Using LinkedIn Analytics to Find the Topics That Actually Land

Fifty published posts is a dataset about what your market responds to, and it is free. Most people either never look at it or look at the impressions column, which is the one that will send them in the wrong direction.

This is for anyone deciding what to write next month.

Compare the Right Thing

Rank your posts by comments from people in your target market. Not impressions, not reactions, not a blended engagement rate.

Comments require effort and a public position, which makes them the only engagement signal that means someone actually cared. And filtering to your target market matters because a post that got 40 comments from other marketers tells you nothing about whether benefits brokers found it useful.

This takes twenty minutes with a spreadsheet and LinkedIn's post analytics export. Do it once a quarter, not weekly.

Look for the Shape, Not the Topic

The common mistake is concluding "posts about deliverability do well, write more deliverability posts." Usually the topic is not what made it work.

Look at the top ten and ask what they share structurally. Were they disagreements? Did they contain a specific number? Were they about a failure? Did they describe a concrete situation rather than a principle?

In our own data the pattern is consistently structural rather than topical: posts making a claim someone would argue with outperform posts explaining something, regardless of subject. That is a far more useful finding than a list of topics, because it applies to everything you write next.

Sample Size, Honestly

Individual post performance on LinkedIn is extremely noisy. The same post published on two different Tuesdays can differ by a factor of five for reasons entirely outside your control, who happened to be online, what else was in the feed, whether one well-connected person commented early.

So do not draw conclusions from fewer than about twenty posts, and never from one. A single post outperforming by 10x is usually variance, not a signal, and restructuring your content strategy around it is the most common analytics mistake on the platform.

The useful question is not "why did that post do well" but "what do my top ten have in common."

The Comments Are the Research

The highest-value data in LinkedIn analytics is not in the analytics. It is in what people wrote.

Read every comment on your top posts and note the vocabulary. People describe their problem in words your marketing does not use, and those words belong on your landing pages, in your outreach, and in your keyword map.

Note the objections too. A comment disagreeing with you is telling you what the market believes, which is more useful than agreement, and it is usually the thing your sales team hears on calls without anyone writing it down.

Use Follower Demographics as a Correction

LinkedIn shows the job titles, industries and company sizes of the people following you and viewing your posts.

Check it quarterly against your ICP. Drift is the thing to look for: if you sell to benefits brokers and your audience is 40% marketing agencies, your content has been rewarding the wrong readers and the algorithm has learned from it.

The fix is to get narrower: more specific vocabulary, more specific examples, more assumed context. Broad posts recruit broad audiences, and a broad audience is expensive because it suppresses your reach to the people who matter.

What to Do With the Answer

Pick two structural patterns from your top ten and write to them for the next quarter. Not topics, structures. "A disagreement with a number in it" is a repeatable format; "deliverability" is a subject you will exhaust in a month.

Then re-run the analysis next quarter and check whether the pattern held. That is the entire loop, it takes an hour four times a year, and it is more than almost anyone does.