Impressions are the vanity metric everyone agrees is a vanity metric and everyone keeps reporting anyway. We have generated more than three billion of them across the network, which earns us the right to say something slightly uncomfortable about what they are worth.
They are worth something. They are just not worth what most decks imply.
The number that hides the answer
An impression counts a post appearing on a screen. It says nothing about whether the viewer read it, cared, or could name the brand afterwards. Two posts with identical impression counts can produce wildly different business outcomes, and the impression column will never tell you which was which.
The metric that consistently separated them, in our own campaigns, was the ratio between engagement and reach — and then, further down, the ratio between link clicks and engagement. A post can travel enormously and convert nothing. That is not a failure of distribution. It is a failure of intent.
Reach tells you the post travelled. Clicks tell you it mattered.
The practical version: if you only look at one number, look at how many people did something after seeing the post, divided by how many saw it. It is less flattering and far more useful.
What the high performers had in common
Going back through the posts that outperformed, the pattern was not what we expected. It was not production value. Several of the highest-performing pieces were visibly low-fidelity.
They were funny before they were commercial
The posts that travelled were, almost without exception, ones a person would have shared with no brand attached. The product was present but not the point. This is the hardest note to give a client, because it sounds like an argument for not marketing. It is the opposite: it is an argument for earning the attention before spending it.
They were native to the format
A video shot vertically for TikTok and cross-posted to X performs worse than the same idea framed for X. Not because the platform punishes it, but because the audience reads it as imported. Cultural proximity is a formal property as much as a tonal one.
They arrived at the right moment
Timeliness beat polish repeatedly. A rough post inside a live conversation outperformed a well-produced post arriving three days later, consistently enough that we changed how we staff around cultural moments.
They gave people a reason to reply
Engagement begets distribution. Posts constructed with an obvious opening for a reply — a question, a claim worth arguing with, a format people can riff on — compounded in a way that broadcast posts did not.
What did not correlate
Equally instructive, and worth stating plainly because so much agency practice assumes otherwise:
- Posting frequency, past a floor. Beyond a certain cadence, more posts did not produce more reach. They produced more posts.
- Hashtags. On X specifically, we have never been able to attribute meaningful lift to them.
- Follower count of the posting account, in isolation. A smaller account posting into an active conversation routinely outperformed a larger one broadcasting into its own timeline.
That last one is the reason we think about distribution as a network rather than an account. One large page is a single point of failure and a single audience. Several coordinated pages reach overlapping but distinct audiences, and can seed the same idea from more than one direction.
How to report it without lying
If you take one operational thing from this, make it the shape of your reporting. A defensible cannabis social report has four rows, in this order:
- Impressions — the ceiling. Context, not achievement.
- Engagement rate — did the content earn its distribution?
- Link clicks — did anyone act?
- Follower growth — did the audience compound?
Reporting impressions alone is how agencies survive quarters where nothing worked. Reporting all four is how a brand can tell the difference between a month that built something and a month that made noise. It is also why every client of ours can see all four themselves, live, in their own portal rather than waiting for a monthly slide.
The uncomfortable conclusion
Three billion impressions did not teach us how to make things go viral. Nothing teaches that reliably, and any agency claiming a repeatable formula is describing survivorship bias with a straight face.
What it taught us is narrower and more useful: you cannot control whether a given post travels, but you can control the odds, and you can control what happens to the attention once it arrives. Most brands are not losing at the first thing. They are losing at the second — sending hard-won attention to a neglected profile, a broken link, or a landing page that answers a different question than the post raised.
Fix the destination first. The creative is the easier half.
If you want the platform-level argument that sits underneath all of this, start with why X is still the highest-leverage platform in cannabis.
