R2 Media
Our own network of media brands. The publishing runs itself, and in three months it drew 99 million views across nine accounts.
- views in three months, nine accounts
- 99M
The problem
Posting to social media at scale looks like a quality problem and is actually a distribution problem.
Nine accounts, each needing something published several times a day, on Instagram, Facebook, TikTok and YouTube. Done by hand that is a full-time job per brand, and the ceiling is however many hours a person has. Through all of 2025, working that way, the network published 446 posts and drew 719 thousand views.
The harder part is that nobody can pick the winners in advance. We have the archive to prove it: of the posts published since the schedule went to full rate, thirty-eight percent drew under a thousand views, and the median post drew 2,322. Judgement applied one post at a time is judgement applied to a coin flip.
The approach
So the system is built for cost per attempt, not for polish per post. Content is generated, queued, and published on a fixed slot schedule without anyone opening a phone. The cheaper an attempt is, the more attempts the same budget buys, and attempts are the only input that reliably moves the total.
Frequency is decided per slot from how much work is waiting, because an empty queue posting on schedule just publishes filler. Twenty or more pending links — two full days at ten posts a day — and every slot fires. Below that, only the five core slots run and the half-hour slots are skipped, with no catch-up. A link added at midday raises that same afternoon's rate.
Underneath sits the boring infrastructure that makes nine accounts one operation rather than nine: a single sign-on and one shared roster, each application holding its own internal token scope so a token minted for one brand cannot act as another. An archive polls the platforms after publishing and stores views, reach and engagement per post — which is the only reason any of the numbers on this page exist as measurements rather than as claims.
The outcome
Between July and September 2026 the network published 1,522 posts across nine accounts and drew 99.1 million views, against 446 posts and 719 thousand views for all of the preceding year.
The shape of that total is the finding worth keeping. The top one percent of posts accounts for sixty-three percent of all views; the top ten percent accounts for ninety-five. One post reached 13.4 million on its own. Meanwhile the median post reached 2,322 people and was, in practice, invisible.
That is an argument against tuning and in favour of throughput. No amount of care applied to a median post turns it into the one that travels — but halving the cost of publishing doubles the number of chances at it. We would rather run a system that is honest about being a lottery and buys more tickets than one that pretends each ticket can be engineered.