Social media analytics
Most social metrics are not decisions
Social media analytics is the practice of reading a small number of metrics that can change a decision, rather than reporting every number a platform exposes.
Social dashboards are generous with numbers and stingy with decisions. The test for any metric is simple and almost never applied: if this number moved twenty percent, what would you do differently? Most of what is reported monthly fails that test, and the reporting continues anyway.
Impressions are the weakest number you will be shown
Impressions measure distribution, which the platform controls, not resonance, which you control. They move for reasons that have nothing to do with your writing, and a post shown to many people who ignored it is not a success. Use impressions as a denominator for engagement rate, and almost never on their own.
Follower growth is only readable as a rate against a baseline
The raw count goes up for everyone who posts at all. What tells you something is the change in rate: how many followers per week now against your normal, and whether it moved after a change you made. A drop in the rate is a signal worth investigating; a drop in the absolute count is worth investigating urgently, because it usually means something else is wrong.
Replies and profile clicks beat likes
A like is close to free, so it measures very little. A reply cost someone effort, and a profile click means the content did the one job that leads anywhere: it made a stranger curious about who wrote it. Those two are the closest social metrics get to intent.
Attribute conversations, not conversions
Social attribution is genuinely hard and most claims of precision about it are false. What you can track honestly is conversations started and where they came from. If you can say that a third of your inbound conversations started as a reply, that is a real finding, and it is more defensible than a number pulled out of a model.
How to do it
Pick four metrics and delete the rest
A reasonable set: engagement rate, follower growth rate, profile clicks, and conversations started. Every other number goes into a folder you look at quarterly if ever.
Record a baseline before you change anything
Four weeks of normal. Without it you cannot tell an improvement from a good week, which is how most social experiments end up claiming victory.
Change one thing at a time
Format, timing, or topic. Change two together and the result tells you nothing you can reuse.
Review monthly, not daily
Daily variance on social is enormous and reading it daily produces reaction rather than learning. Monthly is enough to see a real change.
Watch the account health numbers separately
A falling follower count or collapsing reach is a warning about the account, not a content result, and it needs a different response: stop, then diagnose.
Common mistakes
- Reporting impressions as a headline number, which measures distribution rather than resonance.
- Reading daily variance as signal, which produces constant reactive changes and no learning.
- Claiming precise attribution for social, which is not credible and undermines the honest findings next to it.
- Changing several variables at once, so a good month cannot be repeated.
How Quillen handles social media analytics
Quillen tracks impressions, engagement, follower growth and profile clicks per post and per account, and shows them on a dashboard with period on period deltas rather than raw totals alone. Impressions come from X; LinkedIn through the third party provider does not expose them, so that column is empty for LinkedIn rather than estimated. Analytics are aggregated across the workspace and split by platform rather than broken out per connected account, and each chart can be exported as CSV. There is also account health monitoring that warns on a follower drop, and a daily report summarising what went out.
Common questions
- What is a good engagement rate on X or LinkedIn?
- Benchmarks published as single numbers are mostly useless because they vary enormously by audience size and niche. Your own trailing four week average is the only benchmark that tells you anything about your account.
- Why did my impressions collapse?
- Usually distribution rather than content: a change in how the platform ranks, a period of lower posting, or a shift in who engages early. Check whether replies and profile clicks fell by the same proportion; if they did not, the content is not the cause.
- Can I attribute revenue to social replies?
- Rarely with precision, and a tool claiming otherwise is overstating. What you can do honestly is track which conversations started where, and report conversations rather than a modelled revenue figure.