runs weighted CTR, lift, CPC efficiency, and spend share on every creative slice — then writes the specific play (clone this hook, pause that segment, test this variant) into a brief you paste into ChatGPT, Claude, or Cursor.

004Engine Vectors04 Metrics

Every recommendation cites these four.

No vanity metrics. No invented numbers. Each output below ties back to deterministic math the engine runs on your campaign — not LLM guessing dressed as analysis.

  • CTR lift

    Weighted slice CTR vs the campaign baseline. Positive = winning.

  • CPC efficiency

    Cheaper or pricier clicks vs the campaign average — positive is cheaper.

  • Spend share

    How much of the campaign actually rides on this slice — exposure read.

  • Evidence strength

    directional → promising → reliable → strong. Gated by ad count and impressions.

005Directive Outputs04 Surfaces

Four outputs. One engine. Zero guesswork.

Same evidence shows up four ways — on the notebook home, in every pattern dossier, on every ad signal, and inside the paste-ready runbook. Cited against the four engine vectors above.

01

Decision Brief

Notebook home · UploadPhoto Org

Live in your workspace

0.470% workspace CTR · 10 slices · 10 win · 0 under· Map · 10 wins · 0 under

Lean into

  • Headline shape — Question headline

    CTR 0.497% vs campaign 0.470% · CTR lift 5.7% · CPC eff. −0.3%

    6 ads · 248k impr

  • Headline shape — Statistic-led

    CTR 0.479% vs campaign 0.470% · CTR lift 2.3% · CPC eff. −1.4%

    6 ads · 71k impr

Watch

  • No losing slices

    0 underperformer slices in this notebook.

    More volume may surface drag

Next move

Clone top-performing question headline ads and test a new question variant, measuring for at least 500 additional clicks to confirm CTR improvement before scaling spend.

Hover map tiles for engine detail · Patterns to launch tests.

02

Pattern Dossier

Pattern Studio · Question headline

Live in your workspace
WinnerHeadline shapeQuestion headlineHigh evidenceReliable
Do this next

Clone the top-performing question headline ad and test a new question variant. Run until each arm clears ~500 additional clicks, then compare CPC-normalised CTR before scaling spend.

What we're seeing

Cluster headline_hook · question beats the campaign CTR baseline — weighted CTR 0.497% vs 0.470% (+5.7% lift) on a meaningful share of total spend.

Try next

Duplicate your top headline into one new audience with one controlled change (image OR first line, not both). Compare CPC-normalised CTR after the clicks floor.

03

Signal Spotlight

Per-ad drill-in · T1-Ad 2 · 7/9 levers

Live in your workspace
Signal · Headline shapeYou're here

"You led" headlines are marginally cheaper per click than the campaign average.

CTR vs baseline
+5.7%
CPC efficiency
+0.3%
Of campaign spend
23.2%
Do this now

"You led" headlines are working but CPC is marginally higher than the campaign average. Continue using this approach and test variations to improve cost efficiency.

Then testClone top-performing "you led" ad and pair with a question variant.

Each play comes straight from the numbers above.

04

Runbook

Paste-ready Markdown · /Runbooks/cpn-create-reddit-ads…md

Live in your workspace
# LLM EXECUTION COMMAND

When the evidence is locked, the engine writes the whole brief — opening with the same "do not invent metrics" preface every model needs, then Brand Memory, Inputs, Winning Patterns, Goal, Constraints, and the exact Output Format your assistant should return.

One click sends it into ChatGPT, Claude, or Cursor with the campaign context already attached.

See the runbook preview

006RunbookPaste-ready · .md

From evidence to paste-ready brief in one click.

The engine writes the whole runbook — opening with the "do not invent metrics" preface every model needs, then Brand Memory, Inputs, Winning Patterns, Goal, Constraints, and the exact Output Format your assistant must return. Same pattern engine as the Patterns tab, every time.

  • Create New Reddit Ads from Winning Patterns
  • Create Reddit Copy Only from Winning Patterns
  • Turn Winning Hook Into Meta Ad Variants
  • Create Image Generation Prompts from Winning Patterns
  • Create Video / Motion Prompts from Winning Patterns
  • Evidence-backed Reddit Import Sheet
  • Create Site Message Tests from Winning Patterns
  • Create Landing Pages Brief from a Winning Ad
  • Optimize from a Specific Pattern
  • Summarize Campaign Learnings for Future Tests
All 10 templates
Runbooks/cpn-create-reddit-ads-from-winning-patterns.md

Prepended when you open in ChatGPT

"You're helping me execute the creative-testing runbook below. Read it end to end, then produce the requested deliverable in the exact Output Format specified. Restate any numbers exactly as written — do not invent metrics."

# LLM EXECUTION COMMAND
You are a performance creative strategist. Treat this runbook as the
only grounded source for campaign metrics, named ads, and pattern lifts.
## Context
Creating new Reddit ads for UploadPhoto · campaign Selfie → ID 2026-05.
## Brand Memory
- @Brand/Brand.md
- @Go To Market.md
- @…/performance-learnings.md
## Winning Patterns (engine slices)
- Question headline (headline_shape) — +5.7% CTR lift · reliable · 6 ads
- Statistic-led (headline_shape) — +2.3% CTR lift · promising · 6 ads
## Output Format
Markdown table with columns: Ad Name | Audience | Hook | Headline | Body | CTA | Hypothesis,
then a fenced csv block ready for the Reddit Ads importer.

007NotebookWorkspace

One workspace. Every campaign. Every learning.

Markdown-native files, a Cmd-K command palette, structured ad inventory, and the pattern engine — all in one place, all under your control.

creativepatterns.app / pagestash · reddit-2026-04
⌘K
Workspace
▸Brand
▸Go To Market
▾Campaigns
▾Reddit · 2026-04
Mcampaign-brief.md
Cimported-report.csv
Mcreative-inventory.md
Mpattern-analysis.md
Mperformance-learnings.md
Mnext-tests.md
Mchatgpt-runbook.md
Creddit-import.csv
pattern-analysis.md
HIGH CONFIDENCE

"Permanent archive" messaging is outperforming generic save-page language.

Ads
3
Weighted CTR
0.56%
vs Campaign avg
+34%

Evidence. Three ads tagged promise = permanent_archive produced 0.56% weighted CTR across 9,822 impressions vs the campaign baseline of 0.42%.

Recommendation. Create variants that keep the archive promise but vary persona — OSINT, journalism, threat research.

Suggested
Find patterns0 credits
Generate next tests10 credits
Create ChatGPT runbook10 credits
Export Reddit CSV5 credits
Captures
Reddit AdsLIVE
12 ads · 3 ad groups · 8 previews

008Own Your Data.md · .csv · .zip

Everything the engine writes is yours to keep.

No lock-in, no proprietary format, no "export coming soon." Portable Markdown, CSV, or a single ZIP of the whole workspace — same files the app reads from.

  • .md

    Markdown tree

    Every file the engine writes — Brand, Go-To-Market, Pattern Intelligence, Runbooks, Campaigns. Plain `.md`, version-control-friendly.

  • .csv

    CSV inventories

    Creative inventory with metrics + tag columns, plus a Reddit next-batch scaffold sheet ready for the Ads Manager importer.

  • .zip

    ZIP archive

    Download the whole workspace as `.zip`, folder tree intact. Move to your storage, your repo, your designer — anywhere.

009PricingCredits, not seats

We sell creative credits, not seats.

One credit balance per account. Spend them when you analyze ads, generate runbooks, or export campaigns. If a paid job fails on our side, we refund the debit.

Founder Plan

Limited · 100 only

$49 once · 1,000 credits/mo for 12 months · then 250/mo forever.

New workspaces get 100 trial credits on signup. Request founder access below — we read every message. Sign-in stays a magic link.

Already have an account? Sign in

Free

Try the full loop on real data.

$0/ month
  • 100 trial credits at signup
  • 1 workspace
  • Find patterns + runbooks
  • Reddit Ads sync or CSV import
Start free

Starter

Steady testing volume for solo founders.

$12/ month
  • 250 credits / month
  • 3 workspaces
  • Full patterns + runbooks
  • Reddit Ads sync
Choose Starter

Pro

Popular

High-volume growth teams.

$29/ month
  • 1,000 credits / month
  • Unlimited workspaces
  • Advanced tagging
  • Multi-campaign comparison
Choose Pro

Studio

Agencies and multi-brand shops.

$79/ month
  • 5,000 credits / month
  • Team workspace
  • Multiple brands
  • Priority support
Choose Studio

010Colophon

Stop guessing. Ship the next play with evidence.

First 100 founders only. Lifetime credits, instant onboarding, killer recommendations dogfooded on real Reddit Ads campaigns — no analyst, no SQL, no spreadsheet rituals.

Claim founder access · $49 lifetimeSee how it works

Reddit sync → Decision Brief → Pattern Dossier → Signal Spotlight → Runbook → CSV / ZIP.