LinkedIn Algorithm 2026: Why 838 Impressions Beat 93,000 (And How to Fix Your Audience)

If your LinkedIn reach is declining, or if you get plenty of impressions and no leads, you are not imagining it. The algorithm changed what it optimises for. The good news is that the fix is concrete, and it starts with reading your own numbers differently.

Below: the three posts that made this obvious, what 360Brew is actually doing, and the three-step framework for realigning your audience so you are speaking to buyers rather than to an audience of peers.

The Three Posts That Make the Point

These are three posts from a single creator, all published within a two-week window. The figures are from that account's own LinkedIn post analytics, tracked in ContentIn.

• Post A: 93,350 impressions | 174 likes | 68 comments | 15 followers gained
• Post B: 74,441 impressions | 287 likes | 75 comments | 19 saves | 29 followers gained
• Post C: 838 impressions | 56 likes | 58 comments | 13 followers gained

Read the last column, not the first. Post A reached more than 93,000 people and gained 15 followers. Post C reached 838 people and gained 13 — almost the same outcome from roughly 1% of the reach.

Post C also drew 58 comments against Post A's 68, on a hundredth of the audience. The comment-to-impression ratio differs by nearly two orders of magnitude.

This is one creator, so treat it as a worked example rather than a study. But it is the pattern we keep seeing across the accounts connected to ContentIn, and it is the reason the impressions number is the least useful thing on your dashboard.

What Actually Made These Posts Different

It was not the format, the hook, or the posting time. It was who the content was written for.

Post A: the wide-reach trap

Post A was universal leadership content — a piece about executives, team sports and what one shapes in the other, built on third-party research and relatable to essentially every professional on the platform.

The algorithm matched it to everyone who engages with leadership content, which is a very large group. It went wide because it appealed broadly.

The problem: those 93,000 people were not this person's buyers. They were other professionals who enjoy reading about leadership — coaches, managers, students, peers. People who will almost certainly never become clients.

Post C: the precision target

Post C covered daily, weekly and monthly habits for digital visibility in regulated sectors — specifically healthcare and financial services.

If you don't work in a regulated sector, that post means nothing to you and you scroll past. If you are a CEO or a marketing director in healthcare or financial services, it was written for you specifically. You engage. You follow. You might get in touch.

Only 838 people saw Post C. They were the right 838 people.

Why 360Brew Changed the Rules

LinkedIn's current ranking system is built on 360Brew, which LinkedIn's own research describes as a decoder-only foundation model that replaces a stack of separately engineered ranking models with one system that reads text — profiles, posts, interactions — and reasons about relevance directly.

The practical consequence is that it is a relevance engine, not a reach engine:

• Your profile text — headline, about section, experience, skills — is a primary classification signal
• The system reads it to decide who should see your posts
• Generic positioning gets you a generic, diluted audience
• Specific positioning gets you a specific, valuable one

If you have been posting generic leadership tips, motivational content or broad "five things I learned" posts, the algorithm has matched you to the people who engage with that kind of content. Most of them are other creators, coaches, consultants and peers. Not buyers.

The echo chamber compounds

The irony is that the better those posts look on the surface, the deeper into the wrong room they pull you:

1. You optimise for reach
2. The algorithm learns what drives your engagement
3. It shows you to more of those same people — the wrong people
4. Your audience fills up with non-buyers

Each cycle makes the next one worse. The flywheel runs in both directions, and by default it runs against you.

The Three Metrics That Actually Matter

Instead of watching impressions, watch three numbers that tell you something about audience quality.

1. Followers gained per post

The clearest signal that someone found your content worth following for. They want more of it. That means they are in the room you want to be in. It is also the column that separated Post A from Post C, and the one almost nobody looks at.

2. Profile views generated

Someone who clicked through to your profile was curious enough about you specifically to go and look. A reaction is passive. A profile visit is intent.

3. Saves

Someone decided this was worth coming back to. That is durable value rather than a scroll-by acknowledgement, and it is the signal most closely associated with content people intend to act on.

The ratio that matters is not impressions. It is how many of these three things you get relative to the impressions you generated. A post with 838 impressions and 13 followers gained is not a small post. It is a high-quality post that found the right room.

If you want the wider context on what a normal engagement rate actually looks like, we've published medians across 100,459 posts from 1,054 professional accounts — and the same point applies there: the median tells you what typical looks like, not whether the typical audience is yours.

The Three-Step Framework to Realign Your Audience

Everything below rests on one of two things: what we see in the data from the quarter of a million posts ingested through ContentIn — of which 100,459, from 1,054 accounts, form the analysable subset behind our published benchmarks — or on how 360Brew is documented to work. No speculation.

Step 1: Realign your profile so 360Brew classifies you correctly

This happens before you write a single post. Your profile is the anchor for everything else, because 360Brew reads all of it — headline, about section, experience, skills — when deciding who to show your content to.

What not to do: "Helping businesses grow through marketing."

That gets you classified as a generic marketing professional, and your content matched to people who follow generic marketing content. A diluted room for whatever specific thing you actually sell.

What to do instead. Your headline needs three elements: a specific audience, a specific problem, and a specific proof point. Not just what you do, but who you do it for and what they get.

The creator behind Post C is the example. His headline and profile point unambiguously at digital visibility for healthcare and financial services. When he posts niche content about AI discovery in regulated sectors, 360Brew already knows who to show it to.

Your about section is the second signal:

• Name your specific expertise
• Name your specific audience
• Include something only you could say
• Cut generic phrases like "passionate about helping people achieve their goals"
• If you have a methodology, put it in there

Then the skills section. Reduce it to the high-value skills you actually work with. Every signal is read; a long tail of stale skills is noise in the classification.

Step 2: Reposition your content from peer-facing to ICP-facing

This is the harder step, and where most people get it wrong.

The instinct is to post what gets reach: leadership insights, AI trends, productivity tips. Things that broadly resonate and occasionally go viral. Those posts get reach to the wrong people.

The reposition is simple to describe and difficult to execute: stop writing for everyone, start writing for your specific ICP.

Peer-facing: research on sport and leadership. Broad. Every professional can relate. Goes wide, to the wrong room.

ICP-facing: daily habits for digital visibility in regulated sectors. Only CEOs and marketing directors in healthcare and financial services care. Small room, right room, near-identical follower gain.

The question to ask before publishing: would my specific ICP find this useful, or would any professional in the general space find it relatable?

• Useful to your ICP = right room
• Relatable to everyone = wrong room

Yes, the more specific your content, the fewer people it reaches. And the more of the right people it converts. That is the trade, and it is worth making.

When you commit to it, the same flywheel that was working against you starts working for you. 360Brew learns from your engagement patterns; when your posts consistently attract ICP-type engagers, saves from buyers, substantive comments from your target market and followers who match your target profile, it starts distributing your content to more people with the same profile.

Step 3: Fix your commenting — speak in the right rooms

Where you comment matters more than most people think, because 360Brew builds its model of you from your full activity: your posts, your profile, and the comments you leave on other people's content.

The rooms you show up in train the algorithm on which room you belong in.

If you spend your engagement energy on generic LinkedIn tips, motivational coaches and career-advice posts, that is the cluster you get placed in.

Comment instead on posts your ICP is reading, industry publications, sector-specific thought leaders, and discussions inside your actual market.

Five substantive comments a day in the right place will do more for your audience quality than fifty one-liners scattered everywhere.

The Audience Quality Layer, Live in ContentIn

The problem with LinkedIn's native analytics is not that the signals are missing. It is that they are scattered, and you have to pull them manually and work out what they mean.

That is the gap the audience quality layer closes, and it is live in the product now — not a roadmap item. It does three things.

1. An audience quality score for your profile

Based on who engages with your content over time. Are the people interacting with your posts in the ICP bucket, or in the peer-and-noise bucket?

2. Post-level scoring

Every post gets an audience quality signal. Not just impressions and engagement rate, but who engaged — and whether they were buyers.

3. Lead scoring

We take the engagers — people who comment, connect, or react — and score them into four buckets: ICP, near ICP, peer and noise. When someone engages with your content, you can see immediately whether they are worth following up.

ContentIn also tracks followers gained and profile views per post automatically, which is the column that made the three posts above legible in the first place. And it feeds all of it back into your content strategy, so your post ideas and drafts are optimised to score higher with the right audience rather than simply to score higher.

Start your free trial at contentin.io →

The Bigger Picture: LinkedIn as a Business Tool

LinkedIn became a reach game for too many people. The visible metrics — impressions, likes, follower counts — optimise for visibility, not for business outcomes. For a long time the algorithm rewarded exactly that, and those were the only numbers we could see.

360Brew changed the rules, and while everyone is complaining about it, there is an advantage buried in the complaint. A relevance engine rewards you for being clearly, specifically and consistently useful to a defined group of people. That is a game a founder with 900 followers can win against an account with 90,000.

838 impressions to the right people, every week — that is what compounds.

What to Do Today: Audit Your Last Five Posts

Pull up your last five posts. Don't look at impressions first. Look at followers gained, profile views and saves.

Then ask the only question that matters: were the people who engaged actually my buyers, or were they peers, creators and other professionals who will never become clients?

That ratio is your audience quality score, and it is telling you whether you are in the right room.

If you'd rather start smaller, pick one of these and do it now:

• Fix your headline so it names a specific audience, problem and proof point
• Take your next draft and ask whether it is for your ICP or for everyone
• Pick three accounts your ICP follows and leave a substantive comment on each

One step today. The rest follows.

Want to see this in action? Our free LinkedIn post optimizer takes a draft and tells you what's weak before you publish — no account needed. Our guide to reading its feedback walks through each checkpoint. For the audience quality layer itself, start a free trial.