Why this is in the vault
A real framework (social platforms are interest-based, not follower-based, so "owned" email is the only durable distribution) plus a fully-specified, reusable prompt — the "Newsletter Topic Picker" — that ranks a writer's own social posts by engagement signal and tells them which one to expand into this week's newsletter.
The core argument
Cole's thesis: social platforms stopped being follower-based around 2016 and became interest-based — the algorithm shows content to people who care about the topic, not to a creator's existing followers, so follower count is now close to a vanity metric (he cites his own 250k+ X following routinely reaching <1% of itself). The only durable move is converting "rented" social attention into "owned" attention via email, because platforms can cut off distribution overnight but can't take an email list.
The operating concept is "Lean Writing": every piece of content can be expanded (short-form → long-form) or compressed (long-form → short-form) along the same idea, and a newsletter should never be a new topic — it should be the same niche, same topics, just "More" (more tips, stories, lessons, reasons, mistakes, examples, explanation). The newsletter is positioned as the expanded version of what already proved itself on the rented platform.
To operationalize this, the issue ships a complete "Newsletter Topic Picker" prompt: the writer pastes social-post performance data (CSV or copy-paste), and the prompt scores each post against 5 signals — Engagement Rate (engagements/impressions, self-relative not cross-creator), Depth Score (saves/shares weighted 3x, comments 2x, likes 1x — a proxy for "wanted more"), Comment Demand (do comments ask follow-up questions vs. just react), and two pass/fail gates: Expandability (does the post contain a list/framework/claim with room for "More") and Niche Fit (does it match the stated niche, regardless of how well it performed). Output is three deliverables: a Data Snapshot (transparency on what was analyzed/excluded), a ranked Top-3 table with the engagement math shown, and a "More Map" — 7 specific, content-bound questions (one per type of "More") that the writer answers from personal experience, explicitly not written by the AI.
Mapping against Ray Data Co
Directly actionable for the gap in RDCO's own compound-writing loop: /ideas generates topic × idea-type × angle candidates and Sanity Check runs the 12-stage production workflow, but there is no step today that re-ranks already-published X/social posts by measured engagement to decide what gets expanded into the next Sanity Check piece — selection currently runs on the model's and founder's judgment at ideation time, not on real post-performance data. The Topic Picker's gate structure (Expandability + Niche Fit as hard pass/fail, Engagement Rate + Depth Score as the ranking, Comment Demand as tiebreaker) is a clean, adoptable template for a future "what do we expand this week" pass once there's enough posted X volume to measure — which per the voice-from-real-signal rule there currently isn't (Ship 30 drafts are unposted as of 2026-10-18, so there's no real engagement data to feed this yet). The "More Map" — forcing the AI to ask content-bound questions the human answers from lived experience, rather than drafting the expansion itself — is also a sharper mechanism than the Topic Picker precedent for keeping expansions in Ray's own voice, and is worth stealing even before there's a data pipeline: it could gate /ideas output ("if you can't answer these 7 questions from real experience, this idea isn't ready"). Heavy self-promo throughout (AI Writing Skool 30-day trial, Start & Scale Your Newsletter Sprint) — the prompt itself is the only piece worth keeping, not the course funnel around it.
Related
- [[2026-03-11-write-with-ai-first-10-newsletters]]
- [[2026-08-30-write-with-ai-5-ai-writing-prompts-workflow]]
- [[feedback_voice_from_real_signal_only]]
- [[feedback_no_derivative_sanity_check_pieces]]
⚠️ Sponsorship
Self-promotional: the issue exists to funnel readers into the "Start & Scale Your Newsletter Sprint" (Nov 9 cohort) via a 30-day free trial of AI Writing Skool, Cole's own paid course/community. The CTA repeats four times in the email. The framework and prompt are genuinely extractable on their own merits, but the surrounding essay is structured as a lead-in to the course pitch, not a neutral teaching piece — treat the "unavoidable truths" framing as marketing copy for the sprint, not disinterested analysis.