Key findings
- Em dash posts: 3.5% (2023), 12% (2024), 26% (2025), 20% (2026).
- In a fixed panel of 345 people, heavy em dash users went from 5% to 53% to 35%.
- Em dash posts are not penalised (+7%); AI-sounding posts are (-22%).
- Buzzwords fell; AI structures like "It's not X, it's Y" and arrow bullets kept rising.
Contents
The short answer
Em dash use on LinkedIn rose sevenfold after ChatGPT arrived, from 3.5% of posts in 2023 to 26% in 2025, then dropped to 20% in 2026 as people learned it was a tell. But the dash itself is not what costs you engagement.
Same people, different punctuation
3.5% to 26%
Finding 5
Em dashes went up sevenfold, then fell
Posts with at least one em dash: 3.5% in 2023, 26% in 2025, 20% in 2026. In a fixed panel of 345 people who posted in both 2023 and 2025, the share who used one in at least 1 post in 5 went from 5% to 53%, then down to 35% in 2026.
A fixed panel matters here: it rules out the idea that new, AI-heavy users simply joined the sample. The same people changed how they punctuate.
Do em dashes hurt engagement?
+7%
Finding 6
The em dash itself is not what readers punish
Posts with an em dash got +7% engagement vs the same author's posts without one (95% CI +3% to +10%). The dash is a marker that often travels with AI text, not the cause of the penalty. Deleting dashes from an AI draft does not make it read human.
The other AI tells, tracked
1.3% to 9.6%
Finding 7
People dropped the famous AI words but kept the AI structures
"Delve", "game-changer" and "ever-evolving landscape" fell from 4.2% of posts in 2023 to 0.5% in 2026. Over the same years "Here's why" style set-ups rose from 1.3% to 9.6%, "It's not X, it's Y" contrasts from 1.2% to 5.9%, and arrow bullets from 0.6% to 12.3%. Writers learned to delete the words everyone mocks, not the rhythm that gives AI away.
| Pattern | 2022 | 2023 | 2024 | 2025 | 2026 |
|---|---|---|---|---|---|
| Em dash | 2.9% | 3.5% | 12.5% | 26.0% | 20.3% |
| "It's not X, it's Y" | 0.0% | 1.2% | 1.4% | 2.3% | 5.9% |
| "Here's why / here's the thing" | 0.3% | 1.3% | 4.1% | 5.9% | 9.6% |
| Arrow bullets | 0.3% | 0.6% | 4.0% | 11.8% | 12.3% |
| "Delve", "game-changer", "landscape" | 0.4% | 4.2% | 3.9% | 2.0% | 0.5% |
Posts of 150+ characters. 2026 runs to September.
Does your next post read as AI?
Paste it into the free LinkedIn AI Detector. It scores the post and points at the phrases and patterns that give it away. Nothing you paste is stored.
Common questions
- Is the em dash a sign of AI writing?
- It is a weak signal. Em dash use jumped from 3.5% to 26% of LinkedIn posts as AI tools spread, but plenty of people used them before ChatGPT (6% of posts in 2021).
- Should I stop using em dashes on LinkedIn?
- You do not need to for engagement: posts with an em dash got +7% vs the same author's other posts. If you want to avoid looking AI-written, fix the voice, not just the punctuation.
- Why does ChatGPT use so many em dashes?
- Language models learned from edited text such as books and journalism, where the em dash is common, and they use it to join clauses smoothly. On LinkedIn, where most people type quickly on a phone, it stands out.
Download the data
Suggested citation
LigoSocial Research (2026). Em dashes and AI: how often LinkedIn posts use them. https://ligosocial.com/research/em-dash-ai
Free to reuse under CC BY 4.0: quote the numbers, embed the charts or rework the data, with credit to LigoSocial Research and a link to this page. Every figure is aggregated and anonymised; no individual, profile or post can be identified.
How we collect, clean and anonymise the data: methodology. Questions or a cut we have not published? Ask us.
Related research
- Annual reportThe State of LinkedIn 2026: what works for everyday professionals40 findings from 134,522 posts, plus what works at each follower count.
- ReportDo AI-written LinkedIn posts get less engagement?AI-sounding posts get 22% less engagement, and the share of them peaked in 2024 to 2025.
- ReportLinkedIn post formats compared: images, carousels, video, text, links and pollsImages +36%, link previews -15%, polls -50%. Carousels only pay past 10 pages.
- ReportHow long should a LinkedIn post be? What 134k posts showLonger posts win: 2,400+ characters +24% vs 600 to 900.
- ReportDo hashtags work on LinkedIn? What 134k posts showHashtags: no penalty, small lift. Tagging people: +35% to +126%.