A note on framing: this is a practical workflow guide, not a report of a formal, measured experiment. An earlier version referenced a "30-day AI subject line experiment" and presented time and output figures as if they were logged data from a controlled test. Those weren't measured results — they describe what a typical session looks like. The workflow itself is genuinely useful advice, cross-checked against Sprout Social's published guide to repurposing content for social media and Buffer's own AI Assistant documentation.
The real problem isn't content volume — it's leverage
You're not short on ideas. You're short on time to execute those ideas across six channels in six different formats that each have their own character limits, norms, and optimal posting times.
Most repurposing advice tells you to "just turn your blog posts into social content." That's not a workflow. That's a chore. What actually works, per Sprout Social's guidance on repurposing best practices, is selecting high-performing content deliberately and tailoring format to each platform — not cross-posting the same copy everywhere.
This is the system I run weekly, built around that principle. The rough output for a typical week: 1 blog post, 2 social posts per platform (X + Bluesky), 1 newsletter teaser, and a backlog of post variations for slower weeks. AI does the drafting; the editing and platform judgment stay with me.
🛠 Tools Used (all free tiers)
ChatGPT or Claude (free tier) · Buffer, including its AI Assistant (free, 10 posts per channel queue) · Your email platform's composer (Beehiiv, Mailchimp, etc.) · A simple notes doc or Notion page for your content backlog
Step 1: Write the article with repurposing in mind
The biggest mistake is treating the article as the final output and repurposing as a bonus step. Flip that. The article is your raw material. Every section should generate at least one derivative piece.
Before you write, identify three things:
- The one-line insight. What would you say if someone asked "what's the point?" in a lift? That's your X post hook.
- The most counterintuitive finding. The thing that surprises even you. That's your Bluesky contrarian take.
- The one thing readers will want to save. A stat, a framework, a checklist. That becomes your newsletter pull quote.
Flag these as you write. I literally put [X POST], [BLUESKY], [NEWSLETTER] in brackets inside my draft. By the time the article is done, the repurposing map is already built.
Step 2: Extract the social posts with a structured prompt
Once the article is published, paste the full text into ChatGPT or Claude with this prompt structure:
The prompt template:
"Here is an article about [topic]. Write me: (1) An X post of under 250 characters that uses the most surprising finding as the hook. No links in the body. Voice: practitioner sharing what they found, not a brand. (2) A Bluesky post of under 290 characters that challenges a common assumption in this space. Same voice. (3) Two backup variations of the X post using different angles. Label each clearly."
The character limits are tighter than the platform maximums. That's intentional. AI has a tendency to write right up to the limit, which often means padding. Tighter constraints produce tighter posts.
Review the outputs. Fix anything that sounds like a brand announcement. If it starts with "We're excited to share..." delete it and try again. It should sound like something a senior marketer would say on their personal account.
See our full library of ChatGPT prompts for social media for more templates you can adapt.
Step 3: Write the newsletter teaser — not a summary
Most newsletter teasers summarise the article. That's wrong. A teaser should create curiosity, not satisfy it.
The formula that works: hook in the first line (a stat, a claim, or a question) → one paragraph of context → one pull quote from the article → a hard CTA to read the full piece.
Prompt for the newsletter teaser:
"Write a 150-word newsletter teaser for this article. Tone: friendly and direct, like a colleague recommending something they found useful. Do not summarise — create curiosity. Open with a surprising finding or stat from the article. End with: 'Read the full breakdown →' followed by nothing else."
150 words is non-negotiable. Newsletter readers have low patience. If your teaser is longer, it starts doing the article's job, and the click-through rate will suffer.
Step 4: Build a post backlog per article
One article shouldn't generate one batch of posts. It should fuel content for 2–3 weeks. Different angles, different formats, different readers.
After the initial batch, I run one more prompt:
Backlog prompt:
"Based on this article, give me 3 more social post ideas I haven't used yet. Each should use a different format: one question (invite replies), one list with → arrows, one finding-first framing ('Here's what the data shows'). All under 250 characters. Same practitioner voice."
These go into a simple backlog doc. When I'm behind on content one week, I pull from the backlog instead of scrambling. A month of consistent posting is just a handful of articles' worth of material, strategically deployed.
Step 5: Schedule in one session, not daily
Batching is where solo marketers reclaim the most time. Don't post in the moment. Write and schedule a week of content in one sitting.
The sequence: publish article → run repurposing prompts → review outputs → queue everything in Buffer → done. In a typical session this whole process, after the article is already live, takes well under an hour once the prompts are dialed in.
On a free Buffer plan you get 10 posts per channel in the queue. With 2 posts per day per platform, that's roughly a 5-day runway — enough to stay consistent through the week without touching Buffer again until next week's article is done.
If you're new to building an efficient content workflow, this 6-step ChatGPT workflow for content marketing covers the production side in more detail.
The figures above describe a typical session using this workflow, not a measured or logged experiment — actual time and output will vary by article and how much editing you do.
What breaks this workflow
Three things consistently derail repurposing systems. Know them before you hit them.
Weak source material. AI cannot rescue a thin article. If your blog post is generic, your repurposed social content will be generic times six. Invest in the quality of the original piece, and repurposing multiplies that quality. Skip it, and repurposing multiplies mediocrity.
Bad briefs. "Repurpose this article" is not a brief. The prompts above are specific because specificity is the entire job. AI needs the format, the character limit, the voice, the angle, and the restriction ("no links in the body"). Remove any of those constraints and you'll get output you can't use.
Treating repurposing as a separate task. The system works because flagging [X POST] mid-draft takes two seconds. Trying to do repurposing as a separate project three days later takes far longer and happens less. Build it into the writing process, not after it.