How I Actually Automate My Social Media (And What's Not Worth Automating)

I Used to Batch Content on Sundays. It Was Miserable.

Every Sunday night I'd sit down with a blank Notion page and try to come up with a week's worth of social media posts. LinkedIn, X, Instagram. Sometimes I'd get through it in two hours. Most weeks I'd get halfway done, hate everything I'd written, and abandon it by Tuesday.

The posts that actually performed well? They were never from the batch. They were the random things I fired off in the moment because something genuinely annoyed or excited me.

So when people say "just batch your content," I get it. It sounds efficient. But for me it produced the most lifeless, forgettable posts imaginable. I needed a different system.

The Stack I Actually Use (And Why Most Tools Are Overkill)

I've tried probably fifteen social media tools in the last year. Buffer, Hootsuite, Later, Sprout Social, Metricool, Publer, Hypefury, Typefully. Some of them are good. Most of them solve problems I don't actually have.

Here's what I landed on:

Claude for drafting and ideation. Not ChatGPT. I've used both extensively, and Claude produces stuff that sounds less like it was written by a press release generator. ChatGPT is fine for brainstorming topic lists, but the actual writing quality from Claude is noticeably better. Fight me on this.

Canva for visuals. The AI features (Magic Design, background remover) are genuinely useful, not gimmicky. I make carousels in about 5 minutes now. The free tier is enough if you're just doing social graphics.

Buffer for scheduling. I tried Later, I tried Hootsuite. Buffer is simpler, and simpler wins when you're one person. The free plan covers three channels. That's all I need.

That's it. Three tools. Everything else I tried added complexity without adding results.

My Actual Weekly Workflow

I don't do the "Sunday batch" thing anymore. Instead, I have a rolling system that takes maybe 20 minutes a day.

Every morning, before I check email, I spend 10 minutes on X and LinkedIn just reading. Not posting, not engaging. Just reading what people in my space are talking about. If something sparks a reaction, I open Claude and say something like: "I just read a post arguing that AI-generated content is killing SEO. I disagree because [my reasons]. Write me a punchy LinkedIn post from my perspective."

Claude gives me a draft. I edit it. Sometimes heavily, sometimes just tweaking a line or two. Then I drop it into Buffer for that afternoon.

The key insight: I'm not asking AI to come up with ideas for me. I'm asking it to help me articulate ideas I already have. The difference in quality is massive.

Twice a week, I do a longer content session. Maybe 45 minutes. This is where I write threads for X or create a carousel for LinkedIn. For carousels, my process is:

1. Pick a topic I have a strong opinion on
2. Ask Claude to outline 7-8 slides with punchy headers and one key point per slide
3. Open Canva, use a template from my brand kit, paste in the copy
4. Tweak the design, export, schedule in Buffer

Total time per carousel: 15 minutes. They used to take me an hour when I was designing from scratch and writing the copy myself.

On Fridays, I check Metricool (the free tier is solid for basic analytics) to see what performed that week. I don't obsess over this. I'm looking for obvious patterns. Last month I noticed my "hot take" posts consistently outperformed my "helpful tip" posts by 3-4x on LinkedIn. So I shifted my ratio. Simple.

The Tools I Stopped Paying For

Hootsuite. Way too complex for a solo operator. It's built for teams managing multiple brands. If that's you, fine. But I was paying $99/month to use maybe 10% of the features. Buffer does everything I need for free.

Sprout Social. Beautiful analytics, genuinely impressive reporting. But $249/month? For social media analytics? I can look at my posts and see which ones did well. I don't need a sentiment analysis dashboard to tell me that my carousel about automation tools got more engagement than my text post about morning routines.

Opus Clip. I wanted to love this. The idea is great: you feed it a long video and it spits out short clips for Reels and TikTok. In practice, the clips it chose were rarely the most interesting moments. I spent more time sorting through mediocre clips than I would have spent just cutting the video myself. Maybe it's improved since I last used it, but I'm not going back.

ManyChat. The "comment GUIDE to get the free download" DM automation thing. It works. I've seen the data, it genuinely drives engagement and leads. But I hated how it felt. Every post becomes a lead gen trap instead of actual content. My audience started expecting a freebie with every post, and the comments section turned into a wall of people typing "GUIDE" and "LINK" instead of having real conversations. I pulled the plug after three weeks.

What I Wish Someone Had Told Me

AI-generated captions need heavy editing. Even Claude, which I think is the best option for writing, produces stuff that's a little too polished. Real social media posts have rough edges. Sentence fragments. Typos you leave in because fixing them would make it sound too corporate. I usually take Claude's draft and make it messier on purpose. Shorter sentences. Cut the transitions. Start mid-thought sometimes.

Consistency beats quality. I know this is cliche but I didn't believe it until I saw it in my own data. My account grew more in two months of posting daily (some posts mediocre, honestly) than in six months of posting "only when I had something great to share." The algorithm rewards showing up.

Engagement automation is a trap. I briefly experimented with auto-liking and auto-commenting tools. My engagement numbers went up. My actual relationships went down. People can tell when a comment is automated, even if it's a "good" automated comment. The only engagement worth automating is the posting itself. The replies and conversations should be you.

You don't need a posting schedule. Every social media guru says you need to post at optimal times. I tested this for two months. Optimal times vs. random times. The difference in reach was around 5-8%. Barely noticeable. Post when the content is ready. Don't hold a great post until Thursday at 2:15 PM because some tool told you that's when your audience is online.

The Honest Numbers

Before this workflow: I posted maybe 3-4 times a week across all platforms. Engagement was flat. I had about 2,000 followers on LinkedIn and 800 on X.

After three months of this system: I post 5-7 times per week. LinkedIn grew to about 5,500 followers. X grew to about 2,400. Not viral numbers. Not "quit your job" numbers. But steady, compounding growth that feels sustainable.

The time investment went from sporadic 3-hour Sunday sessions (that I dreaded and often skipped) to consistent 20-minute daily sessions that I actually enjoy because I'm responding to real conversations and ideas, not staring at a blank content calendar.

Monthly cost: $0. Buffer free tier, Canva free tier, Claude free tier (I use the paid tier for other work, but the free version is enough for social media drafting). Metricool free tier for analytics.

What I'd Do Differently Starting Today

If I were building this from scratch right now, I'd skip every tool except Claude and Buffer. I'd post only on one platform for the first month. Probably LinkedIn, because the organic reach is still insane compared to other platforms. X is noisier and the algorithm is harder to crack if you're starting from zero.

I'd focus on writing about my actual work and opinions rather than trying to be educational or helpful. The posts that grew my following fastest were the ones where I said something slightly controversial or shared a real failure. Not the "5 tips for better productivity" posts.

And I'd ignore every piece of advice about content pillars and posting frameworks. Those are useful for brands with marketing teams. If you're one person, your content pillar is "whatever you're thinking about today." That authenticity is your advantage over every brand with a content calendar and an approval process.

The Real Secret Nobody Talks About

The biggest unlock wasn't any tool. It was realizing that social media is a writing habit, not a marketing strategy. Once I treated it like journaling in public (here's what I'm working on, here's what I think about X, here's where I screwed up), the content came naturally and the growth followed.

AI tools make the execution faster. Buffer makes the scheduling painless. But the thing that actually matters is having something to say. No amount of automation fixes that.

Start there. The tools are the easy part.