The short answer
Analytics pages are designed to look thorough, not to be useful. The result is creators who can recite their follower count and impression totals but cannot say which of their last ten videos was actually good.
The five
1. Watch-through on the first three seconds
The percentage still watching at three seconds is the closest thing short-form has to a single quality score. Everything downstream — reach, follows, shares — is gated behind it, because platforms only push a video further if the first slice of audience stays.
If this number is bad, nothing else matters and no posting-time optimisation will save it. It comes from each platform's own native insights.
2. Saves and shares
Likes are cheap and mean roughly "I did not dislike this". A save means someone intends to come back; a share means they attached their own reputation to it. Both are far stronger signals of value, and both correlate better with sustained reach than likes ever do.
3. Follows per thousand views
Not follower count — the rate. This tells you whether the people arriving actually want more of you, which is the difference between a video that went viral and a video that grew an audience. A million views with a low follow rate means you made something people enjoyed once.
4. Posts published per week, actually
Boring and the most predictive of the five over months rather than days. Almost every account that stalls stalls because posting became irregular, not because the content got worse. It is also the only metric on this list you fully control.
This is the number a calendar is genuinely for — seeing an empty week coming while there is still time to fix it.
5. Your own best posting windows
Not a generic chart — the hours *your* audience engages, from your own published history. Worth checking quarterly rather than obsessing over, and worthless until you have a few weeks of posts to read. We go into why the popular infographics disagree in this post.
What to ignore
- Follower count as a headline. A lagging indicator you cannot act on. The rate matters, the total does not.
- Impressions. Inflated, inconsistently defined between platforms, and rarely decision-changing.
- Engagement rate as one blended number. It mixes a like and a share into one figure and hides the only part that mattered.
- Daily fluctuation. Short-form performance is spiky by nature. Read weeks, not days, or you will chase noise.
Where to actually look
Three of the five — watch-through, saves and shares, follow rate — live in each platform's native insights, and that is the right place for them. What a scheduler adds is the cross-platform view: the same post compared across channels, per-post and per-account history over time, and your own best posting windows once there is enough history to mean something. Consistency is visible on the calendar itself.
Questions people actually ask
What social media metrics actually matter?
Watch-through in the first three seconds, saves and shares, follows per thousand views, how consistently you publish, and your own best posting windows. Everything else is either a lagging indicator or too noisy to act on.
Is follower count a vanity metric?
As a headline number, yes — it lags and you cannot act on it directly. Follows per thousand views is the useful version, because it tells you whether new viewers actually want more.
Why are likes less useful than saves?
A like is a low-cost reaction. A save is intent to return and a share is a small reputational risk, so both are stronger signals of value and track sustained reach more closely.
How often should I check analytics?
Weekly for performance, quarterly for posting windows. Short-form results are spiky day to day, so reading daily numbers mostly means reacting to noise.