Rank your marketing on outcomes, not engagement

Open any social dashboard. The biggest number on it will be engagement rate, impressions, or follower count. Somebody chose to put that number in the largest font, and they did not choose it because it was the truest thing they could show you. They chose it because it was the easiest thing to collect.

That would be a harmless annoyance if dashboards were only ever read. They are not. They decide what gets made next.

Two posts, same week, opposite verdicts

Here is the problem in one table. Two posts from the same week, and which one you call the winner depends entirely on which column you look at. These are illustrative numbers, not research findings, but the shape is one every content team eventually recognizes.

Post A won the room. Post B won 19 customers off a twelfth of the reach.

Rank these by engagement and you will commission four more of Post A next month. You will also be able to prove, with a chart, that things are going extremely well. The chart will be accurate.

Inputs, outputs, outcomes

Every marketing metric sits in one of three tiers, and mixing them up is the most common measurement error there is. Inputs are what you spend. Outputs are what the platform hands back. Outcomes are what the business can actually bank.

Why the ranking metric is the only one that really matters

The metric you rank on is the one your program optimizes toward, whether you intended that or not. Which is why it matters far more than whatever else you put on the screen. Rank last quarter by engagement, feed the winners into next quarter's brief, and you have not built a learning loop. You have built a taste machine, and you have trained it on a taste that does not pay.

What makes this hard to catch is that engagement-optimized content is never visibly bad. It is topical. It is confident. It is mildly provocative in the way that gets quote-tweeted by people who will never buy anything. And it performs beautifully on the exact metric that selected it.

It just does not sell. Your dashboard will not mention this, because the number it is watching is going up.

What to do before you can prove revenue

Drop one tier and label the fallback wherever the number is read. Almost nobody can attribute revenue in month one, and pretending otherwise is worse than admitting the gap. Rank on the highest tier you can genuinely prove, and say which tier that is.

1. 1

   Get click attribution first

   Wrap outbound links so a click ties to the exact post that earned it. No product changes, no engineering ticket, and it moves you off pure engagement immediately.

2. 2

   Close the loop to signup

   Send a small conversion event from your post-signup step, or ask one 'how did you hear about us' question. Either one connects a click to a real account.

3. 3

   Attach value

   Once signups are attributable, revenue follows from your own pricing. This is the point where a content program becomes arguable in a budget meeting.

4. 4

   Set a threshold before you promote a tier

   Do not start ranking on conversions off four data points. Decide the minimum volume in advance and stay on the lower tier until you clear it.

5. 5

   Say which tier you are on

   In the dashboard and in every report. A reader who does not know the tier cannot calibrate the claim, and will assume the strongest one.

Attributed is not incremental

Attributed revenue means a customer touched a post before they converted. It does not mean the post caused the conversion, and the difference is usually large. Some meaningful share of those customers would have found you anyway, and no attribution model can separate them, because the thing you would need to observe never happened.

Proving incrementality means deliberately withholding marketing from a comparable group and measuring the gap. Most teams at content-marketing scale cannot run that experiment credibly. So report attributed revenue, call it attributed every single time, and never let it be read as lift.

The test to run on every number on your dashboard

Ask what you would do differently if it went up. If the answer is a concrete decision, the metric has earned its place. If the answer is that you would feel good, it is a vanity metric regardless of how sophisticated the calculation is.

• Impressions up 40%: what changes? Usually nothing, unless paired with a downstream rate.
• Click-through rate up 40%: the hook is working. Write more like that one.
• Conversion rate up 40%: the offer or the audience match improved. Go find out which.
• Followers up 40%: what changes? Nothing you could bank.

Keep the output metrics on the dashboard. They are how you tell a distribution problem from a persuasion problem, and a post with no impressions needs a completely different fix from a post with impressions and no clicks. The rule is not to hide them. The rule is to never headline one as a result.

Sheevook enforces this in the product rather than leaving it to willpower. As of 2026 it ranks on the highest tier a brand can actually prove, refuses to compute return on marketing investment when there are no conversions to compute it from (not computable, which is a different answer from zero), and labels reach and engagement as proxies everywhere they appear, including inside the prompts that generate the next round of content. A learning loop is only as honest as the metric feeding it.