All companies

Pylon

Customer Support AI · 36.7K followers · LinkedIn benchmark, last 90 days (original)

data through 2026-07-30

Report card vs prior 90 days

Posts

37

-74%

Posts/wk

2.9

-74%

Median eng.

42

-47%

Smart avg

54.5

-52%

Total eng.

2.0K

-88%

Baseline vs market

  • Median engagement 42 top 33% of 70 tracked companies · top 50% of Customer Support AI (n=10)
  • Posts per week 2.8 top 70% of 70 tracked companies · top 60% of Customer Support AI (n=10)
  • Engagement per 1K followers 1.51 top 61% of 70 tracked companies · top 50% of Customer Support AI (n=10)

What worked

  • Customer Highlight — smart avg 55.8 across 24 posts (1.3× typical)

Best format: Image (smart avg 55.1, 29 posts). Best day: Tuesday. Best time block: 3pm–6pm UTC. Floors: 8+ posts per category, 8+ per timing bucket.

Who carried it

No employee posts in this window (company-page only).

Top posts

  • 2026-05-26 · Company page · Industry Insight168 eng. (156 likes · 11 comments · 1 shares)

    The spotlight is on 🔆 Sarah Taubner 🔆 today! For nearly a decade, Sarah has scaled post-sales teams in SaaS. Now a CS leader at Hummingbir

    View post →
  • 2026-07-15 · Company page · Product Update126 eng. (108 likes · 10 comments · 8 shares)

    Customer support is going agentic. Today, we’re introducing the new Pylon, the first platform built for Agentic Customer Support. Available

    View post →
  • 2026-05-21 · Company page · Customer Highlight114 eng. (100 likes · 13 comments · 1 shares)

    Today we have 🔆 Lakshmi Narayanan 🔆 in the spotlight! Lakshmi has over a decade of experience building and scaling customer success teams.

    View post →

Summary

  • Posting pace fell 74% vs the prior 90 days (11/wk → 2.9/wk).
  • Typical-post engagement fell 47% (median 79 → 42).
  • Customer Highlight posts earned 1.3× the company's typical post (24 posts).
  • Median engagement of 42 sits in the top 33% of 70 tracked companies.
2026-05-042026-08-02 · Benchmark by Imagine AI · methodology at benchmark.imagineai.me/dashboard/methodology