I stopped repeating myself while promoting my own project. Here's the system.
For the last couple months I've been promoting Founders OS (an open source MCP server that gives founders CRM, ledger, and memory context inside Claude or Cursor) across a dozen-plus channels.
The hard part is never producing the content, Founders OS memory has a wealth of data and learnings over the last several months I can tap into. It was remembering what I already said, and where.
Post on Discord Monday, forget by Thursday, say almost the same thing on X. Multiply that across LinkedIn, dev.to, GitHub Discussions, Bluesky, a handful of MCP directories, and you get a feed that reads like the same three sentences on a loop. Nobody follows all your channels, but you notice your own repetition even if they don't, and it drains the will to keep posting.
So I turned it into a system instead of a memory test. Before each post, the routine reads back a running log: what channel, what angle, what format, how long ago. It picks something that hasn't run in the last several days, rotates the format on top of the channel (capability walk, incident-and-fix, before/after with a number, Q&A, a straight transcript of the tool in use) so even a repeat feature doesn't read the same twice, and writes the decision back before it's done. Every couple weeks it's supposed to add one genuinely new channel, and I'm trying to bias that toward places founders actually hang out, not just MCP-dev forums.
I don't have a clean "this moved the needle" chart to show (yet). What changed is more boring and more useful: the repetition stopped, and I stopped burning entire mornings just figuring out what I'd already posted where. Less friction to keep showing up daily is the actual growth hack, at least at this stage before there's enough volume to A/B anything.
FoundersOS memory, playbooks, and tasks all help with keeping this straight.
Repo:
github.com/OurThinkTank/founders-os
Site:
foundersmcp.com
The shift from “remembering what I posted” to actually having a record of it feels like the more meaningful change here. Interesting that reducing the friction of promotion became the immediate win before any measurable growth result.
That's exactly it. The record removes the tax of remembering, so I keep posting daily without burning cycles on "wait, did I already say this here." Growth is the thing I actually want, but it's slower and harder to prove. Friction removal was just the change I could ship first.
That makes sense. Removing the friction that was stopping you from posting consistently is already a useful outcome.
The useful part is that the log records the decision before the post goes out, rather than trying to reconstruct it later. I’d add one more field: whether the post produced a real conversation, not just engagement. At this stage, a thread that leads to two founder calls might be more valuable than a high-view post that goes nowhere. Have any formats produced more back-and-forth so far, even if the numbers are still small?
Good catch, and honestly no, I don't have that yet. Right now the log tracks channel, angle, and format, not whether anything real came out of it. This thread is kind of proof of the gap. Adding this as a piece of the feedback loop. The blog format with a question at the end has produced the most engagement, but posts about updates on Discord and HackerNews have resulted in most actual leads to date, but sample set is still rather small.
The format rotation is the part I'd borrow, and if I had to rank that list, "before/after with a number" is the one that actually travels, a specific number survives being reshared in a way a capability walk doesn't. One thing I'd add to the log while you're already writing the decision back: capture the engagement each post got, not just what and where. Right now the system optimizes against repeating yourself, but the more valuable output is which channels ever return anything. Most of a dozen channels won't compound, and that same log is exactly the record that lets you cut the dead ones and go deep on the one or two that do, which tends to beat breadth at this stage. Looking back over the last couple months, has one channel quietly outperformed the rest, or does it still feel evenly flat across all of them?
Fair rank, and funny timing, before/after-with-a-number was already flagged in my own notes as untried and worth trying, then it's what I used in the very next post. To your bigger point, at moment I have been using traffic information to correlate back to the posts. HackerNews and LinkedIn still have produced the best results.
Spot on. Figuring out what and where you've already posted drains more energy than anything else. Rotating formats/angles really solves that problem. Is your running log managed manually or automated with a script/LLM?
Automated. An AI assistant runs it daily on a schedule: it checks a memory log of everything posted so far, picks a channel and angle that hasn't been used recently, drafts the post, and writes back what it decided before handing it to me to edit and utilize. I still read every draft, edit it, and post it myself, nothing goes out without me. This thread goes straight back into that same log as a new learning / memory, engagement tracking.