The Social Metrics That Matter and the Ones That Flatter You

Cost per acquisition and what a customer is worth to you. Everything else is diagnostic at best and decorative at worst. If a report leads with impressions, reach or engagement rate, it is measuring activity rather than outcome, and activity is what you get when nobody asked what the money produced.

The uncomfortable part is that the flattering metrics are not lies. They are true numbers that answer a question nobody needed answered.

What is a vanity metric, exactly?

Any number that reliably goes up when you spend more, without telling you whether spending more was a good idea.

Impressions go up with budget. Reach goes up with budget. Clicks go up with budget. None of them can go down when you increase spend, which means none of them can tell you that increasing spend was wrong.

A metric earns its place if it can deliver bad news. Cost per acquisition can rise. Return on ad spend can fall. Those numbers are useful precisely because they are capable of telling you to stop.

That is the test I would apply to any line in a report: under what circumstances would this number make us do something differently? If there is no answer, it is decoration.

Which metrics actually matter?

Four, and they nest.

The relationship between the first two is the entire game. A $180 cost per acquisition is excellent if a customer is worth $2,000 over three years and catastrophic if they are worth $40 once. The number means nothing on its own, which is why benchmark comparisons are so misleading.

I went through why published benchmarks disagree so violently in the Meta ads comparison. The short version: they are measuring different account mixes and publishing a single average as though it describes your business.

What is the most common measurement mistake?

Optimizing for an event that is not the thing you sell.

Campaigns get set to optimize for landing page views because that event is cheap and plentiful, and the dashboard looks healthy as a result. Meta obligingly finds you the people most likely to load a page and leave, because that is literally what you asked for.

What you want is a custom conversion fired on the action that makes you money, with server-side tracking through the Conversions API behind it. That gives the system a real signal rather than a proxy, and it gives your reporting a number that corresponds to revenue.

This is the single highest-return fix in most accounts I audit, and it is invisible until someone asks what the conversion event actually is.

What about attribution?

Treat it as directional, and stop trying to make it exact.

Every platform reports conversions it believes it caused, using its own attribution window. Add up the conversions claimed by Meta, Google and your email tool and you will frequently find more conversions than you had customers. Nobody is lying. Each is counting touches it was involved in.

Two practices make this manageable.

Pick one source of truth for the number that matters, normally your own system rather than any ad platform, and use platform numbers for optimization rather than for financial reporting.

And watch blended cost per acquisition: total marketing spend divided by total new customers, across everything. It is crude, it ignores channel nuance, and it cannot be gamed by attribution settings. When platform-reported performance improves and blended cost per acquisition does not move, something in the attribution is flattering you.

What is incrementality and why does nobody measure it?

It is the only question that actually matters, and it is genuinely hard.

Incrementality asks what happened because of the ads that would not have happened otherwise. Retargeting is the clearest case: an ad shown to someone who already decided to buy will be credited with the sale, and would have happened without the ad.

The proper way to measure it is a holdout, meaning a portion of your audience deliberately not shown ads, compared against those who were. Meta offers conversion lift tests for this.

Most businesses never run one, for two understandable reasons. It requires deliberately not advertising to some people, which feels like burning money. And there is a real risk the answer is uncomfortable, particularly for retargeting and branded search.

I am not going to pretend most small accounts have the volume to run these properly. What I would say is that if your reporting shows a spectacular return on retargeting, that number is the most likely one in the whole report to be overstated.

What should a monthly report contain?

One page, four numbers, and a sentence.

Spend, new customers, cost per acquisition, and cost per acquisition compared to what a customer is worth. Then one sentence saying what you are changing next month and why.

That is it. Not thirty slides of platform screenshots. If someone wants the detail it should be available, but the report itself should be answerable at a glance, because a report nobody reads to the end is a report that never changes a decision.

Add trend rather than snapshot. A single month's cost per acquisition is noisy, especially at small volumes. Three months of it tells you whether something is actually happening.

How much data do you need before believing anything?

More than people act on, and this is where most bad decisions come from.

Meta's own delivery guidance is the useful anchor: an ad set needs roughly 50 optimization events a week before costs stabilize. Below that threshold, the variation you are seeing between campaigns is mostly noise, and reacting to it makes things worse.

Which means a small account should be changing things less often than it wants to. Turning off an ad after four conversions because its cost per acquisition looks bad is a decision made on nothing. I set out the volume arithmetic in the ad budget guide.

The discipline is uncomfortable: at low volume, patience beats optimization.

What about organic metrics?

Judge them on their own terms, and do not mix the reports.

Organic social is not a direct response channel and reporting it as one produces nonsense. Engagement rate is a legitimate organic metric because there is no budget lever making it go up automatically.

What organic is genuinely accountable for is whether it passes the credibility check, whether the audience is growing, and whether it produces the language that later works in ads. Those are real and mostly not numeric.

The one number I would watch is response time on messages, which almost nobody tracks and which directly costs local businesses customers.

What I see in reporting

Reports built to be defended rather than to be acted on.

The pattern is a monthly deck full of impressions, reach, engagement rate and click-through rate, every one of them up, and no line anywhere for what a customer cost. The report is unfalsifiable, which is exactly why it survives.

The second pattern is a cost per acquisition presented without any reference to customer value, so nobody in the room can say whether it is good. The number gets compared to a benchmark from a blog instead, which is comparing your business to an average of businesses that are not yours.

Rebuilding reporting so it can deliver bad news is usually the first thing the paid team at Inertia Digital Marketing does, and it is frequently the least popular.

If you have never checked the account end to end, the audit is the place to start, since a reporting problem is usually a symptom of something further upstream.

What to do on Monday

Open your ad account and find what conversion event your campaigns optimize for. If it is a landing page view, a click, or anything that is not the thing that makes you money, that is your entire optimization problem and it is fixable this week.

Then work out what a customer is worth to you over the life of the relationship. Not the first purchase. If nobody in the business can answer that, no cost per acquisition figure can be judged, and every conversation about ad performance is guesswork wearing a spreadsheet.

FAQ

Which social media metrics actually matter? Cost per acquisition and customer lifetime value, together. Neither means anything alone: a cost per acquisition is only good or bad relative to what a customer is worth. Return on ad spend and incrementality sit above them.

What are vanity metrics? Numbers that go up automatically when you spend more, such as impressions, reach and clicks. The test is whether a metric can ever deliver bad news. If it cannot, it cannot inform a decision.

Why do my ad platform numbers not match my sales? Because each platform counts conversions it was involved in, using its own attribution window, so totals overlap. Use your own system as the source of truth for financial reporting and platform numbers for optimization.

What is incrementality? What the advertising actually caused, as opposed to what it was credited with. Measured with a holdout group who are deliberately not shown ads. Retargeting figures are the most likely to be overstated without it.

How long should I wait before judging a campaign? Until it has enough volume to mean something. Meta's guidance is around 50 optimization events a week per ad set, so below that the differences you are seeing are mostly noise.

About the author

Adam Palmer is President of Inertia Digital Marketing, a digital marketing agency working with clients from small business to multi-location enterprise and franchise operations. Across 15 years he has managed more than $50 million in paid digital campaigns, and he writes here about what actually moves numbers rather than what sounds good in a strategy deck. Inertia built and maintains Social Fonts, which is free and always will be.