GrowthBook LinkedIn Topic Review — Organic (all-time, boosted excluded)

Ranked by median engagement per post. Engagement rate is a weak secondary signal only.

1. Top 3 Best-Performing Topics (by median engagement)

1. Industry commentary — Graham (CEO): median 38 across 4 posts. ⚠️ Low-confidence (post_count < 5).
2. Industry commentary — Ashley: median 36 across 2 posts. ⚠️ Insufficient (post_count < 3) — do not act on this yet.
3. Product announcement — Graham (CEO): median 31 across 4 posts. ⚠️ Low-confidence (< 5).

The strongest well-sampled performers are: Thought leadership (expert-cited) — Ashley (median 16, n=18), Customer story — Ashley (median 16, n=11), and Thought leadership — Graham (median 15, n=50). These are the safest bets by sample size.

2. Top 3 Under-Exploited Opportunities (high median, low volume)

1. Industry commentary — Graham: median 38, only 4 posts. High signal, low volume — worth scaling to build confidence. (Low-confidence.)
2. Product announcement — Graham: median 31, only 4 posts. Strong per-post for the CEO's page; underused. (Low-confidence.)
3. Industry commentary — Ashley: median 36, only 2 posts. Promising but insufficient — worth testing a few more, not yet recommending.

3. Candidates to Cut (high volume, low median)

• Podcast (Experimentation Edge) — Company: median 5 across 14 posts. Highest-volume, lowest-return organic topic. Trim or rework.
• Product announcement — Company: median 10 across 33 posts and declining — very high volume for a modest return.
• Customer story — Company: median 9 across 17 posts; keep an eye on it (it's rising), but currently low-yield per post.

4. Notable Rising / Declining (30d)

• Rising: Customer story — Company (n=17).
• Declining: Product announcement — Company (n=33), Podcast — Ashley (n=38), Customer story — Ashley (n=11), Thought leadership (expert-cited) — Company (n=8), Product announcement — Graham (n=4, low-confidence).

5. Caveat on Confounds

"Industry commentary" tops the board for both Graham and Ashley, but every instance is on tiny samples (2–4 posts). We can't separate topic effect from author effect or from a couple of lucky posts. The three-way overlap (same topics across three authors with very different volumes and audiences) means high medians on n<5 rows should be treated as hypotheses to test, not proven winners.

What changed since the last update (prior: 2026-07-27)

Ad clicks: 442 in the last 7d vs 398 prior (+11%), through 2026-07-22 (ad data lags a few days). CTR ticked up to 1.11% (from 0.92%) — a secondary signal, and it improved on lower spend. Top campaigns by clicks were Consideration | LaunchDarkly Audience | Engagement | Single Image (104 clicks) and Consideration | Statsig Audience | Engagement | Single Image (102 clicks); note the Target Accounts + Function | Document campaign also drove 102 clicks at a standout 4.55% CTR on tiny impressions.

On posts, the clearest genuine trend is GrowthBook Product announcement declining — median 8 across 10 recent posts (vs 11 across 2 last period); the "Double down on Product announcement for GrowthBook" rec no longer holds and should soften given the drop on solid volume. Ashley's Podcast is also declining — median 8 across 13 posts (vs 15 across 5). GrowthBook Customer story is rising — median 10 across 2 posts (vs 6 across 3) — a mild positive but small n. Ashley's Industry commentary and Thought leadership show only single/first-time posts this window (Industry commentary: first posts this period, n=1, median 45; Thought leadership: n=2 vs 0 prior), so the "Double down on Industry commentary for Ashley" rec stays anecdotal, not yet earned. Graham's Product announcement is n=2 vs n=1 — too thin to call a trend.

Influenced pipeline and ROAS held steady.

Top & worst posts (last 7 days)

Ashley

• Top: "Fanatics, UPS, and DoorDash all prove the same thing about testing." (26)
• Worst: "B2C: "It's black, it fits my apartment, it makes hot water for my tea." (12)

Graham

• Top: "We don't have product managers at GrowthBook. That is deliberate..." (17)
• Worst: "We ran an internal contest at GrowthBook a while back. The setup was s" (14)

GrowthBook

• Top: "How many experiments did you not run this week because setup felt like" (14)
• Worst: "Most experimentation programs stay stuck with one team, which caps how" (5)