Cosmetic KPIs vs. revenue

Impressions, likes, and reach all measure the same thing: attention. They're easy to report, easy to feel good about, and easy to grow. They also tell you almost nothing about whether the content paid for itself.

This isn't a new complaint, but it persists for a practical reason. Attention metrics are abundant and immediate - every platform hands them to you for free, in real time. Revenue metrics require connecting content to an outcome that happens later, somewhere else, often in a different system. So teams optimize what's in front of them, and the dashboard slowly drifts away from the P&L.

The same open rate, two very different outcomes

Consider two email sends with an identical 24% open rate. On a vanity dashboard they're twins. In reality, one went to a persona with high purchase intent and a message matched to what that group actually cares about, and produced several dollars of revenue per send. The other went broad, got opened out of habit, and produced almost nothing.

Open rate couldn't distinguish them, because open rate was never measuring the thing that mattered. Revenue per send could - instantly, and without argument.

Score the channel on what the channel is judged on

The fix isn't to abandon attention metrics. It's to stop treating them as the scoreboard. Every channel already has a metric that survives a finance review, and it's rarely the one the platform surfaces first:

Search advertising is judged on conversion rate and cost per acquisition. Email is judged on revenue per send. Programmatic is judged on cost per acquisition. Organic search is judged on qualified traffic that converts, not raw sessions. When content is scored against the metric the channel is actually accountable for, the prioritization changes - sometimes dramatically.

Attention still matters - as an input

None of this makes reach worthless. Attention is a real input: you can't convert someone who never saw the content. The error is treating an input as an outcome, and then optimizing the input in isolation until it stops correlating with anything downstream.

A useful test is whether a metric would survive being read aloud in a board meeting without a follow-up question. "We generated four million impressions" invites "and what did that produce?" "We generated $340,000 in attributable revenue at a $42 acquisition cost" does not. The second number ends the conversation because it already contains the answer.

What this looks like in practice

A revenue-rooted scoring system won't let a post pass just because it got a lot of likes, even if none of those likes came from anyone your CRM would recognize as a customer. It won't celebrate a traffic spike from an audience that never buys. And it won't let a high-performing piece go unnoticed because it happened to be quiet on social.

That discipline is easier to state than to run, because it requires the content system to know who the content was for and what happened after it shipped. Most stacks can't close that loop - the writing tool doesn't see the revenue, and the analytics tool doesn't see the brief.

Traffic is not the trophy. Rankings are not the trophy. AI mentions are not the trophy. Revenue is - and a scoring system built by marketers, not engineers, was never going to mistake one for the other.