Contents7
Thursday evening, the person who runs the company page realises nothing has gone out for two weeks. The new article is waiting, next week’s trade show has not been announced, and the last post dates from before the holidays. It is not a lack of ideas. It is a lack of rhythm, because posting always comes after everything else.
This guide extends our guide to marketing automation for small businesses, which places social media among the other marketing tasks. It is written for teams of 20 to 500 people who already post, or would like to, without spending their evenings on it. And to know what those posts bring in, not just how many people see them, our digital marketing in one block offer links every contact to its source in the CRM.
In short
Whatever repeats can be scheduled: posts prepared in advance, sharing every article or product, sorting incoming messages.
The networks forbid bots that like, comment or write messages on your behalf; scheduling with their own tools remains allowed.
AI suggests versions of a message already approved; a person chooses the one that goes out and replies to comments.
Every link you post carries campaign parameters, otherwise nobody will know what your posts brought in.
Why automate your social media?
Because social media is part of daily life for most companies, and the hardest part is not writing a post: it is getting one out every week, without fail, for months.
of EU enterprises with at least ten people use social media
of small enterprises, with 10 to 49 people
among medium-sized enterprises, with 50 to 249 people
of large enterprises, with 250 people or more
The share has risen by more than 21 points since 2016, according to the same source. Companies use it for their image and marketing, to listen to customers, to communicate or to recruit.
Automating does not mean posting more. It means that what has been decided goes out on time, even in the week nobody thinks about it, and that the team keeps its time for what a tool will never do for it: replying, and choosing what to talk about.
What to hand to a tool, and what to keep?
The line comes down to one question: if a mistake showed up in public tomorrow, who would answer for it? A scheduled post follows a rule the team wrote and reviewed. A reply to a comment commits the brand to a conversation, and a mistake there is seen and shared.
What can be handed to a tool
Scheduling the month’s posts, on the day the team prepares them
Sharing every new article or product, with a tagged link
Sorting incoming messages by topic and alerting the right person
Suggesting, with AI, several versions of a message already approved
What stays with a person
Choosing the version that goes out
Replying to comments and messages
Approving every figure, every promise and every mention of a customer
Pausing a scheduled post that the news has made inappropriate
These rules fit in each network’s own scheduling tool, in the usual publishing tools, or in an automation tool such as n8n, which we run on our own server. We compare these options in n8n, Make or Zapier: which one for a small business?.
What do the networks forbid you to automate?
Scheduling a post is allowed, and often built into the network itself. LinkedIn lets you schedule a post between 10 minutes and 3 months ahead (LinkedIn help on scheduling posts). What it forbids can be read in its user agreement, among the things every user agrees not to do:
Use bots or other unauthorized automated methods to access the Services, add or download contacts, send or redirect messages, create, comment on, like, share, or re-share posts, or otherwise drive inauthentic engagement.
The word that matters is “unauthorized”. Scheduling with the network’s tool, or with a tool that publishes through its official access, stays within the rules. A bot that likes, comments or sends invitations in an employee’s name does not, and it is that person’s account that is exposed: the same agreement reserves LinkedIn’s right to restrict, suspend or close an account that breaches it.
Each network has its own terms. Before connecting a tool, read those of the network concerned, looking for what they say about automated methods.
What can AI write, and where should it stop?
AI provides three real services on social media: adapting the same message to several lengths or tones, summarising an article into a post, suggesting a hook from a text already reviewed. It replaces neither the substance nor the review, and the networks say so themselves. LinkedIn, which offers its own AI writing features, warns in its terms that generated content “might be inaccurate, incomplete, delayed, misleading or not suitable for your purposes”: it asks you to “review and edit such content before sharing with others”, and the user remains responsible for what they post.
Three rules avoid unpleasant surprises.
- AI starts from an approved text. You give it the article, the announcement or the product page already reviewed; it adds no figure, no promise and no customer name.
- A person chooses the version that goes out, and rereads it in full, links included.
- Nothing replies on its own. A reply suggested by AI waits for approval, especially under a post that upsets people.
Should you then disclose that a post was written with AI? The answer depends on what you publish: we cover it in our article on text written by AI, and it is summarised in the frequently asked questions below.
How do you know what your posts bring in?
Follower counts and likes are reassuring, but they do not say what brings in business. To find out, every link you post carries campaign parameters (source, medium and campaign), and the form that creates a contact in the CRM copies them onto the contact record. The day the deal is signed, you know it came from a post, and which one. The figures to track and how to connect the tools are in our guide to automated marketing reporting.
From the field
The pipeline we built for our own posts
Each post is a file, reviewed like a blog article before it goes out. Approving it means publishing it: a program sends it on its date, first checks that the article it shares is actually live, rereads our latest posts so as never to publish the same one twice, and stops at two a day. A post more than seven days late expires instead of going out: better to stay silent than to share old news.
Where to start?
Four steps, in this order:
- List what gets published.
Note what comes back every week or every month: articles, new products, events, job openings. That list is what gets automated, not inspiration.
- Schedule with the network’s own tool.
Prepare the month’s posts in one session and schedule them in the network’s tool, before connecting any other.
- Tag every link.
Add the source, medium and campaign to every link you post, and check that the CRM records them on the contact record.
- Add AI last.
Give it variations of an approved text, never replies, and keep a person who chooses.
If your team does not yet know which tasks to hand to a tool, our digital and AI maturity audit reviews them with you and ends with a written GO or NO-GO, project by project. And to connect your posts to your sales, our digital marketing in one block offer tracks every contact through to the sale, with its source, and tells you every month what each channel brought in.
Frequently asked questions
Can you schedule LinkedIn posts in advance?
Yes, LinkedIn provides for it itself, up to three months ahead. What its user agreement forbids are bots and unauthorised automated methods that create, like, comment on or share posts, or send messages on your behalf. A publishing tool that goes through the network’s official access stays within the rules.
Can AI reply to comments for us?
It can suggest a reply, not publish it on its own. A reply commits the brand to a public conversation, and a mistake there is seen and shared. On LinkedIn, a bot commenting in an account’s name also falls under what the user agreement forbids. Keep AI for drafting a reply, and a person for sending it.
How often should you post?
No frequency suits every company. A rhythm the team keeps up all year is worth more than a flashy start followed by three months of silence. Start with what you know you can feed, for example every new article and every new product, then look in the CRM at what each post brings in before adding more.
Should you disclose that a post was written with AI?
For a commercial post reviewed by a person, generally not under the EU AI Act, which targets text published to inform the public on matters of public interest. Deepfakes, those images, sounds or videos that imitate real people, places or events, must always be disclosed. In every case, a person answers for what is published.
Sources
- LinkedIn help on scheduling posts (new tab)linkedin.com
Jim · founder of Brain Plus AI
Also in digital marketing
Practical guide11 min The topic guide
Marketing automation for SMEs: with or without AI
Emails, social media, SEO in several languages, support, reporting, catalogue: most of an SME's marketing tasks can be automated with simple rules. AI is justified at a few precise steps, and a person reviews whatever commits the brand.
Practical guide8 min
Automated marketing reporting: track what pays off
A useful marketing report does not copy out figures: every week it says what each channel earned, straight from the tools and unedited. The figures to track, how to connect Analytics, Search Console and your CRM, what AI can comment on, and the data pitfalls to know before you start.
Stepping back8 min
Should you disclose that a text was written by AI?
Transparency owed to the reader, or a detail of how the text was made, like a spell checker? The EU AI Act settles it only halfway, with a criterion that is useful well beyond the law: does a human answer for the text?