PETRICHOR.

Reach Reality · Claims Index

LinkedIn claims, checked against the source.

A human-readable evidence layer for 21 repeated claims. Every record carries a verdict, real finding, primary source, action, evidence status, and review date.

Research version 2026-08-03 · Last audited August 3, 2026

Evidence library

Find the claim behind the advice.

21 of 21 records shown.

MYTH

LinkedIn cut company-page organic reach 60-66% between 2024 and early 2026 to force companies into paid ads.

Metricool's 673,658-post, 63,108-account global study found company-page impressions per post down about 10% year over year (Jan-Feb 2025 vs Jan-Feb 2026). Larger account tiers grew; tiny and small tiers fell 27.52% and 12.19%. No transparent study supports a platform-wide 60-66% collapse.

Primary source: Metricool, LinkedIn Study 2026 · 2026-04-14
Evidence status: High · Source link: Available · checked 2026-08-08

Do instead: Benchmark your own account's median reach across at least 10 posts before assuming a platform-wide collapse.

Category: Reach & Distribution · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

CONTESTED

LinkedIn replaced the Relationship Graph with an Interest Graph; only 31% of the feed is first-degree connections now.

Out-of-network semantic retrieval is confirmed (an LLM-based retriever picking roughly 2,000 candidates from hundreds of millions of posts in under 50ms), but LinkedIn still calls the network a key signal and no primary source publishes a 31% figure. The number traces to a paid report that does not expose its sampling method.

Primary source: LinkedIn Engineering, Engineering the next generation of LinkedIn's Feed · 2026-03-12
Evidence status: High on mechanics; low on the 31% figure · Source link: Available · checked 2026-08-08

Do instead: Write for both your existing network and topical relevance. Don't abandon relationship-building to chase a keyword score.

Category: Platform Mechanics · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

Dwell time is the dominant signal: a post read for 0-3 seconds gets 1.2% engagement, one read for 61+ seconds gets 15.6%, a 13x cliff.

LinkedIn explicitly rejects a single universal dwell threshold and normalizes long-dwell scores daily by content type, creator type, and distribution context. The 1.2%/15.6% table traces to an uncited July 2025 blog post with no named sample, dates, or method, and is internally inconsistent about whether the data even exists.

Primary source: LinkedIn Engineering, Leveraging dwell time to improve member experiences on the LinkedIn feed · 2024-10-01
Evidence status: High · Source link: Available · checked 2026-08-08

Do instead: Write posts worth a full read. Don't build a strategy around a timer that isn't real.

Category: Content & Format · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

There's a 60- or 90-minute golden hour. If a post doesn't get engagement in that window, it's dead.

No disclosed test window, audience size, or pass/fail score exists. LinkedIn deliberately relaxed recency in a mid-2025 feed test so older relevant posts could resurface, and Metricool found about half of lifetime impressions arrive within 48 hours, not one hour.

Primary source: LinkedIn Engineering, Engineering the next generation of LinkedIn's Feed · 2026-03-12
Evidence status: High · Source link: Available · checked 2026-08-08

Do instead: Measure post performance at 48 hours and 7 days. Stop treating minute 60 as a verdict.

Category: Timing & Engagement · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

There's a silent penalty that can drop your reach from 8,500 to 340 impressions overnight, with a 60-90 day recovery.

LinkedIn confirms it can limit reach for suspected artificial engagement, but publishes no standard impression loss and no standard recovery duration. The 8,500-to-340 story is source-laundered: no named creator, no visible native analytics, and a broken citation chain (one link post-dates the article citing it).

Primary source: LinkedIn, Authentic content and conversations · 2026-03-13
Evidence status: High · Source link: Available · checked 2026-08-08

Do instead: If reach drops, check for an explicit LinkedIn notice first, stop any pod or automation activity, and track medians across at least 10 posts before diagnosing a penalty.

Category: Enforcement & Penalties · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

CONTESTED

A founder's personal profile always out-reaches the company page by 4-5x.

Metricool found roughly equal average impressions per post between pages and profiles, with profiles earning 63% higher engagement and 238% more comments. AuthoredUp, across more than 3 million posts, found profiles at 1.72x the page median in impressions. Neither supports a universal 4-5x multiplier.

Primary source: Metricool, LinkedIn Study 2026 · 2026-04-14
Evidence status: High on direction; medium on magnitude · Source link: Available · checked 2026-08-08

Do instead: Post as a founder because it drives real conversation and trust, not because of a guaranteed multiplier.

Category: Channel Strategy · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

LinkedIn has a disclosed 'topic-authority score' that rewards staying in one content lane.

Semantic matching of content, professional identity, and member interests is confirmed. A named, disclosed topic-authority score is not; it does not appear in any LinkedIn engineering or policy disclosure.

Primary source: Tim Jurka, How does LinkedIn Feed work (confirms semantic matching, not a named score) · 2025-08-11
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Pick two or three durable content lanes because it helps readers recognize you, not because you're gaming a named metric.

Category: Platform Mechanics · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

Comments are weighted exactly 2x likes in the ranking algorithm.

The production ranker is a sequential, multi-objective model that reads more than 1,000 historical interactions per member. LinkedIn does not publish fixed action weights, and a comment's value is explicitly contextual, varying by member, post, author, and session.

Primary source: Feed-SR paper (LinkedIn-authored), submitted Feb 12, revised May 29, 2026 · 2026-05-29
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Aim for comments that add real information. There's no fixed math target to hit.

Category: Platform Mechanics · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

If the author replies to comments within 30 minutes of posting, visibility jumps up to 35%.

Buffer found a 30% cross-platform association, not a LinkedIn-specific or causal 30-minute effect, between author replies and engagement, across nearly 2 million posts and 220,000+ accounts on six networks. Buffer itself warns of reverse causality: successful posts may simply attract more author replies.

Primary source: Buffer, State of Social Media Engagement 2026 · 2026-03-05
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Reply promptly when there's a real conversation happening, not to hit a manufactured multiplier.

Category: Timing & Engagement · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

Engagement pods get caught by specific detectable thresholds, like five comments inside three minutes.

LinkedIn confirms it detects coordinated engagement, bots, scripts, and automation, and can limit reach, de-rank automated comments, remove pod groups, warn members, or restrict accounts. It discloses no timing regularity, graph-density threshold, text-similarity threshold, or rate limit.

Primary source: LinkedIn, Authentic content and conversations · 2026-03-13
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Don't use pods or auto-engagement tools at all. There is no safe threshold to game, because none is published and the ones being repeated are fabricated.

Category: Enforcement & Penalties · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

CONTESTED

LinkedIn's AI-content detector is 94% accurate at flagging generic AI slop.

LinkedIn itself states 94% correct identification in its initial testing of a generic-content classifier, but discloses no sample size, class balance, precision/recall breakdown, or independent audit. It is a first-party claim about an internal test, not an externally verified accuracy rate.

Primary source: LinkedIn, Keeping conversations real on LinkedIn · 2026-06-04
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Publish AI-assisted content that carries your own specific perspective. Don't assume a bare percentage protects you either direction.

Category: Enforcement & Penalties · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

Posting more than once a day gets you throttled. There's an optimal posts-per-week ceiling.

Buffer's analysis of more than 2 million posts from 94,000+ accounts, using within-account normalization, found more posts per week associated with more impressions and engagement, not less (2-5 posts/week: +1,182 impressions; 6-10: +5,001; 11+: +16,946, all relative to one post per week).

Primary source: Buffer, How Often to Post on LinkedIn · 2025-08-28
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Post as often as you can sustain real quality. There is no credible evidence more volume triggers a throttle.

Category: Timing & Engagement · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

The Relationship Graph is dead. LinkedIn only shows you strangers now.

Out-of-network interest-based retrieval is real and additive, but LinkedIn's hybrid feed inventory still includes network posts and followed sources, and network affinity remains an explicit ranking signal.

Primary source: LinkedIn Engineering, Engineering the next generation of LinkedIn's Feed · 2026-03-12
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Keep building your network deliberately. Interest-matching adds distribution; it doesn't replace what your connections already give you.

Category: Platform Mechanics · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

Collaborative articles are a good format to invest in for reach.

LinkedIn stopped creating new collaborative articles and made existing ones read-only. The format no longer generates new reach for anyone.

Primary source: LinkedIn Help (collaborative articles, current in 2025-2026) · 2026-08-03
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Retire this format from your content plan. It no longer generates new reach.

Category: Content & Format · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

360Brew is LinkedIn's current production feed algorithm.

360Brew was a 150-billion-parameter research model with a small early-2025 test that was stopped. The 2026 production ranker is the separately disclosed Feed-SR system.

Primary source: Tim Jurka, on record (360Brew test stopped, not current ranking) · 2026-04-01
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: When a post cites 360Brew as proof of how the algorithm works today, treat that as outdated framing, not current mechanics.

Category: Platform Mechanics · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

MYTH

Running LinkedIn ads on your account boosts your organic reach on the same account.

LinkedIn's relevance policy states third-party payments do not influence ordinary organic distribution, except for content that is clearly labeled as promoted.

Primary source: LinkedIn Help (relevance policy, current in 2026) · 2026-08-03
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Treat ad spend and organic reach as separate systems. Don't expect a paid campaign to lift unpaid posts.

Category: Channel Strategy · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

CONTESTED

Asking 'comment YES below' or similar engagement-bait questions is a safe, reliable reach hack.

Metricool found questions associated with 77.39% more comments and comment CTAs with 80.07% more, in one observational dataset. LinkedIn now explicitly targets engagement bait for suppression, so the tactic that 'worked' in that dataset carries live platform risk.

Primary source: Metricool LinkedIn trends · 2026-04-14
Evidence status: Reviewed primary source · Source link: Available · checked 2026-08-08

Do instead: Ask questions only when the answer changes your view or genuinely helps the audience. Skip the empty 'agree?' CTAs.

Category: Content & Format · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

VERIFIED

LinkedIn can and does limit reach for accounts running suspected fake engagement.

This one is true. LinkedIn's March 2026 policy disclosure confirms it can limit feed reach, de-rank automated comments, remove engagement-pod groups, warn members, remove program badges, and restrict accounts for suspected artificial engagement.

Primary source: LinkedIn, Authentic content and conversations · 2026-03-13
Evidence status: High (LinkedIn-disclosed policy) · Source link: Available · checked 2026-08-08

Do instead: Don't join pods or run automation. There's no safe version of the thing this claim is warning about.

Category: Enforcement & Penalties · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

VERIFIED

Dwell time is a real, meaningful ranking signal.

Also true, confirmed across LinkedIn's 2020 and 2024 engineering disclosures. The myth isn't that dwell matters, it's the fixed-cliff number attached to it (see dwell-cliff-13x). Dwell is normalized daily by content type, creator type, and distribution context, not a flat threshold.

Primary source: LinkedIn Engineering, Leveraging dwell time to improve member experiences on the LinkedIn feed · 2024-10-01
Evidence status: High (LinkedIn-disclosed) · Source link: Available · checked 2026-08-08

Do instead: Write for genuine dwell, content that earns a full read, instead of trying to game a threshold that isn't published.

Category: Content & Format · First reviewed 2026-08-03 · Last reviewed 2026-08-03 · Version 2026-08-03 · Initial reviewed record

Check this claim interactively

Evidence maintenance

The Index remains an evidence layer, not another scorecard.

Material platform disclosures trigger review. The active dataset is reviewed at least quarterly. A verdict or finding change receives a change note and research-version update. A broken primary link creates a review flag rather than silent confidence.

How to read the Index

What does Contested mean?

The repeated advice is broader or more certain than the source supports. Use the real finding and dated action instead.

Does absence prove a claim is false?

No. It means the current library does not contain a reviewed record. Use the checker wording as a research lead, not a verdict.

Can a verdict change?

Yes. A material disclosure or stronger primary source can change a verdict. The record then receives a change note and research-version update.

Check a specific claim · Inspect the method · Read the evidence answer · Read the Tool Trust Charter