How to find your best performing posts
Your most viewed post is usually your oldest or your luckiest, not your best. Compare every post against your own middle instead. Take the middle value of your last thirty posts, divide each post by it, and the ones sitting well above the line are the ones that actually worked. Those are the only ones worth doing again.
Why is the most viewed post the wrong answer?
Because it is usually your oldest post, your luckiest one, or the one that happened to go out the week something else of yours was being passed around. None of those tell you what to do again on Thursday.
What you want is the post that did the most relative to what you normally do. That margin is the part caused by the post. Everything underneath the margin was caused by your account already existing, and you cannot repeat an account.
How to compare a post against yourself.
Write down the view count for your last thirty posts. A spreadsheet, a notes app, the back of something. Thirty is enough to have a shape and few enough to type.
Sort them and take the middle value. Use the middle, never the average. One post that went wide drags an average so far up that everything else looks like a failure.
Divide each post by that middle number. Now every post is a multiple of a normal day for you, and your account size has been cancelled out of the comparison.
Pick a line and hold it. Half again above the middle is a fine place to start. The point is that a line exists, not where you put it.
The answer you are looking for is not a number. It is the four or five posts sitting above the line, and the thing they have in common.
Which number should you divide?
Pick one and stay with it, because switching halfway is how people talk themselves into believing whatever they already wanted to believe.
Views or reach answers how far it travelled. Saves and shares answer whether it was worth watching, and they are the better number if you sell something, because a person who sends your post to somebody else is doing your selling for you. Likes answer almost nothing and are the easiest to get, which is exactly why they are the number everybody looks at.
If you only ever track one, track shares.
Finding the number on each platform.
| Platform | Where to look |
|---|---|
| Insights, inside the professional dashboard on the account. Content you shared, then sort by the metric you picked. | |
| TikTok | Analytics under the creator tools, in the Content tab. |
| YouTube | YouTube Studio, Analytics, Content. Sort the list rather than reading the summary cards. |
| Post analytics on each post, or the analytics section on the profile for the roll-up. | |
| Professional dashboard on the Page, in the content list. |
These menus move. If one is not where this says it is, the metric still exists, it has just been renamed again.
What do the winners usually have in common?
Almost never a topic. Nearly always a shape.
When people run this for the first time they expect to find that their audience likes a subject, and what they find instead is a length, an opening move, or a kind of promise. Four posts that all start with a flat statement of a problem. Three that are all under twenty seconds. Two that both showed a thing going wrong before showing it going right.
That is more useful than a topic, because a shape can be applied to anything you have to say next week, and a topic runs out.
What do you do with the winners?
Three things, in this order.
Put them back out once they are properly finished travelling, with a new opening. Then build the next month around the shape they share, because that is the only content plan you will ever have that was made out of your own evidence rather than somebody else's. Then make the best one easy to find: pin it, put it on the profile, use it as the thing a stranger meets first.
A post that already proved it converts strangers is the wrong thing to bury by date.
One caution about small numbers
If you have posted twelve times, this exercise tells you something about twelve posts and nothing about your account. Small samples are mostly noise, and the temptation is to rebuild everything around one post that did well for reasons that had nothing to do with the post.
Keep the spreadsheet anyway. It gets useful somewhere around thirty, and by then you will already have it.
Do this today
- Put your last thirty posts and their view counts into a spreadsheet, oldest to newest.
- Take the middle value, divide every post by it, and highlight everything half again above the line.
- Write one sentence naming what those posts have in common that is not their topic.
Short answers.
Why the middle value instead of the average?
One post that went wide drags an average so far up that everything else looks like a failure. The middle value ignores it, which is exactly what you want when you are hunting a pattern rather than an outlier.
How many posts do I need for this to mean anything?
Around thirty. Under that you are mostly reading noise, and the risk is rebuilding your whole approach around one post that did well for reasons that had nothing to do with the post.
Views, saves or shares?
Pick one and stay with it. Views answer how far it went, shares answer whether it was worth watching. If you sell something, shares are the better number, because the person sharing is doing the selling.
What do I do with the ones below the line?
Nothing. That is the point of the exercise. You are not trying to rescue them, you are trying to stop spending the same effort on that shape again.
Where to check this yourself.
Rules on these platforms move, and a guide is a snapshot of the day it was written. When something here matters to a decision you are making, read it at the source first.
Every guide here is free. So is the part of Cosmoose that does it for you.
Two channels, scheduling, and an alert the moment a post fails, with no card. When you want every channel, the repair engine and the rest, it is $49 a month and the founding hundred keep that price.