Social media automation: what to automate and what not to

For a small business, social media is usually someone's job "when they have time". The result: three posts one week, then a month of silence, and messages answered two days late. Automation won't fix a lack of ideas, but it handles everything repetitive around them. Here's what's worth automating, with which tools, and where to stop.

What's worth automating

The rule is simple: the machine does what repeats identically, people do what needs judgement.

Scheduling: the free option

If you only post on Facebook and Instagram, Meta Business Suite schedules posts, stories and Reels on both for free. For more networks, a scheduler puts everything in one calendar. At Buffer, the free plan covers three channels with 10 scheduled posts per channel; paid plans cost $5 per channel per month (Essentials) or $10 (Team, with unlimited users).

For a business posting two or three times a week, that's usually all you need. The rest of this article is for cases where posts start somewhere else.

When a custom workflow is worth it

A workflow built in n8n or Make pays off when the content already exists elsewhere and just needs to reach social media:

• a new blog post becomes a post for each network;
• a new product in your store, or one back in stock;
• the daily special or the week's offer, from the file where you already write it;
• an event in your calendar, with a reminder the day before;
• a good Google review, turned into a post (with the reviewer's permission).

A typical workflow, with approval:

1. A new article appears on your site (the workflow picks it up from the RSS feed).
2. An AI model writes a version for Facebook, Instagram and LinkedIn, following a tone guide you provide.
3. The drafts arrive by email, Slack or Telegram with "Approve", "Edit" and "Reject" buttons.
4. Once approved, each post is scheduled for the best time on its network.
5. On Friday, a report of what went out and how it did.

Text drafted by the model usually costs under a cent per post. The rest is the build and hosting; details are in how much AI automation costs, and choosing a platform in n8n vs Make vs Zapier.

Platform limits

Automatic publishing goes through the networks' APIs, which have rules:

• Instagram publishes through the API only to professional (business or creator) accounts and limits publishing to 100 posts per account in 24 hours; a carousel counts as one. For a small business the limit doesn't matter, but it shows API publishing is allowed and controlled.
• Facebook publishes to Pages, not personal profiles.
• LinkedIn publishes to company pages through approved apps; the big schedulers are already approved.

What's allowed nowhere: fake accounts that like your posts, automated comments on other people's posts, mass following. Every network blocks them and can suspend your account. In the US, the FTC's rule on fake reviews and testimonials also bans buying or selling fake followers and views for commercial purposes.

AI, labels and disclosure

Three rules about AI-made content:

1. Meta says it may require an AI label on Facebook, Instagram and Threads when a post contains photorealistic video or realistic-sounding audio that was digitally created or altered, including with AI; the rule doesn't apply to still images. Meta may also add the label itself when a file carries AI metadata.
2. The EU AI Act has required, since 2 August 2026, that AI-generated images, video and voices that could pass as real be disclosed as such. If you sell into the EU, see our EU AI Act guide.
3. California's bot law (B&P Code § 17941) makes it unlawful to use a bot that hides its artificial identity to push a sale to people in California; a clear disclosure that it's a bot keeps you on the right side of it.

Text written with AI's help doesn't need a label. But a generated photo of "customers" who don't exist is misleading, label or not.

Messages and comments

Automatic replies are worth it for the questions that come in daily: hours, address, prices, the booking or order link. An AI assistant can answer them in Messenger, Instagram DMs or WhatsApp, provided it says up front that it's AI and hands over to a person when it doesn't know. How that works on WhatsApp is in WhatsApp chatbot for business.

What stays with a person: complaints, questions about a specific order and any public conversation where your business is on display. A wrong automatic reply under a negative review does more harm than no reply.

Messages contain personal data, so a workflow that collects them and sends them to an AI model needs care: keep only what you need, for as long as you need it, with providers that don't train on your data.

Where to start

1. Make a simple calendar: what kind of post, which day, which network.
2. Schedule a week's posts at once, in Meta Business Suite or Buffer.
3. Write a one-page tone guide: how you talk, which words you use, what you avoid. The AI will use it.
4. After a month, look at what repeats: that's the first workflow to automate.

More automation ideas for small businesses, beyond social media, are in AI automation examples.

We build n8n workflows for social media that start from your website, store or calendar, draft posts with AI in your business's voice and publish only after you approve. Details are on our n8n automation page.