The People Who Praise Your Writing May Not Have a Vote

On June 15, I opened an export from my WeChat Official Account dashboard and recalculated the previous twelve days.

Forty articles had received 11,116 reads. Recommendations accounted for 10,780 of them, or 97 percent. The harsher number was this: only eleven articles received any recommended traffic at all. The other twenty-nine showed zero.

People had told me the writing was good during that period. Those twenty-nine zeroes still made one thing plain. Praise can be sincere without determining the outcome.

Once you start making content, it is easy to confuse “this person has an opinion” with “this person has a vote.”

A friend may sincerely like the title. A peer may have a fair point when they call the angle shallow. But if the first person did not click or share, and the second is not among the readers the article is meant to serve, their comments describe personal taste.

Taste is worth hearing. It should not automatically set the direction.

The weight of feedback depends on whether it can change the result you are pursuing.

I used to sort feedback by volume. The more specific and confident someone sounded, the more they resembled an expert. That hides a problem: the people most willing to speak are not necessarily the people closest to the outcome.

Someone who never pays can give detailed lessons on pricing. Someone who never shares can repeatedly explain which headlines will spread. Their words may be reasonable, but they did not cast a vote on this result.

Speaking up has its own selection bias. People close to you are more likely to comment. People with strong emotions are more likely to comment. People who think something is merely fine usually finish reading and leave. The voices that remain in your ears are not a random sample.

That does not make comments useless. They can expose an awkward sentence, a misleading example, or a blind spot I cannot see. They measure one person's experience. They cannot be promoted into the overall result without further evidence.

Dashboard data is not the final truth either.

High recommendation traffic only shows that the platform distributed an article more widely, or that readers were more willing to open it at that moment. It does not prove that the idea was deeper or every sentence was correct.

Low recommendation traffic does not automatically make an article worthless. Distribution, the response of the first group of readers, and fatigue around a topic can all affect the result.

Condemning an article with one metric is the same kind of shortcut as accepting one compliment as proof.

The first question is what you are trying to learn.

If I want to know whether an article reached a larger traffic pool, recommendation sources are the judge. If I want to know whether the title delivered on its promise, completion and time spent are more useful. If I want to know whether readers carried something away, shares, saves, and comments come closer.

As for whether the article says what I truly wanted to say, there is no dashboard field for that. I remain responsible for it.

Not all feedback should be ignored, and not all feedback belongs in the same table.

That is what those forty articles ultimately left me. The 10,780 recommended reads showed that platform distribution set the ceiling on reach during that period. The twenty-nine zeroes showed that publishing frequently did not create an automatic bridge into the recommendation feed.

Neither number told me what to flatter. They only revealed who was deciding which kind of result.

This distinction matters to a creator. Editorial conviction and distribution performance can be examined together, but neither should replace the other. If I erase what I mean for the sake of traffic, the metric has exceeded its authority. If I claim to want reach but use “I like it” to avoid the dashboard, the author has exceeded his.

I now divide feedback into three rough layers.

The first is feeling: like, dislike, smooth, awkward. It helps me understand a particular person, but it does not represent every reader.

The second is behavior: opening, continuing, sharing, saving. People who say nothing still leave choices here.

The third is the goal. What should this article produce: reach, trust, or a clear statement of one judgment? The meaning of the same number changes with the goal.

Before publishing, I now write down one signal to wait for. If I am testing a title, I watch recommendation entry and opens. If I am testing usefulness, I watch saves and shares. If the article exists to preserve an honest judgment, I do not let traffic rewrite it afterward.

This does not make the next result better by itself. It prevents one familiar trick: waiting for the data, then choosing whichever explanation feels most comfortable.

When someone says, “Good piece,” I am still pleased. I just do not immediately record that as proof that the direction is right. Criticism gets the same treatment. I keep it, but I do not rebuild everything on command.

I would rather wait until the next article is out and see whether the signal that actually matters has moved.

Applause makes a sound. Votes often do not.