Post Optimization

Last updated: May 3, 03:08 AM

Content Optimization Engine

Data-driven insights on what content features drive performance

Posts analyzed

64

From Nov 7, 2024 to Apr 30, 2026. Out of 128 total posts in the database, 64 fall outside the 18-month window and 0 have fewer than 100 impressions captured.

How to grow this sample
  • Capture more analytics — open each recent post on LinkedIn and let the browser extension grab the in-app analytics drawer. Posts without captured analytics show 0 impressions and get filtered out.
  • Post more often — only the last 18 months are included (older posts ran under different LinkedIn ranking rules and would muddy the signal).
  • Lower the impression floor — currently set at 100. Posts below this threshold are usually too fresh for LinkedIn to have tallied yet, but the floor can be tuned in src/app/optimization/page.tsx if you'd rather include them.

Image Impact

+36%

Impressions vs. posts without an image

Question Impact

+0%

Impressions vs. posts without a question

Sweet Spot

101-150

Best-performing word-count range

Best Day

Thursday

9,897 avg/post · n=15

Best Slot

Thu · Noite

11,656 avg · n=8

Feature Impact on Impressions

% change vs. posts without the feature. Gray bars = signal not trustworthy (need ≥15 posts on each side AND |effect| ≥15%; below either bar, the difference can be confounding from post type rather than the feature itself).

  • External Link negative is consistent with LinkedIn's well-documented preference for keeping users on-platform — community consensus.
  • Bold Unicode: external research finds no direct algorithm penalty; bold-as-mathematical-symbols can hurt screen-reader accessibility but doesn't trigger reach throttling. The slight negative in our data is more likely confounding — your non-bold posts include several Bemobi announcements that the algorithm boosts independent of formatting. Treat as inconclusive until we stratify by post type.

Feature Stacking Impact

Impressions by number of features included

Word Count Impact

Best-performing length: 101-150 words

Topic Performance

Writing Style Performance

When you post (day × time)

Both rows show avg impressions per post (São Paulo local time). Top: aggregated across the whole day. Bottom: same days, broken down by time of day. Read each column top-to-bottom: the bar tells you the best day overall, the cells tell you when within that day to publish. Each day's bar equals the count-weighted average of its three buckets.

  • Best day: Thursday (9,897 avg impressions/post).
  • Strongest slot: Thursday · Noite (11,656 avg/post, n=8).
  • Confidence caveat: only 9 of 21 cells have ≥3 posts; the other 12 are too thin to read as signal — treat anything outside the meaningful cells as noise.
Mon
Tue
Wed
Thu
Fri
Sat
Sun
Per post
full day
6,899
n=14
5,975
n=8
6,105
n=7
9,897
n=15
8,266
n=17
3,004
n=1
2,630
n=2
Manhã
06–12h
6.6k
n=9
6.0k
n=6
5.3k
n=4
7.2k
n=5
7.7k
n=9
3.0k
n=1
2.3k
n=1
Tarde
12–18h
6.4k
n=2
5.2k
n=1
8.8k
n=2
9.6k
n=2
7.2k
n=4
3.0k
n=1
Noite
18–06h
8.2k
n=3
6.8k
n=1
4.0k
n=1
11.7k
n=8
10.5k
n=4