Publishing a good blog post and sharing it once is the most common way small businesses waste content. The post took you two hours; the single social share gave it about six hours of visibility before the feed buried it. This guide covers the repurposing workflow I use with Claude to turn one article into roughly a month of platform-native social posts — without the copy-paste sameness that gets ignored on every platform at once.
Why Repurposing Beats Publishing More
Most owners respond to weak content results by producing more content. That is usually backwards. If a post is genuinely useful, the constraint is not supply — it is that almost nobody saw it. A single article typically contains six to ten distinct ideas, each of which can stand alone as a social post. Extracting them costs minutes and multiplies the return on work you have already done.
There is a search argument too, though it needs stating carefully. Social posts are not a direct ranking factor, and anyone who tells you links from Facebook boost rankings is overselling it. What distribution actually does is produce the conditions that lead to ranking: people who read your work, cite it, search for your business by name, and link to you from their own sites. Branded search and genuine citations are meaningful signals. Social is how you generate them.
Step 1: Extract the Ideas Before Writing Any Posts
Do not ask for social posts yet. First ask Claude to inventory what is in the article:
Here is a blog post I published. Extract every standalone idea worth sharing on its own — specific claims, counterintuitive points, numbers, mistakes, and steps. For each, note in one line why someone would stop scrolling for it. Do not write social copy yet.
You will typically get eight to twelve items. Delete the weak ones. What remains is your content calendar for the next few weeks, and because each item is a distinct idea rather than a reworded summary, the resulting posts will not feel repetitive to someone who follows you in more than one place.
Step 2: Write for One Platform at a Time
The single biggest mistake in AI-assisted social content is asking for “social media posts” and getting one bland block of text pasted everywhere. Platforms differ in ways that matter:
- LinkedIn rewards a specific opening line, a short lesson, and a plainly stated takeaway. Professional, not stiff. Links in the post are fine for most business pages.
- Instagram is visual first — the caption supports the image or video. No clickable link in the caption, so the call to action points at your profile.
- Facebook still works well for local businesses, especially in community groups, and tolerates longer, more conversational posts than people expect.
- X rewards a single sharp claim. Threads work when each line stands alone.
- Short-form video (Reels, TikTok, Shorts) needs a spoken hook in the first two seconds. Claude is genuinely useful for scripting these — ask for a fifteen-second script with the hook written as the literal first sentence.
Prompt per platform, and give Claude the constraint explicitly: “Write this for LinkedIn. First line must work as a standalone hook. Under 150 words. No hashtag block. Do not use the words unlock, leverage, or game-changer.” Constraints are what separate usable output from the generic default.
Step 3: Keep Your Voice Out of the Averaging Machine
Left alone, models drift toward a bland median voice — upbeat, hedge-free, faintly corporate. The fix is to show rather than describe. Paste three or four social posts you wrote yourself and liked, then:
These are my posts. Describe my voice in five specific rules — sentence length, how I open, what I avoid, how formal I am. Then rewrite the drafts above to follow those rules.
Save the resulting rules into a Claude Project so every future request inherits them. This one step does more for consistency than any amount of prompt tinkering, and it means the posts still sound like you six months from now.
Step 4: Schedule Around a Rhythm You Can Sustain
A workable pattern for one article per week, without burning out:
- Publish day — the article, plus one post announcing it on your strongest platform.
- Days 2 to 5 — one extracted idea per day, each posted natively with no link, purely to be useful.
- Day 7 — the strongest idea rewritten as short video or an image quote.
- Week 3 — revisit the best-performing idea with a fresh angle or a customer example.
Note that most posts carry no link. Posting value without always asking for a click is what makes the occasional link worth following, and it avoids training your audience to scroll past you.
Step 5: Close the Loop With Real Data
After a few weeks, paste your actual engagement numbers back into Claude and ask which ideas performed and what the winners have in common. You are looking for patterns you can repeat — a format, a topic area, an opening style — not a verdict on individual posts.
Feed those patterns back into your article topics too. If a specific idea consistently outperforms on social, it is telling you there is unmet demand worth a full post. That is a tighter feedback loop than most agencies ever gave their clients, and you are running it yourself in about twenty minutes a week.
What Not to Automate
Two things should stay human. Replies are the first: comments and DMs are where social actually converts for a small business, and templated responses are obvious and off-putting. Anything sensitive is the second — complaints, apologies, pricing disputes, anything involving a named customer. Draft with AI if it helps you get started, but read every word and make it yours before it goes out.
If you want the local-search side of distribution as well, the same extracted ideas feed directly into a weekly Google Business Profile posting workflow. And if you are still working out how to produce articles worth distributing, start with using Claude to write blog posts that actually rank.