I Published 40 Posts. 29 Got Zero Recommended Reads.

A while back I was cleaning up the backend data for my WeChat account. I pulled the export into a spreadsheet to see how the posts I'd pushed out over the past month or so had actually done.

I scrolled to the “reads from recommendation” column. Out of 40 posts, 29 of them had a 0 in that cell.

I paused when I got to 29. What I'd been doing that month was basically “publish more, syndicate harder”: AI drafts the first version, a script posts on schedule, a syndication tool copies it to a few other platforms with one click. I'd started treating “how many did I ship this week, which channel did I hook up next” as the account moving forward.

That spreadsheet was telling me the moving-forward part was smaller than I thought.

First, how those 40 posts got made.

It used to be that publishing one post meant doing the layout myself, sitting around waiting for the right time to hit send, then copy-pasting into each platform one by one — two or three a day, tops. Over the last couple of years the tools took all of that over: AI writes the draft, a human passes over it once and it's ready; a script pushes it at the scheduled time; once one platform is done, a repost tool spreads it to a few others in seconds. Output really did multiply several times over. I won't pretend otherwise.

The logic behind it is simple: getting seen = number of posts × number of channels. So push both of those numbers up as high as you can.

The catch is, in that multiplication, the only things I can turn up by hand are “number of posts” and “number of channels.” The number that actually decides how far a post travels isn't in my equation.

What decides how far a post travels is the platform's recommendation feed.

Same spreadsheet: over that month or so, total reads were 11,116, of which 10,780 came from recommendation. Reads from existing followers opening it themselves didn't even reach the rounding error. Open rates on WeChat accounts have been sliding for years, and the platform long ago moved the bulk of traffic from “follower subscription” to “algorithmic recommendation” — which means it took the decision of “who gets seen” back out of the writer's hands and into its own.

The recommendation feed doesn't look at how many posts I shipped this week, or how many accounts I syndicated to. It test-pushes to a small batch, watches whether people in that batch finish the piece, whether they forward it or linger — if the numbers hold up it keeps amplifying, if they don't it cuts it off.

The “post more, syndicate more” levers I control myself have no socket on that valve. However high I stack the numbers on my side of the equation, the valve on the other side doesn't open an inch wider.

It reminds me of a line about selling wontons: if your wontons are good, ten customers are happy, and word spreads on its own; if the wontons aren't good, standing in the doorway shouting “but I shouted really hard!” gets you nowhere.

Running an account works the same way. When the content has substance, publishing more is an amplifier — one post takes off and drags a batch along with it. When the content has no substance, publishing more is closer to a mirror held up to it: the recommendation feed tests it on a few hundred people, sees a flat reaction, stops pushing, and the harder you publish, the faster that “no good” gets confirmed.

There is a time when publishing more is worth it: before you've figured out what your readers want, publishing more is running tests, spending a batch of posts to buy back the information of “which kind of topic gets picked up.” At that stage, volume is scouting the path. Still carpet-bombing after the scouting is done is mistaking the means for the end.

So these days I've put “figure out which topics get into recommendation” ahead of “ship two more this week.”

Next time the urge to publish more comes up, I'll pull that spreadsheet up first and take a look: of the last batch I sent out, how many actually got pushed, actually got read to the end?

If it's still 29 to 11, then the two extra posts are most likely just adding two more rows to that column of zeros.