The LinkedIn Playbook

There is a lot of LinkedIn advice on the internet. Most of it is written by people who post for a living, drawn from data they don't share, generalized from accounts that don't resemble yours. This paper is the opposite. One account. One year. Every number transparent. The hope is that the patterns are big enough to translate.

Section 1The data, what it is, what it isn't

This study draws on three datasets. The first is the LinkedIn Creator Analytics export covering May 2, 2025 to May 1, 2026 — 365 days of impressions, engagements, and new follower counts. The second is a 158-post classified corpus where each post is labeled by theme (AI, family, professional, founder, etc.) with full engagement breakdown (reactions, comments, reposts, impressions). The third is a deeper 30-post intelligence file where each post is tagged by hook pattern, format, and length.

What the data does not include: post text from every post (only top performers), exact publish times, and per-post follower attribution. Theme and hook classifications were performed manually before this analysis began, which means the labels reflect a single human's judgment — a limitation worth flagging up front.

Section 2The shape of one creator's year

Headline numbers first. The account generated 698,795 impressions, 15,262 engagements, and converted to 4,305 new followers across the year. Engagement rate of 2.18% sits 9–45% above the published benchmark for personal LinkedIn profiles. The follower base grew 155%, from 2,782 to 7,087.

Figure 1. Followers grew from 2,782 to 7,087 — a 155% increase. The slope steepens in the second half.

The growth curve never goes negative. There is no point in the year where the audience contracts. But the trajectory is not flat — it accelerates in the second half of the year, even as raw impressions decline (more on this in Section 12).

Section 3When to post: a deeper rule than "Tuesday morning"

The conventional advice is that you should post on Tuesday mornings. The data here does not support that as a primary rule. The right rule has three parts.

Figure 2. Sunday's engagement rate is 2.36% — higher than every other day. Saturday is the worst at 1.98%.

Part one: Sunday is the engagement king

Sunday's average engagement rate is 2.36% — the highest of the week. 13 of the year's 60 top posts (22%) were published on Sunday, more than any other day. This contradicts mainstream advice, but the mechanism is plausible: weekend feeds have less competition, and the audience that scrolls Sunday has opted in to see content rather than been forced to.

Part two: Saturday is structurally dead

Saturday delivers 1,191 average daily impressions — 38% below Thursday. Only 4 of 60 top posts ever landed on a Saturday. The same content moved from Saturday to Sunday gains 64% in reach.

Figure 3. Three reasons Sunday wins and Saturday loses — same audience, same content, different days.

Part three: Tuesday acquires the most followers

Tuesday delivers 15.0 new followers per day — the highest of the week. This appears to be a newsletter-halo effect (a Tuesday newsletter drives traffic to LinkedIn). Tuesday is not the best engagement day, but it is the best conversion day. Tuesday and Sunday play different roles. Tuesday converts existing readers into followers. Sunday deepens the relationship with people already following.

Section 4The 5-day algorithmic halo

Every top-performing post creates a tail. The day immediately after a top post averages 3,721 impressions — nearly double the daily baseline of 1,915. By day +2 the lift is 26% above baseline; by day +5 it is still 14% above. The algorithm warms up the next post you publish for almost a full work week.

Figure 4. After a top post, daily impressions remain elevated for 5 days. Posting into this window compounds.

Your next post is most powerful in the 48 hours after your last hit. Most creators wait. They shouldn't.

The implication runs against most posting calendars. The optimal cadence is not "Mon-Wed-Fri at 9am" — it is "post a strong piece, then post again the next day or two while the audience is primed." May 2025 in this dataset produced a streak of 5 top posts in 7 days, which together account for 16% of the year's top-60 posts. The streak wasn't coincidence; it was momentum compounding.

Section 5What to write about: theme analysis

From the 158-post classified corpus, themes break down by engagement rate as follows. The order may surprise creators chasing trend topics.

Figure 5. Engagement rate by theme across 158 posts. Founder/startup content tops; AI/tech content bottoms.

The pattern is consistent: specificity beats abstraction. A founder shipping an actual product outperforms a thinker offering opinions. A graduation visit with a sibling outperforms "five lessons on family." A hurling explainer outperforms a generic culture post.

Section 6How to open: hook patterns ranked

Of 30 deeply-tagged posts, four hook patterns emerge. The order is unambiguous.

Figure 6. Personal news openings beat declarative openings by 36%.

The bold declarative — the format LinkedIn-influencer YouTube tutorials recommend — is the worst-performing of the four. Personal-news openings beat them by 36%. The mechanism: a personal news opener implicitly grants the reader a relationship. A bold declarative just announces a position.

The opener tells the reader what kind of relationship this post wants. "I'm sharing something" beats "I'm asserting something" every time.

Section 7How long: post length and engagement

Length analysis splits the corpus into Short (under 100 words), Medium (100–300 words), and Long (300+ words). The pattern is non-monotonic.

Figure 7. Medium posts win on engagement rate. Long posts win on impressions.

Medium-length posts have the highest engagement rate at 2.16% — beating both Short (1.62%) and Long (1.66%). Long posts get more raw impressions (5,819 vs 5,054 for medium), suggesting the algorithm rewards depth, but proportionally fewer of those readers engage.

The implication: length does not produce engagement; it produces reach. If the goal is conversation, write medium. If the goal is broadcast, write long. Short posts underperform on both — they don't give the algorithm enough to work with and don't give the reader enough to lean in.

Section 8Format choice: structures that work

The 30 deeply-tagged posts surface clear format winners.

Figure 8. Personal news with photo is the runaway winner. Pure repost is dead air.

Two structural patterns dominate. First, posts with photos beat photo-less posts in nearly every category — but the difference is smaller than usually claimed (1.79% vs 1.80% in this dataset). Photos help framing, not engagement directly. Second, posts that combine personal news framing with a photo of the moment are the highest-leverage format on the platform.

Section 9Comments are the viral trigger

Reactions are easy. Comments are the signal LinkedIn's algorithm actually rewards.

Figure 9. Posts crossing 30 comments average 14,500 impressions — 3× posts under 6 comments.

Posts with 0–5 comments average 4,900 impressions. Posts with 30+ comments average 14,500 — almost three times more. The relationship is monotonic and steep. Comments don't just correlate with reach; they cause it. A comment is read by LinkedIn as a higher-quality engagement signal than a reaction, because comments require effort.

The strategic implication: posts engineered to trigger conversation outperform posts engineered to trigger likes. Open questions, calls for opinions, and "what would you add?" closers all sit in this space. The same content with a question at the end will outperform the same content with a statement at the end.

Section 10Three cohorts of top posts

The 60 top posts of the year split cleanly into three behavioral groups, and each group rewards a different content strategy.

Figure 10. Three top-post cohorts behave differently across the week.

The takeaway is operational: match the post type to the day slot. Sunday/Tuesday/Thursday for posts designed to travel widely (story, milestone, opinion). Monday/Wednesday for posts designed to spark deep discussion in your existing audience (frameworks, contrarian takes, complex analysis).

Section 11Reach versus resonance

Most creators conflate reach (how many people see the post) with resonance (how many people lean in). The data here shows they are independent variables.

Figure 11. The 60 top posts plotted. Above the green line: resonance hits. Far right with low engagement: broadcast hits.

The largest single post by reach (31,654 impressions) had an engagement rate of 1.29% — well below the average. The post with the highest engagement rate (4.41%) had only 12,480 impressions. The biggest reach post and the deepest connection post are at opposite ends of the chart.

Chasing reach gets you broadcast hits. Chasing resonance gets you community. The two strategies look the same from the outside and produce different businesses on the inside.

Section 12Conversion: the metric that matters

The single most important number in this dataset is followers per 1,000 impressions. It is the operational measure of audience-content fit.

Figure 12. New followers per 1,000 impressions, by month. The audience is getting more refined over time.

The first six months averaged 4.77 followers per 1K impressions. The last six months averaged 7.76 — a 62.7% improvement. February 2026 set the absolute peak at 10.49 followers per 1K. Each impression became more valuable as the audience got refined. The wrong followers from the launch month churned out, the right ones stayed, and the algorithm started pushing content to better matches.

Figure 13. Vanity metrics fell. Business metrics rose. The headline of year one.

This is the inversion that defines the year. Impressions fell 13%. New followers grew 41%. If you measured this account on impressions alone, you would conclude the year was worse in the second half. Measured on follower acquisition, it was clearly better. The first measurement is wrong.

Section 13Twelve lessons, ranked by leverage

1. End every post with a specific question. Comments are the viral trigger. A specific question gets a specific answer. Generic questions get nothing. This is the single highest-leverage habit change in the playbook.
2. Move every Saturday post to Sunday. Saturday delivers 38% less reach with no offsetting benefit. Sunday is the year's best engagement day. This is a free lift.
3. Default to "personal news + photo" format. The highest-converting structure in the dataset by a wide margin. Use it for every milestone, role change, family moment, completion, win.
4. Post into halos. When a post crosses 3K impressions in 24 hours, publish another piece within 48 hours. The algorithmic boost compounds for a full work week.
5. Aim for 250–400 words. Medium posts have the highest engagement rate. Reserve short for breaking news, long for broadcast plays.
6. Open with news, not theses. "I'm sharing something" beats "I'm asserting something" by 36%.
7. Track followers per 1K impressions monthly. Stop tracking raw impressions. The conversion rate is the leading indicator of audience-content fit.
8. Specific stories beat strategic frameworks. One story about a graduation visit outperforms ten posts on "how to think about family." Specificity is the engine.
9. Tuesday is for newsletter-aligned content. If you have a newsletter, Tuesday gets the LinkedIn post that announces it. Tuesday is the highest-converting day for new followers.
10. Eliminate cold streaks. Going 15+ days without a strong post degrades algorithmic confidence. The 2026 February slump in this dataset traces back to a cold streak that started in January.
11. AI/tech content drives reach, not engagement. If you post about AI tools, do it for traffic, not connection. Layer personal stories on top to recover engagement rate.
12. The post you don't write is a real cost. Posting volume in this dataset correlates 0.43 with follower growth — moderate but positive. The marginal post is now worth 2.7× what it was in the launch month.

Section 14Methodology and limitations

Data sources. Three datasets were combined: the LinkedIn Creator Analytics export (May 2, 2025 to May 1, 2026), a 158-post classified corpus with theme labels and full engagement breakdowns, and a 30-post deeply-tagged file with hook pattern, format, and length classifications.

Sample size and inference. Most findings draw on n ≥ 13 (Personal/Family theme) and n ≥ 6 (length analysis). Format analysis is necessarily anecdotal where n = 1 — those rows are reported with the count for transparency. Hook pattern analysis (n = 7–13 per group) is robust. Day-of-week analysis (n = 365 days) is the strongest dimension in the study.

Posting cadence inference. The export does not include total post count. "Estimated posts" is inferred from days where impressions exceeded 2,000 — substantially above the 1,400 median baseline. This produces a lower bound on cadence; days with multiple posts collapse to one spike.

Single-account caveat. All findings come from one creator's account. Day-of-week patterns, theme rankings, and hook performance may not transfer cleanly to creators with different audiences (e.g., B2C vs B2B, US vs European focus, technical vs creative). The portable lessons in Section 13 are written cautiously to flag where transfer is most likely (day-of-week, comment behavior, length norms) and where it is less certain (specific theme rankings).

Algorithm drift. LinkedIn's algorithm changes. Findings reported here describe May 2025 – May 2026 behavior. Multi-year stability is not guaranteed.

Reproducibility. Source data, classification labels, and all analysis code are stored in the project archive. Charts in this whitepaper can be regenerated from the saved datasets.