LinkedIn for Engineers: What Actually Works, and What Is Folklore

Most LinkedIn advice is folklore with a number attached

Every few weeks a list of LinkedIn "cheat codes" does the rounds. Post on Sunday. Use emojis, they perform 220% better. Never write fewer than fourteen paragraphs. Some of it is sound. A surprising amount traces back to no study at all, and a few of the most confident claims are the opposite of what large datasets show.

For engineers this matters more than it seems. You do not need to become a creator. But a clear profile and an occasional useful post compound quietly: recruiters find you for the right roles, people remember what you are good at, and opportunities arrive that you never applied for. Doing that well takes a few hours a month, provided you spend them on what actually works.

So here is the list, checked. What LinkedIn has said publicly, what large studies found, and what is folklore.

How the feed works now

In March 2026 LinkedIn published a detailed engineering write-up of its rebuilt feed. Three points from it shape everything else in this post.

It reads your post and your profile. Each post is represented using its text along with the author's headline, company and industry. Each reader is represented by their skills, work history and the posts they have engaged with. The system then matches the two.

It shows people content from outside their network. The ranking deliberately surfaces posts from authors you do not follow, on topics you care about. Your reach is no longer capped by your follower count, but it depends heavily on being clearly about something.

It measures attention, not just clicks. Signals include how long someone stays on a post (LinkedIn has written separately about dwell time), likes, comments, shares and skips. A post people scroll straight past gets fewer chances.

In the same week, LinkedIn's head of feed said the company would reduce "Comment 'Yes' if you agree" posts and repetitive, click-driven content. In July 2026 it added a button to report AI slop and began limiting low quality AI-written posts shown outside the author's network.

Put together: be clearly about something, write things worth stopping for, and do not game it.

Your profile is the landing page

Every post that works sends people to your profile. If they cannot tell in five seconds who you are and what you do, the post did its job and your profile wasted it. The profile also feeds ranking directly, so this is not just cosmetics.

Headline. Say what you do and for whom, in words a recruiter would search. "Backend engineer · payments and distributed systems · Go, Postgres, Kafka" beats "Passionate technologist | Lifelong learner". Your headline travels with every post and comment you make, so it is the most read line on your profile.

Photo and banner. A clear, well lit photo of your face. A banner that reinforces your niche, even if it is just a line of text, beats a default blue gradient.

About. Three short paragraphs. What you work on and the problems you are good at. Two or three concrete results, with numbers where you have them. What you are interested in now, and how to reach you.

Experience. Outcomes, not responsibilities. "Cut checkout latency from 1.8s to 400ms by moving session reads to Redis" tells a reader what you can do. "Responsible for checkout performance" tells them nothing.

Skills. Keep the list accurate and relevant, since skills are one of the things the feed and recruiter search both use. Twenty honest skills beat fifty aspirational ones.

Featured. Pin your best work: a project, a write-up, a talk, a popular post. There is no evidence it affects ranking, but it is the first proof a visitor sees.

Pick a lane, at least for now

The advice to post about one thing until you have an audience holds up, and the reason is mechanical rather than mystical. A system that matches posts to readers by topic can only match you if your topic is recognisable. An engineer who posts about Kubernetes on Monday, productivity on Wednesday and their holiday on Friday gives it nothing to work with.

Pick a lane you can sustain: the stack you work in, a problem you keep solving, the stage of career you are in. "What I learned moving a monolith to services", "Postgres things I wish I had known sooner", "notes from my first year as a tech lead". You can widen later. A narrow, consistent start is how people learn to associate your name with something.

Writing posts people stop for

The first line does most of the work

On mobile, roughly the first 140 characters show before "see more", and around 210 on desktop. LinkedIn does not publish these limits, and line breaks can cut the preview shorter, so treat the first sentence as the entire pitch.

Good openers are specific and create a small, honest gap:

• "We cut our AWS bill by 38% and the biggest win was deleting things."
• "I reviewed 200 pull requests last quarter. The same five comments came up every week."
• "The worst production incident of my career was caused by a single default timeout."

Weak openers are generic ("Excited to share...") or promise something the post does not deliver. The second kind is worse, because a reader who clicks and immediately leaves is itself a negative signal.

Write the way you would explain it to a colleague

There is no evidence that LinkedIn rewards a "conversational" tone as such. What the data does support is plain language: short sentences, familiar words, one idea per paragraph. Posts from the most followed accounts tend to read at a very low grade level, not because readers are simple but because people scan the feed on a phone between meetings.

Break text into short paragraphs so it is easy to scan. The "fourteen paragraphs" rule, like most exact numbers in this genre, has no source. Use as many as the idea needs.

Length: long enough to say something

LinkedIn posts can be up to 3,000 characters. A 2026 analysis of 372,000 posts by AuthoredUp found the highest median engagement in the 1,300 to 2,500 character range, with very short posts lowest. That is a useful range, not a rule. A crisp 600 character post with one sharp insight beats 2,500 characters of padding.

One post, one idea, one takeaway

This is the best advice on the original list. If a post has two ideas, it is two posts. Decide before writing what a reader should think or do differently after reading, and cut everything that does not serve it.

Proof of work beats proof of authority

For engineers especially, the posts that travel are the concrete ones. A diagram of how you designed something, the numbers before and after a change, a mistake and what it taught you, a code snippet that solves a common problem. "Five tips for clean code" reads like a thousand other posts. "Here is the refactor that removed 4,000 lines from our billing service, and the one test that made it safe" does not.

Results over credentials. Nobody needs to know your title to learn from what you built.

Formats: what the data shows

Document posts. Since LinkedIn retired native carousels in late 2023, the "carousel" is really a PDF uploaded as a document. In Socialinsider's 2026 benchmarks across 1.3 million company page posts, documents had the highest engagement rate, followed by multi-image posts and video, with text-only and link posts lowest. That is company page data and engagement rate rather than reach, but creator datasets point the same way. For a step by step explanation, an architecture walkthrough or a checklist, a document post is a good choice.

Images. LinkedIn accepts aspect ratios from 3:1 wide to 4:5 tall, and recommends a width of 1080 pixels (LinkedIn Help). A 4:5 portrait image such as 1080x1350 takes up the most room in a mobile feed, which is a sensible default. Claims of a specific reach boost from portrait images are not backed by anything solid.

Infographics. The claim that infographics are "the most likely format to go viral" traces back to generic, years-old social media statistics, not LinkedIn data. A clear diagram is valuable because it explains something quickly, which is reason enough.

Video. LinkedIn has pushed video hard, including a dedicated vertical video feed, but video views on company pages have fallen in some datasets. If talking to camera suits you, it can work. It is not a shortcut.

Small things that are mostly myths

Emojis "perform 220% better". No study could be found for this number. It appears to be a garbled version of old statistics from other platforms. The largest LinkedIn specific analysis available, from MagicPost across 1.2 million posts, found a modest lift of around 15% in median likes with one or two emojis, and nothing extra beyond that. Use one if it helps the reader scan. Do not decorate every line.

Hashtags. LinkedIn removed hashtag following and hashtag feeds in 2024, and studies find little to no reach effect. The original list is right: skip them, or use one or two at most.

External links. LinkedIn has said it does not deliberately penalise links. Reach data consistently shows posts with a link preview card get fewer impressions, likely because they send people off the platform. One recent analysis found the drop came from the preview card itself rather than the URL. There is no evidence for a "maximum three links" rule. If the link is the point, include it; if it is a nice-to-have, put it at the end or in a reply.

Generic calls to action. Here the original advice is strongly supported, and LinkedIn has said so directly. "Agree?", "Comment YES" and "Repost if this helped" are exactly what the feed is being tuned to demote. A genuine question you want answered ("How does your team handle flaky tests?") is fine, and often produces the best comments.

When and how often to post

Sunday is not the best day. This claim is one of the clearest misses. Buffer's analysis of 4.8 million posts (September 2026) and Sprout Social's analysis of around two billion engagements both found weekends among the weakest days, with midweek strongest. Buffer's best single slot was Wednesday afternoon.

Forget "11AM to 1PM GMT". Both large studies measure the audience's local time, not a single global clock, and they disagree with each other: Sprout found strong performance across Tuesday to Thursday working hours, while Buffer found the peak shifting later into the afternoon and early evening. If your audience is mostly in one region, post during their working week and test a few slots. Your own analytics after a dozen posts are worth more than any study.

Frequency. Buffer's study of over two million posts found two to five posts a week is a practical sweet spot, and that posting more often did not reduce how each individual post performed. The popular rule about never posting twice in 24 hours is not confirmed by LinkedIn or the larger datasets.

For most engineers, though, the honest answer is: pick a pace you can sustain for six months. One good post a week, every week, beats three a week for a month and then silence. Consistency without quality is noise, but quality without consistency is invisible.

Engagement is a two way street

Commenting thoughtfully on other people's posts is the most underrated growth habit on the platform, and it does not need a hack to justify it. Your headline appears with every comment, so a useful comment on a popular post in your field is seen by exactly the people you want to reach. It also teaches the system which topics you care about.

A useful comment adds something: an experience, a counterexample, a clarifying question. "Great post!" adds nothing. Ten good comments a week will do more for most people than an extra post.

On your own posts, reply to comments properly, especially early. Conversations keep people on the post, and time spent is one of the signals LinkedIn has confirmed it uses. The idea of a strict "golden hour" that decides a post's fate is weaker than it used to be, since posts now often find readers over several days.

The advice to "build relationships with authority figures so their comments signal quality" has no evidence behind the mechanism. Building real relationships with people further along in your field is simply good for your career, and that is reason enough.

Using AI without sounding like it

AI can help you outline, tighten and proofread. It is a poor substitute for having something to say. LinkedIn now lets members report AI slop and limits low quality AI-written posts shown outside your network, and readers have become quick at recognising the style: a rhetorical opener, three bolded lessons, a question at the end.

The fix is the same as for writing well in general. Start from something that only you could write: a specific system, a specific number, a specific mistake. Use AI to sharpen that, not to generate it.

Build a swipe file

Save posts in your field that make you stop scrolling. After a month, look at them together. You will see patterns in the openers, the structure and the kind of detail that makes something credible. That is a better education in writing for this medium than any list of rules, including this one.

The original list, scored

Holds up: write plainly, lead with a strong first line, have a clear niche, optimise your profile like a landing page, skip hashtags, avoid generic calls to action, one post one idea, comment on others' posts, show proof of work, save posts you admire, and stay consistent.

Partly true: document posts perform well (as engagement, mostly measured on company pages), portrait images take more space (reach boost unproven), 1,250 to 3,000 characters is a reasonable range (not a rule), and two to five posts a week is sensible (not a minimum of three).

Not supported: Sunday as the best day, 11AM to 1PM GMT, emojis performing 220% better, career posts being three times more likely to go viral, infographics being the most viral format, fourteen or more paragraphs, a maximum of three links, and an eight to ten year old reading level as a hard target.

The short version

Make your headline say what you do in searchable words. Rewrite your experience as outcomes. Pick one lane. Put the whole pitch in the first 140 characters. Write plainly, one idea per post, with something concrete only you could say. Use document posts or a clear diagram when the idea is visual. Skip hashtags and engagement bait. Post midweek in your audience's working hours, at a pace you can keep for six months. Leave good comments on other people's posts. And ignore any tip that comes with a suspiciously precise percentage and no source.

LinkedIn growth is not complicated. It is being clearly useful about one thing, repeatedly, until people start to associate your name with it.

Cover photo by Shutter Speed on Unsplash.