Why Did My LinkedIn Reach Drop? Audit Headline vs Posts

In this article

1. Rule out the platform-wide drop first
2. What the ranker reportedly reads about you
3. A note on the LinkedIn 360Brew algorithm
4. Step 1: Write down what your headline claims
5. Step 2: Tag your last 20 posts by topic
6. Step 3: Find the mismatch and fix one side
7. A 30-day plan built from real work notes
8. Week 1: align the profile
9. Week 2: write for saves
10. Week 3: clean up endings and links
11. Week 4: re-run the audit

If your LinkedIn impressions fell this year while you kept posting at the same pace, one likely cause is that your profile and your posts describe two different people. LinkedIn's feed now reportedly uses language models that read your headline alongside each post and the reader's history, so a scattered posting record gives the ranker little to match. Below is a diagnostic you can run in an afternoon, followed by a 30-day plan for acting on what it finds.

Rule out the platform-wide drop first

Before rewriting anything, check how much of your decline belongs to the platform. Sources measure it differently. All In's summary of the 2026 changes cites Richard van der Blom's data showing views down about 50% and engagement down 25%. It also cites a separate analysis showing visibility down 47% and engagement down 39%. ZoomSphere says reach dropped hardest around March 2026.

Export your post analytics and compare three things:

1. Median impressions per post in the months before March 2026.
2. Median impressions per post from March onward.
3. The topics of your best and worst posts since March.

If your median fell by roughly the reported platform amount and your topic mix held steady, you are seeing the same shift as everyone else. If it fell further, or if some topics held while others collapsed, the rest of this article applies to you.

Rule out two cheaper causes at the same time:

• Engagement bait endings. A MagicPost study of 687,000 posts, reported by Missive Digital, found that posts ending with "Comment KEYWORD" went from 10% above their author's average reach to 47% below after 27 July 2026. Posts that closed with a genuine question did not lose reach.
• External links. Van der Blom's 2026 research on 1.3 million posts found that one external link in the post body cut median reach by 18.8%, according to Melanie Goodman. Goodman adds that comments containing external links can lose up to 80% of their visibility, so moving the link to the first comment is not a clean workaround.

Founders should run one more check. ZoomSphere, citing Whitehat's analysis of Socialinsider data, puts company page organic posts at about 2% of what appears in feeds. If most of your posting happens on a company page, do the topic work below on your personal profile.

What the ranker reportedly reads about you

According to Upgrowth, LinkedIn confirmed on 12 March 2026 that its feed uses LLM-based retrieval and a Generative Recommender ranking model, and that it is acting against engagement pods and bait. Commentify dates the same announcement to October 2026. If you searched for a LinkedIn algorithm October 2026 change, that conflict explains some of the confusion. SocialPilot reports no new ranking mechanic between April and September, and no source here confirms a separate October ranking change. Missive Digital does report that in October 2026 LinkedIn's Tim Jurka described AI-slop detection on every post distributed beyond your immediate network.

The inputs matter more than the date. Social Media Today reports that LinkedIn's Engineering Blog says the feed uses profile information, skills, geography and the content a member engages with, and that the system adapts to changing interests instead of relying on historical markers. Siela lists the author's name, headline, company and industry as confirmed retrieval inputs, and notes that Jurka said in August 2025 that ranking factors in the author's professional identity.

A note on the LinkedIn 360Brew algorithm

Much of the advice you will read calls this system 360Brew. The name comes from a January 2025 research paper describing one 150B-parameter decoder-only model trained on more than 30 LinkedIn ranking tasks. The Hugging Face paper page now says the paper has been withdrawn, and Kunal Dwivedi's breakdown points out that the paper calls V1.0 a pre-production research model and does not claim it ranks the feed. SocialPilot says LinkedIn announced 360Brew's deployment in March 2026, while Siela and Upgrowth say LinkedIn's March posts never name it. Treat "360Brew runs your feed" as unconfirmed.

The paper still offers one useful idea. It describes a model that, conditioned on a member's profile and interaction history, finds patterns personalized to that member without hand-crafted features. Whether or not that exact model is live, a text-reading ranker that sees your headline next to your post has more to work with when the two agree.

There is a limit to this reasoning. Postiv labels profile-content alignment a community-observed signal, not one LinkedIn has confirmed. The diagnostic below is a cheap bet with no guarantee attached.

Step 1: Write down what your headline claims

Copy your headline, company and industry fields into a blank document. Under them, write the 2 or 3 topics a stranger would expect you to post about after reading only those lines. Record what the text says rather than what you meant by it.

Founders and engineers usually land in one of three places:

• The headline is a title and a company, such as "Co-founder at [company]", and names no topics.
• The headline lists several roles and interests, so no single topic stands out.
• The headline names a topic you stopped posting about months ago.

Write topics as nouns a reader would recognize from their own work, such as "Postgres migrations", "pricing for B2B SaaS" or "hiring the first engineers". Phrases like "building great products" give a text model nothing specific to match.

Step 2: Tag your last 20 posts by topic

Pull your last 20 posts into a spreadsheet with these columns: date, topic, impressions, comments, and saves if your analytics show them.

Give each post one topic only. Tag by what the post argues or teaches rather than by format, so a carousel about a late hire is tagged "hiring" and not "carousel" or "lessons". Forcing one tag per post exposes drift. If a post honestly needs two tags, it probably tried to make two points.

Then count three things: how many posts fall under a headline topic, how many distinct topics you used, and the median impressions for on-headline posts compared with the rest.

Step 3: Find the mismatch and fix one side

Compare the sheet with your headline notes. Most results fit one of these patterns.

The real decision is which side to change. Fix the headline when your posts reflect work you do now and want to be known for, and the headline is stale. Narrow the topics when the headline is right but your posts follow whatever came up that week.

LinkedIn niche consistency has a cost. Goodman argues that LinkedIn has moved from a relationship graph to an interest graph, so follower count and reach are now decoupled and hashtags count only as a lightweight topic signal. The off-topic posts your existing network used to like may stop earning reach, and dropping them can feel like a loss in the first weeks.

A LinkedIn headline for reach names the role and then the topics. A backend lead might write "Backend lead at [company]. Postgres migrations, on-call design, hiring senior engineers." Postiv also recommends making the About section state the same topics.

A 30-day plan built from real work notes

Joinvalley claims the model takes about 90 days to map your expertise, though it gives no method. Treat these 30 days as the first stretch of a longer process.

Before day one, start a note bank for each of your 2 or 3 topics. A useful note records something real: a decision and why you made it, a mistake and what you changed, a number from a project, or a tool you dropped and what replaced it. Every post in the plan starts from one of these notes.

Week 1: align the profile

Update the headline and About section so they name the same topics. Post only on those topics this week, with each post drawn from a note.

Week 2: write for saves

Hootsuite cites an analysis of more than 3 million posts finding that a save can drive about five times the reach of a like and roughly twice the reach of a comment. Checklists, decision criteria, config snippets and step-by-step write-ups of real work are the kinds of posts people keep.

Week 3: clean up endings and links

End posts on the last concrete point or on a specific question a peer can answer from experience. Remove "Comment KEYWORD" lines entirely. Keep external links out of the body, and remember that links in comments carry their own reported penalty.

Week 4: re-run the audit

Tag the posts from these 30 days using the same rules as Step 2. Compare median impressions per topic against your earlier numbers, then keep the topics that held.

Sources disagree on AI detection. ZoomSphere says LinkedIn cannot detect AI-written text, while Missive Digital reports Jurka describing AI-slop detection. Either way, a post built from a specific note about your own work avoids the generic copy that sources associate with lower reach.

The step most likely to slip by week two is drafting from notes when you are busy, and that is the part HireSteward automates. You keep a bank of notes about real projects, mistakes and decisions. It drafts LinkedIn and X posts in your voice using only those notes, never invented anecdotes, so each post stays tied to the topics your headline claims. You can review and edit every draft before it publishes. You can start a free trial for 7 days without a credit card.

Frequently asked questions

Why did my LinkedIn reach drop in 2026?

Part of the drop is platform-wide. Reported declines range from about 47% to 50% depending on the source and method, with the sharpest fall around March 2026. Beyond that, the feed reportedly reads your headline, company and industry alongside each post, so a profile that does not match your topics may cost you more. Engagement bait endings and external links in the post body are two other common causes.

Is the LinkedIn feed run by 360Brew?

That is unconfirmed. 360Brew is a 150B-parameter research model described in a January 2025 paper that has since been withdrawn. One source says LinkedIn announced its deployment, while others say LinkedIn's March 2026 feed posts never name it.

Did the LinkedIn algorithm change in October 2026?

One source dates LinkedIn's feed announcement to October 2026, while most date it to 12 March 2026. No source confirms a separate ranking change in October. One source does report that Tim Jurka described AI-slop detection that month for posts shown beyond your immediate network.

How many topics should I post about on LinkedIn?

Practitioners commonly recommend 2 or 3 topics that your headline and About section also name. LinkedIn has not confirmed this alignment as a signal, and it is treated as community-observed. It is still cheap to test with a 30-day audit.

How long before a headline change affects my reach?

There is no official timeline. One practitioner source claims the model takes about 90 days to map your expertise but gives no method. Treat a 30-day test as an early read rather than a final verdict.

Sources

1. LinkedIn's algorithm reset in 2026: 10 changes that actually moved reach — All In
2. LinkedIn Algorithm 2026: How It Ranks Posts and Why Your Reach Dropped
3. LinkedIn Algorithm Updates 2026: What's New & What to Do (Missive Digital)
4. LinkedIn Algorithm 2026: Why Your Reach Dropped (and 9 Data-Backed Fixes)
5. LinkedIn Algorithm 2026: March Update and 360Brew Fact Check
6. How the LinkedIn Engagement Algorithm Works in 2026: What You Need to Know
7. LinkedIn updates its Feed algorithm (Social Media Today, Andrew Hutchinson)
8. LinkedIn 360Brew Algorithm: What It Is and Why Topics Matter
9. Paper page - 360Brew: A Decoder-only Foundation Model for Personalized Ranking and Recommendation
10. LinkedIn wrote its ranking problem as a prompt
11. How the LinkedIn Algorithm Works: September 2026 Update
12. 360Brew: A Decoder-only Foundation Model for Personalized Ranking and Recommendation
13. LinkedIn Algorithm 2026: What 360Brew Changed (Verified August 2026)
14. LinkedIn's New 360Brew AI Algorithm: Why Your Reach Dropped 47% and What to Do
15. How the LinkedIn algorithm works in 2026 (Hootsuite)