MagicPostM
EN
Book a call

Studies · 40,937 LinkedIn posts · 660 creators · summer 2025 and April to September 2026

AI slop on LinkedIn: posts with no detail or lived experience lose more reach

Since the week of July 27, 2026, posts with no concrete detail and no lived experience fell from 13% to 23% below their author's average post. Sounding generic or like AI costs no more than before July 27, as far as we can detect.

  • Before−13%April 1 to July 26, 2026
  • Since July 27−23%posts with no detail and no lived experience
  • Same creators−15%342 creators, before vs since
Naïlé Titah6 min read
Dot plot, one row per group, with a line joining the before and since dots: LinkedIn posts by what they contain, compared with the same creator's average post, before and since July 27, 2026. No concrete detail and no lived experience: 13% below, then 23% below. A concrete detail only: 5% below, then 9% below. A lived experience only: 0%, then 5% below. Both: 10% above, then 15% above. Random posts: 0%, then 1% below.

The essentials

Key takeaways

  1. What loses reach since the week of July 27, 2026: LinkedIn posts with no concrete detail and no lived experience. They get 23% fewer impressions than their author's average post, up from 13% fewer before.

  2. Among the same 342 creators, these posts fell a further 15% against their average post.

  3. The extra reach that comes with a concrete detail jumped 21% at July 27, more than at any other of the 24 dates we tested.

  4. Against a comparable post by the same creator, a more generic post costs what it did before: 20% fewer impressions since July 27 for a post rated 30 points more generic out of 100.

  5. Sounding like AI already cost about 9% fewer impressions before July 27, against a comparable post, and shows no extra cost we can detect since.

Read the numbersMost percentages compare a post with the same creator's average post, same side of July 27, 2026: “−23%” means 23% fewer impressions than that creator usually gets. Others compare it with a post by the same creator that is similar on everything else measured.

Chapter 1Substance

What counts as AI slop on LinkedIn?

An AI judge read 40,937 public LinkedIn posts and answered two plain questions about each one.

Does the post contain a concrete detail? That means at least one fact a reader could check: a number tied to a real situation, a named person, company, product or place, a date, a precise example. A round number in a list title, like “5 tips”, doesn't count.

Does the author tell a lived experience? That means something that actually happened to them, their team or their clients, in a specific situation. A general “I believe” or “I've learned that consistency matters” doesn't count.

The judge also rated how generic each post is and how much it sounds like AI, from 0 to 100. A post of motivational lines that thousands of accounts could publish unchanged scores high on generic, and often has neither a detail nor a lived experience.

Chapter 2The drop

Do AI slop posts get less reach on LinkedIn?

Yes, and more since late July 2026. Posts with no concrete detail and no lived experience, 17% of the posts before July 27, already got 13% fewer impressions than their author's average post. Since then, they get 23% fewer.

Among the 342 creators who published them on both sides of the date, these posts fell a further 15% against those creators' average post.

Table 1 · Posts by what they contain vs the creator's average post5 groups

What the post containsJun to Sep 2025Apr 1 to Jul 26, 2026Since Jul 27, 2026
No detail, no lived experience−10%−13%−23%
A concrete detail only−4%−5%−9%
A lived experience only0%0%−5%
Both+11%+10%+15%
Random posts (control)0%0%−1%

These posts already sat below average in summer 2025, at 10% below, and the gap widened after July 27. Posts with both a detail and a lived experience moved the other way, from 10% to 15% above. Random posts stayed where they were.

Chapter 3The date

When did LinkedIn posts with no substance start losing more reach?

In the week of July 27, 2026. We reran the model at 24 dates, each with five weeks on either side: eight in summer 2025 and sixteen from May to August 2026.

The extra reach a concrete detail brings jumped 21% at July 27, the biggest jump of the 24. The week before, it had moved 4%. No date in summer 2025 came close for the concrete detail. Over the full period since July 27, that extra reach settles at 14%, up from 2%.

Bar chart of 24 tested dates: the change in the extra reach for a concrete detail peaks at July 27, 2026, at +21%, against +4% the week before. The 8 dates of summer 2025 stay between −11% and +11%.
Figure 1Change in the extra reach for a concrete detail at 24 dates, each with five weeks on either side.

In that same week, on July 30, LinkedIn announced classifiers that identify AI slop and low-quality content, to show less of it in suggested posts and in posts from outside your network. It also added a “Seems like AI slop” report button, and later said that posts it classifies as slop get fewer views.

The timing matches. We don't see how LinkedIn ranks posts, so we don't claim a cause. The extra reach for a lived experience rose too, but it had started climbing in early July, so its date is less sharp.

This drop is separate from the drop of comment-bait posts in the same week: it holds on posts that ask for no comment at all.

+21%Jump in the extra reach a concrete detail brings, at July 27, 2026. The largest of the 24 dates we tested.

Chapter 4Style or substance

Does sounding like AI hurt your LinkedIn reach?

Yes, and it already did before July 27. Against a comparable post by the same creator, with the same substance and the same generic score, a post rated 30 points more AI-like got about 9% fewer impressions before July 27. Since then, it shows no extra cost we can detect.

Turns of phrase flagged by MagicPost's AI-phrasing detector, including “It's not X, it's Y”, cost about 4% before July 27 and show no extra cost we can detect since.

“AI and slop are not the same thing.”

Hari Srinivasan, LinkedIn's chief product officer, in the post announcing the AI slop classifiers on July 30, 2026.See the post

Generic writing works the same way. A post rated 30 points more generic got 22% fewer impressions than a comparable post before July 27, and 20% fewer since: the same cost as before. The two costs overlap and don't add up.

Since July 27, the value of a detail and of a lived experience has risen.

Dot plot, one row per trait, with a line joining the before and since dots. Before and since July 27, 2026, compared with a comparable post by the same creator: a lived experience goes from 7% to 25% more impressions, a concrete detail from 2% to 14% more, and a post 30 points more generic from 22% to 20% fewer.
Figure 2One model for all traits at once, written down before any estimate, within each creator.

What to do with your next posts

Three moves, each backed by a number from this study.

  1. A real number, a name, a date or a precise example. Since July 27, a post with one gets 14% more impressions than a comparable post by the same creator without one, up from 2% before.

    +14%since Jul 27

  2. A specific moment with your team, a client or yourself. A post that tells one gets 25% more impressions than a comparable post since July 27, up from 7%. A general “I believe” doesn't count.

    Write a personal story post →
    +25%since Jul 27

  3. Against a comparable post, a more generic one costs what it did before July 27, and sounding like AI shows no extra cost we can detect since. Before reworking the wording, check that the post has a checkable detail or something that happened to you.

    −20%more generic, stable

About this study

The data

40,937 posts

From 660 creators who posted at least 3 times before and 3 times since July 27, 2026, in English or French, text or image, 40 to 400 words: 11,612 posts from June to September 2025, 19,404 from April 1 to July 26, 2026, and 9,921 from July 27 to September 27, 2026. We left out company pages, accounts under 100 followers and posts under 3 days old.

The method

An AI model (gpt-5.4-mini) read every post with the same instructions: concrete detail or not, lived experience or not, and how generic and how AI-like it is, from 0 to 100. Each post is compared with the same creator's average post: the mean, on a log scale, of their other posts in the same language from 30 days before to 30 days after it, on the same side of July 27. Random posts land at 0% on this baseline. The comparisons with a comparable post come from one model, written down before any estimate, within each creator; we reran it at 24 dates to date the change. Adding the AI score to that model, after the fact, gives the AI-sounding numbers. Intervals come from resampling creators.

The limits

  • We don't see how LinkedIn ranks posts, so we don't claim a cause.
  • Two AI judges and no human reader yet. A second judge, Claude Sonnet 5, rated 10,072 posts from 226 of the creators with the same instructions. At equal generic score, posts with no detail and no lived experience fell 11% with its labels (between 23% fewer and 2% more), against 18% with the first judge on the same posts: same direction, less precision. The judges agree well on how generic a post is (0.70, where 1 is full agreement), less on concrete details.
  • The AI score follows the generic score closely (0.80 with the first judge, 0.85 with the second), so how the cost splits between sounding like AI and being generic is imprecise. The 9% cost of sounding like AI lies between 12% and 6%, and its change since July 27 between 10% fewer and 2% more impressions.
  • Accounts tracked by MagicPost, most of them its users, not all of LinkedIn, and only nine weeks of data after the date. The drop is clear in English and on accounts with 5,000 followers or more, less certain in French, on text-only posts and on smaller accounts. Accounts that rely most on reach outside their network do not lose more (indirect measure).
  • The extra reach for a lived experience started rising in early July, so its date is less precise than that of the concrete detail, and one date in summer 2025 shows a jump as large.
  • We name no creator and never show the impressions of an identified post.

FAQ

What is AI slop on LinkedIn?

In MagicPost's study, AI slop means generic, interchangeable content, whether a person or an AI wrote it. LinkedIn's product chief drew the same line on July 30, 2026: “AI and slop are not the same thing.” Since July 27, posts with no concrete detail and no lived experience sit 23% below their author's average post, up from 13% (40,937 posts).

What actually loses reach on LinkedIn in 2026?

Since the week of July 27, 2026, LinkedIn posts with no concrete detail and no lived experience get 23% fewer impressions than their author's average post, up from 13% fewer before, in MagicPost data on 40,937 posts. Against a comparable post, being generic costs what it did before, and sounding like AI shows no extra cost we can detect.

Does AI-written content get less reach on LinkedIn?

AI-sounding posts already got fewer impressions on LinkedIn before July 27, 2026, in MagicPost's data: against a comparable post with the same substance and generic score, a post rated 30 points more AI-like got about 9% fewer. Since then, there is no extra cost we can detect.

Do AI phrases like “It's not X, it's Y” cost more reach since July 2026?

Turns of phrase flagged by MagicPost's AI-phrasing detector, including “It's not X, it's Y”, cost about 4% before July 27, 2026 and show no extra cost we can detect since. The 4% compares posts using them with comparable posts by the same creator.

Is the LinkedIn reach drop since July 2026 just a summer slowdown?

No, in MagicPost's data the drop shows no sign of being seasonal. Each post is compared with the same creator's posts from the same weeks, so a summer dip lowers the post and its baseline alike. At 8 tested dates in summer 2025, the extra reach for a concrete detail moved 11% at most, against 21% at July 27, 2026.

How do I avoid AI slop on LinkedIn?

Give each post at least one checkable detail and, when you can, something that happened to you. Since July 27, 2026, a concrete detail adds 14% more impressions against a comparable post by the same creator, and a lived experience 25%, in MagicPost's data. Generic wording costs what it did before July 27.

Cite this study

Charts and numbers are free to reuse with a link to this page.

MagicPost (2026). “LinkedIn AI Slop: What Actually Loses Reach in 2026.” Study of 40,937 LinkedIn posts. magicpost.in/studies/linkedin-ai-slop

Embed this chart

Paste this code into your site or newsletter. The chart links back to the full study.

More MagicPost data

Which profile describes you best?

These two answers pre-fill your booking.