Niche accounts reward two things: consistency and a clear voice. Both are easy in theory. In practice, the people who run great niche accounts have day jobs, families, and the standard human ceiling on hours per week.
The pattern we see most often is: someone starts a niche account with energy in month one, publishes five posts a week, runs out of steam by month three, ghosts the account by month six. The audience never gets large enough to compound. The voice was there. The pipeline was not.
This post is about a different pattern: an account that we work with closely. We will call the person behind it Marc.
The setup
Marc runs a daily-history account on X and Instagram. One historic event per day, posted at 7am local time, with the same loose structure: a hook, three or four paragraphs of context, a primary source quote, a follow-on question. He has been at it for three years. He is a software architect at a mid-sized French insurer during the day.
When we onboarded Marc, the account had 12,000 followers, a consistent voice (slightly wry, never breathless, always sourced), and a posting cadence that was starting to slip. He was averaging four posts a week instead of seven. The week he reached out, he had posted twice.
What we fed bbuddy
Marc's library went into bbuddy as the brand:
- His last 200 posts, with the ones he likes most flagged as anchors.
- A list of preferred sources: three university press RSS feeds, the Wikipedia "on this day" feed, two podcast feeds he uses to spot leads.
- A voice document he wrote himself: tone, things to avoid, the specific structure of an opener.
- A no-go list: areas where he refuses to post (a few sensitive recent-history topics he considers out of scope).
Setting this up took ninety minutes of his evening. bbuddy was drafting daily posts the next morning.
The cadence today
Six months in, Marc's account is at 31,400 followers across channels. The structure of his week looks like this:
- Sunday evening, 20 minutes: he reviews the week ahead. bbuddy proposes seven drafts (one per day), each tied to an actual historical date. He approves, edits, or rejects. Rejections feed back into the brief.
- Monday to Sunday, 5 minutes per day: he checks the morning's post before bbuddy schedules it. He occasionally tweaks a phrasing. He almost never rewrites the post wholesale.
- Once a month, 30 minutes: he updates the sources list and reviews the no-go list.
Total time per week: about ninety minutes. Posts published per week: seven plus occasional follow-ups. Engagement rate: up versus the manual era, probably because the cadence is finally predictable.
Where Marc still does the work
This is the part that matters. bbuddy is not writing Marc's voice from scratch. It is generating a draft that matches Marc's voice well enough that his edit on top is short. The voice still comes from Marc.
The bigger Marc-specific work that bbuddy does not touch:
- Choosing which corners of history feel right to write about this season.
- Replying to comments and DMs. Marc reads every one, replies personally. The community knows him.
- Spotting threads worth turning into a longer essay on Substack. bbuddy flags candidates, Marc decides.
What we learned
Two things, both relevant to anyone running a niche account.
One: the bottleneck for niche accounts is not "having ideas". It is "writing a credible draft on a Tuesday night after work". Solving that bottleneck is what unlocks consistency.
Two: niche audiences are sensitive to voice drift. The moment a draft reads as generic, they bounce. bbuddy earns its place on accounts like Marc's only because the brand reading is thorough enough that the drafts read as the account, not as "AI wrote a history post". When the voice profile is shallow, the drafts feel that way. Marc's voice profile is deliberately deep.
If you run a niche account that is starting to slip, the interesting question is not "should I use AI". It is "what would I need to feed an AI for the drafts to read as me". Answer that, and the pipeline takes care of itself.