For six months I tracked 17 social media metrics in a Notion dashboard. Engagement rate, follower growth rate, average shares per post, share of voice, response time, six platform-specific subtotals, and a custom "health score" I made up to feel productive.

By the end I checked the dashboard once a week, glanced at it for ten seconds, and made exactly zero decisions because of it.

Then I deleted everything and kept four numbers.

The genre is bloated on purpose

Hootsuite's "must-track metrics" guide lists 21. Sprout's runs past 7,000 words across 8 categories. Asana lists 16. Every cornerstone analytics post is itself the disease it claims to cure: a giant pile of numbers that exist so the page can rank for "social media metrics."

A solopreneur or small team has a different problem from an enterprise marketing department. You aren't defending KPIs to a CFO. You need to know what to write next week, who to reply to today, and whether the thing you tried last month worked.

If a number can't change what you do on Monday, it isn't a metric. It's furniture.

So here are the four that pass the test. Drop the rest.

1. Saves per post (not likes)

A like is a thumb-twitch. A save is a person deciding "I might want this again."

Saves are the closest thing social media has to a bookmark, and they predict reach better than any other engagement signal because the algorithm reads them as "this was useful enough that the user wanted it back." Instagram, TikTok, LinkedIn, and Threads all expose saves in their native analytics. Bluesky and Mastodon don't have a save action, so there's nothing to miss.

What to do with the number: every two weeks, look at your top three saved posts and your bottom three. The pattern shows up fast. People save how-to lists, frameworks, and strong opinions. They don't save announcements. Stop posting the announcements and your save rate roughly doubles inside a month.

2. Your reply rate (your replies, not theirs)

This is a control metric, not an outcome metric. It measures what you do, not what your audience does, and that's exactly why it works.

Buffer analyzed 52 million posts across 10 platforms in 2026. Their finding: creators who reply to comments outperform those who don't on every platform studied. Not most. Every one.

Your reply rate is the answer to: out of every 10 substantive comments you got this week, how many did you reply to? If it's under 7, that's the cheapest growth lever available to you. Replying takes 30 seconds, requires no new content, and the algorithm reads your reply as a second engagement, pushing the original post out further.

Track this manually for a month. If you find yourself wanting to skip the count, that's already the answer.

3. Click-through rate to something you own

Likes, shares, and saves all happen inside the platform. None of them prove you have a business. The number that does is the click-through rate to a destination you actually control: your newsletter, your site, your store, your booking page.

You don't need expensive software for this. Use UTM tags. Every link in every bio, every link in every post, gets a UTM. Then on Friday morning, open your destination's analytics and count clicks per source.

A working benchmark: if you have under 1,000 followers and you're driving more than 10 clicks per week to something you own, you're ahead. If you have 50,000 followers and you're getting under 50 clicks a week to anything outside the platform, you have a community that doesn't trust you with their attention off-platform. The fix isn't more posts. It's writing differently. Specifically, writing about a problem your owned destination solves, not generic content adjacent to it.

4. The 2-week return rate

This is the metric almost nobody talks about and it's the most important one if you're trying to build a community instead of a megaphone.

It's the count of unique people who appeared in your comments, replies, or DMs at least twice in the last 14 days. Not how many people engaged once. How many engaged twice.

No platform exposes this directly, which means you'll have to count it by hand or with a small spreadsheet. That isn't a bug. It's a feature, because counting by hand forces you to look at people instead of aggregates. After one month of tracking, you'll know your top 30 by name. Those 30 are your business. Everyone else is weather.

When the algorithm shifts and your reach drops by 60% overnight, the 30 stay. Followers don't.

We built Broadr's post analytics around these four signals because every other dashboard we tried was 80% noise.

What to do on Friday morning

Open one tab with each platform's native insights. Note your saves, your reply rate (count it), the clicks to your owned link, and the names that showed up twice this week. Total time: under 15 minutes.

If a number moved, ask why. If three weeks pass and nothing has moved, change what you're posting, not what you're tracking.

The metric you can't act on isn't a metric. It's a number staring back while you scroll past it.