Every day at 20:10, posting to an empty room
I run four Facebook pages to promote my apps. A post goes out automatically on each one every day at 20:10. All four pages have zero followers.
The pipeline itself was healthy. Posts succeeded every day and nothing showed up in the failure logs. The problem was that nobody was there to see them. On a page with zero followers you cannot even form a denominator to judge whether the copy is good or bad. The page posts have a verdict date of October 22, but I already doubt what a verdict without a denominator can tell me.
Do your automated posting channels record that a post succeeded, or that someone actually saw it?
My personal account, on the other hand, has 1.6k friends. Among my Meta surfaces it was the only one where organic reach actually happens.
The only surface that reaches anyone is the one you cannot lose
Once I decided to use my personal timeline, the constraints came first. No automated daily posting. If friends hide me from their feed or unfollow me, the surface itself is gone. The pages had nothing to lose at zero followers; the personal account was the one channel that did.
So the rules were:
- Once a week, one app per post. First person, from the point of view of the person who made it, ending with a request for feedback.
- Store apps only, because the only way to recover money is through the stores.
- I answer comments myself. No auto-replies with friends.
- Order by downloads and revenue. First Plotta, my top app by downloads which had just launched a subscription, then the fortune app that makes up 47% of revenue.
There was also a baseline. On April 28 I had posted an English portfolio post on the same account, written in the third person and listing 21 apps. It got 3 likes. This time I am testing the opposite: Korean, first person, one app.
I wrote down the verdict first
The window is October 3 to 31.
- Denominator: if reactions plus comments on a post reach 10 or more, there is reach. Below that, no verdict, because it is a reach problem rather than a copy problem.
- Primary metric: 3 or more Play installs with
utm_source=facebook_profile. Baseline is 0. - Futility: if after two posts there are 0 Play installs and fewer than 10 reactions, stretch the cadence to every two weeks and reconsider.
The Play links carry a referrer of utm_source=facebook_profile&utm_campaign=maker-story so they are counted separately from the page posts. The English post got its own maker-story-en. iOS installs cannot be attributed through this path. That gap already existed and it is still there.
Two easy levers I chose not to pull
Two items were left open from the start as needing a decision: inviting my friends to like the pages, and posting into groups. Both look like they would make the zero-follower problem go away in one move.
What would you do: invite 1.6k friends to a page with zero followers?
On October 10 I decided to do neither.
- Invitations: the page posts are in the middle of their October 22 verdict window. Sending invitations now would change the treatment mid-test and contaminate the verdict. It would also send a notification to 1.6k people at once, spending the trust of the only channel that actually reaches anyone.
- Group posts: there is a risk of account sanctions and no way to measure the effect.
If the October 22 verdict comes back short on denominator, I will reconsider invitations then.
Traps I hit while posting
- The profile had switched. The browser I use for automation was holding Facebook as the brand page profile. Had I posted as is, the maker post would have gone to a page with zero followers. I added a step to confirm the personal account in the profile switcher at the top right before posting.
- Video did not go up. An mp4 inserted through the browser extension's file_upload sat with a grey preview at Posting for more than three minutes and never published. Videos are now uploaded by hand or replaced with still frames pulled from the short-form video.
- The same bundle ID was a different app. The fortune app is Android only. On iOS the same bundle ID belongs to a separate app with a different name, so calling it the iPhone version would be false. I confirmed this with a store lookup.
- Free had a scope. What the Play listing actually says is that today's luck is free every day and that there is no account and no login. The other readings run on credits, so I did not write that everything is free. For Streaklet I likewise checked the iOS and Play listings for no account required, no ads, and on-device storage, and stayed within that.
After each post I re-read the timeline to confirm exactly one post, three images attached, and no duplicates. Date, app, link and post URL go into a jsonl ledger.
What has gone out so far
- October 3: Plotta in Korean (friends only), with three images added later. The same day, an English version went out publicly.
- October 5: the fortune app (friends only, three Play listing screenshots). Originally planned for October 10 and moved up.
- October 12: Streaklet (both stores, friends only, three screenshots). Picked because it was second in Android installs in September (30), is on both stores, and is a general enough topic for friends.
The next post goes out after October 19.
Self-check
- Have you checked that every automated posting channel has a denominator, followers or reach, above zero?
- Do you cap posting frequency on channels you cannot get back once lost, like a personal account or a small group of real readers?
- Do you have a rule that holds back easy levers that would change the treatment while a verdict window is open?
The honest part
This post has no results. The window closes on October 31, and this record does not yet contain reaction counts or Play install counts. The claim that the personal account beats the pages rests on circumstance, 1.6k friends versus 3 likes in April, not on a measured conclusion. My once-a-week rule also broke in the first week: two posts on October 3 in Korean and English, and another on October 5. The English post had its own campaign, but whether showing up in the same friends' feeds three times in three days led to hides or unfollows, I do not know. And iOS installs could never be counted in the first place.