Straight answer: you automate the timing, not the thinking. Write every post in your own words, batch them in one sitting, then let a scheduler handle the moment it goes live and the platform it lands on. The second you let a tool generate the words for you, or you set a queue and forget it exists, that is the moment your followers start smelling a robot.
Next step on this topic: How to Set Up Social Media Auto Posting Without Losing Your Voice.
The scheduling disaster that taught me the real rule
Back when I was building my following into the hundreds of thousands, I ran a three week queue through Hootsuite for a client’s launch. Good posts, written by me, spaced out sensibly. Then the London Bridge attack happened in June 2017 and the news was full of it within the hour. My queue did not know that. It kept firing a cheerful “3 days left to save 20%” post right in the middle of people’s timelines full of genuine fear and grief. Two people commented asking if we’d “read the news today.” One unfollowed with a public note about tone deaf brands.
Nothing terrible happened to the business. But I felt sick, and I never scheduled anything again without a plan for pulling it. That single afternoon taught me more about automation than any course I’ve ever paid for: the words being good is only half the job. Someone still has to be watching the world while the queue runs.
Automation is not the problem, absence is
Here’s the bit most people writing about this topic won’t say plainly: automation doesn’t kill authenticity, you not showing up kills it. A tool posting your words at 8am on a Tuesday is not fake. A brand account that never replies to a comment, never posts anything unscheduled, never reacts to what’s happening that week, that’s what reads as fake, whether the posts are automated or typed live in the moment.
I know accounts that hand write every single post live and still feel hollow, because nobody’s home in the comments and every post sounds like it was written by committee. I also know solo consultants running a full queue three weeks out who feel completely present, because they reply to every comment within the hour and post live, unscheduled reactions when something in their industry happens. Automation is a tool. Presence is the thing you can’t outsource.
The 70/30 split that keeps you human
Here’s the ratio I’ve used with clients and on my own accounts for years: 70% of your output can be scheduled ahead, 30% has to be posted live, in the moment, reacting to something real. If you post 10 times a week, that’s 7 batched posts and 3 that you write and publish the day of, based on a comment you saw, a news story in your industry, a client win that just happened, or a genuine thought that hit you in the shower.
Practically, that looks like this on a Monday morning:
- Block 90 minutes and write 7 posts in your own voice, in one sitting, while you’re in the mood and the tone is consistent
- Load them into your scheduler across the week, staggered at your best posting times, not all at 9am
- Leave 3 gaps deliberately in your content calendar, unlabelled, no draft sitting in them
- Fill those gaps through the week with something that happened that day, written and posted within the hour
That’s it. No fancy system. The 90 minutes on a Monday is the part people skip, and it’s the part that saves you the ten scattered 20 minute sessions you’d otherwise lose to “what should I post today” panic through the week.
Build a kill switch before you build a queue
Every scheduling tool worth using, Buffer, Hootsuite, Metricool, Later, has a pause button. Use it. My rule now is simple: if there’s a major news story, a national tragedy, or anything sombre trending in my country or my client’s country, I pause every queue for 24 hours and check again. It takes 90 seconds and it has saved me from the exact mistake I made in 2017 more times than I can count.
If you’re comparing tools and trying to work out which one fits your business, don’t just look at the feature list, look at how fast you can pause everything in a panic and how clearly it shows you what’s scheduled to go out today. I’ve written a full breakdown on how to compare social media auto posting tools without getting fooled by the feature list, and the pause and preview functions matter far more than the AI caption generator most of them are trying to sell you on.
What you should never let a tool write for you
This is the uncomfortable bit. Plenty of scheduling tools now offer to generate your captions for you. Don’t use that feature for anything that represents your opinion, your story, or your voice. It’s fine for a generic “here’s a blog we published” caption. It is not fine for anything where a real person is supposed to be speaking.
I tested this myself last year. I fed a tool three of my own old posts and asked it to write ten more “in my style.” The grammar was fine. The structure was fine. But every single one sounded like a LinkedIn influencer parody account, all confident bullet points and no actual opinion. Nobody who’s read my stuff for five years would have believed I wrote them. That’s the tell people miss: it’s not the timing that gives automation away, it’s when the words stop sounding like they came from someone with a specific point of view.
If you want the posts themselves to read as human even when the delivery is automated, the fix is in the writing, not the scheduling, and I’ve gone into that in more detail in how to auto post to social media without it feeling robotic.
Comments, hashtags and the parts you cannot automate anyway
Scheduling the post is the easy 20% of the job. The comment section is where authenticity gets judged, and no tool replies for you convincingly yet. I check comments on my automated posts within an hour of them going live, every time, because a scheduled post with no human replying to it for three days looks exactly like what it is: a brand talking at people instead of with them.
Hashtags get automated badly too. I see accounts using the same block of 15 hashtags on every single post for a year, pasted in by whatever tool they’re using, half of them irrelevant to that specific post. That’s a small thing but it adds up, because it tells a sharp follower that nobody looked at this post before it went out. If you want hashtags that do work rather than hashtags that just sit there looking busy, I’ve broken that down in how to write hashtags that get your posts seen.
Measure the right thing or you’ll fool yourself
Here’s a number worth knowing: when I audited a client’s automated Twitter (now X) posting last year, their post volume was up 40% year on year after switching to a scheduler, but engagement per post had dropped by nearly a third. More output, less connection, and they hadn’t noticed because they were only watching total impressions, which were climbing thanks to sheer volume. That’s the trap. Volume metrics will lie to you about whether automation is working, so track engagement rate per post, not just how much you’re pushing out. I walk through exactly how to check this in how to measure whether automated Twitter posting is working for your business.
A weekly rhythm that’s held up for me for years
If you want something you can copy this week, here’s the rhythm I run for my own accounts and recommend to clients:
- Monday: 90 minutes, write and batch 5 to 7 posts, all in your own words, all reflecting something you think, not generic tips pulled from nowhere
- Tuesday to Friday: 15 minutes a day checking comments on scheduled posts and replying like a person, not a bot
- One unscheduled post a week minimum, written the same day it’s posted, reacting to something current in your industry or your own week
- A five minute pause check on any day something major is happening in the news
That’s roughly two and a half hours a week, and it’s enough to run a business account that feels present, opinionated and human, while still giving you your time back. If you’re building this out for a small business and want someone to sanity check the whole set up rather than guessing, that’s exactly the kind of thing I help clients with through AI consultant support for small business, because most of the automation problems I see aren’t tool problems, they’re process problems.
The hard truth underneath all of this is that automation doesn’t save you from having a voice. It just saves you from retyping the same voice five times a week. If the voice wasn’t there to begin with, no scheduler on earth will invent it for you.
Frequently asked questions
Is it okay to fully automate a business social media account?
You can schedule most of it, but a fully hands-off account, no comment replies, no live reactions, no pause plan for bad news days, will read as abandoned within a few weeks. Keep at least 20 to 30% of activity, replies and unscheduled posts, live.
How many posts should be automated versus written live?
A 70/30 split works well for most small businesses: 70% batched and scheduled ahead in your own words, 30% written and posted the same day in reaction to something real. On a 10 post week that’s 7 scheduled and 3 live.
Can AI write my social media captions for me without it sounding fake?
It can draft a starting point, but hand it your opinion, not the other way round. AI generated captions tend to lose the specific point of view that makes people trust an account, so use AI for structure or a first draft, then rewrite the actual opinion yourself.
What’s the biggest mistake people make with scheduled social posts?
Setting a queue and forgetting it exists. Scheduled posts still need a pause plan for breaking news, a person checking comments daily, and regular audits of engagement rate, not just volume, to make sure automation is helping rather than quietly draining connection with your audience.
Related reading: 10 best tools to automate your social media marketing and Automate, Engage, Analyse, Repeat! – Social Media Tactics.
Official documentation
I go much deeper on this in the AI marketing guide.
