LinkedIn does not publish the weights inside its content ranking model, which is exactly why most “LinkedIn algorithm” advice is noise. The useful question is narrower: how do you write posts that earn the depth signals — reading time, real comments, saves, sends — that any sensible ranking system would value? That is the LinkedIn depth signals framework below.
What “depth” means on LinkedIn (and what stays inferred)
Some signals are visible in your own analytics: impressions, reactions, comments, repost count, profile views. Others are inferred — how long someone actually read, whether they paused on a paragraph, whether the comment they wrote took 90 seconds or three words. Treat the inferred bucket honestly:
- Reading time — visible to LinkedIn, not directly to you. Your proxies are dwell on long-form sections, completion of the “see more” expansion, and replies that quote later paragraphs.
- Meaningful comments — visible as comments, but the quality variable is yours to read.
- Saves — visible to LinkedIn analytics for some surfaces; treat any reported number as directional.
- Shares / sends (including DMs) — partly visible; sends in DMs are almost entirely a private signal.
- Profile clicks and follows — visible in your analytics, but they aggregate intent across many posts.
What we can and cannot know about LinkedIn ranking
Two adjacent truths most algorithm posts collapse together. LinkedIn does publish detailed product documentation for the things it sells — for example, Thought Leader Ads and the full objective-based ad formats catalogue. It does not publish the weights of its organic ranking model in the same documentation hub. Anyone telling you “LinkedIn rewards X more than Y” with confidence is reading tea leaves unless they cite a LinkedIn engineering communication.
So the safe stance for a content operator: design for the signals you can defend, log what changes, and resist clean stories about exactly how the ranker behaves week to week. The reverse engineering hobbyists are entertaining; they are not a content strategy.
Shallow vs deep engagement — and a practical depth-signals framework
Why does this matter? Because the easiest engagement to produce is also the cheapest to produce. A post can be optimised for thumb-stopping reactions and produce almost nothing of compounding value. Assume two posts in the same week from the same author. Post A collects many quick reactions and a handful of one-line “agree” comments. Post B collects fewer reactions but several long replies, a few saves, and a private DM from someone who wants to talk. Post B is the better candidate for compounding distribution and pipeline, even if Post A looks better in a screenshot.
That comparison is an explicit hypothetical scenario. It is not a benchmark of LinkedIn ranker behaviour and is not derived from any first-party account.
Six diagnostic dimensions I write against — presented as an author-owned editorial framework, not LinkedIn’s official scoring:
- Hook clarity. A first line that names the reader and the stake within one breath.
- Reading completion likelihood. Each paragraph earns the next; the post does not flatten into a list of axioms.
- Comment depth. Built-in invitations to share experience, not “agree?”.
- Save-worthiness. Something concrete to refer back to (a checklist, a number-free heuristic, a worked decision).
- Share / send-worthiness. A specific reader on the writer’s mind (“send this to your CFO if…”).
- Commercial relevance. A post that would make sense to a buyer in your category two months later.
Reading the signals — a diagnostic table
The table below is what I use during a post-mortem, with the caveat that the “may indicate” column is interpretation, not LinkedIn-confirmed mechanics.
| Signal | What it may indicate | How to improve it | What not to conclude |
|---|---|---|---|
| High reactions, low comments | Hook resonated, body did not invite participation | Replace closing line with a specific experience question | That the algorithm “loved” the post — reactions are the cheapest signal to produce |
| Comments that quote later paragraphs | Readers stayed long enough to find the substance | Lead more often with the line you want quoted | That this guarantees reach next time — sample of one |
| Saves visible in analytics | Reader believes the content has reuse value | Ship more reference posts; less stream-of-thought | That saves drive reach by a specific multiplier |
| DM sends mentioned by recipients | Reached a private network you cannot see in dashboards | Write for a named persona; make the “who” obvious | That DM volume can be measured directly |
| Profile clicks rising without conversion | Curiosity without offer clarity | Add a single specific next step in the post or About | That you are about to “blow up” — pipeline still depends on offer |
| Reactions falling, comments holding | Audience shifted from passive to participatory | Keep going; do not chase reactions back | That reach must follow — measure reach separately |
A posting system for deeper engagement
A weekly system that produces depth-signal posts more often than not — without inventing analytics:
- One idea per post; if there are two, write two posts.
- A specific audience named in your head before the first line is written.
- A sharper first line — preferably one a stranger could read aloud without ambiguity.
- Proof or example, not a generic claim (“X is hard” is filler; “I lost a Q2 in 2024 because X” is not).
- A concrete opinion you would defend in a real conversation.
- One question that invites experience, not yes/no.
- Repurpose the best comments into a follow-up post — your readers wrote your next hook.
- Track saves and comment quality manually in a small log; do not pretend the dashboard is precise.
For longer pieces and reference content I lean on a separate brief — the SEO Content Brief Generator in the library is a starting point. Generic principles like “write for the reader first” sit comfortably alongside LinkedIn-specific craft; Google’s SEO Starter Guide is a calmer reference for those principles than most social-media commentary.
Two diagnostic examples (hypothetical, not benchmarks)
These are illustrative scenarios used to talk through the framework — not first-party analytics and not evidence of LinkedIn ranking behaviour.
Example A — reaction-heavy, shallow
A founder posts a punchy three-line take. It collects many quick reactions and short “this!” comments. Saves are negligible; DM sends are zero. The diagnostic read: the hook worked; the body did not earn the next layer of attention; the post will not compound. Action: rewrite the body to lead with one concrete example.
Example B — fewer reactions, deeper participation
A consultant posts a six-paragraph diagnostic with one specific anecdote. Reactions are fewer than the founder’s. Comments are longer, two readers quote paragraph four, one saves the post for a Monday meeting, and another DMs to ask about scope. The diagnostic read: depth signals present; commercial relevance present; worth reusing as a reference post and pinning to the profile.
Open questions
The honest section, where speculation lives instead of pretending to be evidence:
- The exact weight of comments, dwell time, saves, and DM shares in LinkedIn’s ranker is not publicly documented.
- Whether organic distribution and paid distribution (e.g., Thought Leader Ads) share signals downstream is not officially explained in the same documentation hub.
- Creator mode behaviour, network density, audience interest history, and topic clustering may all matter — but should be discussed as plausibilities, not weights.
- “LinkedIn 2026 algorithm” claims that float on the platform itself (via Pulse posts) are not LinkedIn’s voice. Treat them as commentary, not specification.
If you would like a second pair of eyes on your LinkedIn posting system — the editorial framework, the diagnostic log, the commercial relevance gate — that is what my marketing audit and consulting engagement covers. Longer notes on B2B content distribution sit in Insights.
Frequently Asked Questions
Is “Depth Score” an official LinkedIn ranking system?
Not in any LinkedIn Help, Marketing Solutions or engineering documentation reviewed for this article. The term circulates in LinkedIn Pulse posts and influencer commentary, which are user-generated rather than LinkedIn’s own product voice. Treat it as a useful editorial framework, not a confirmed scoring mechanism.
Did LinkedIn confirm a 2026 algorithm change that weighs reading time and saves above likes?
No public LinkedIn documentation reviewed here makes that specific claim. The intuition behind it — that deeper signals beat shallow reactions in a serious ranker — is reasonable as a content principle, but it is not a citation.
How do I measure “depth” if LinkedIn doesn’t expose all the signals?
Track what is observable: comment count and quality, saves where reported, repost volume, replies that quote specific paragraphs. Keep a small manual log; one row per post per week. Patterns over a month are more honest than any single post’s analytics card.
Will posting longer always perform better?
No. Length is not the variable; earned attention is. A short post that resolves a question with one specific example often outperforms a long post that rambles. Use the framework dimensions, not word count, as the diagnostic.
Are LinkedIn Thought Leader Ads relevant if I am focused on organic depth?
They are the official paid surface for amplifying selected creator posts, documented in LinkedIn’s help hub. Useful to know exists; not a substitute for an organic depth-signal posting practice.