LinkedIn
Best Practices
A question in a daily meeting, an email that arrived at the right time and a way of approaching work: research, learn and share what you learn.
- Origin
- A team conversation · B2B
- My contribution
- Research, judgement and editorial strategy
- Applications
- Guide and open skill · 2026
- Authorship
- Jmillans
Context
It all started with a team conversation.
Information moves fast on social media. Staying informed while keeping up with platform changes and trends is difficult. We work primarily in B2B, and LinkedIn is part of our daily work, both organic and paid.
During a daily meeting, a colleague opened a debate: how do we build a brand’s authority? What roles should the company Page and personal profiles play? How can we use both voices, and what do we really know about how the platform values them?
The question had a strategic consequence: deciding how to build authority to develop the digital funnel and, through it, the brand. The discussion ran on, and we ended the meeting promising to return to it with more information.
Later, back at my desk, I received a Metricool email about its analysis of LinkedIn interactions: commenting, sharing and when that activity happens. Newton’s apple hit me on the head. The newsletters I already subscribed to could be the starting point for structured, verifiable research.
Challenge
Researching without yet knowing what I would find.
At first, the aim was to gain knowledge and return to that conversation with informed judgement. There was no guide or skill on the horizon. There was curiosity and the possibility of finding something that could help us decide.
The first task I gave Codex was simple: review Metricool emails in Outlook, identify relevant LinkedIn content and gather studies, analyses and resources worth reading. Starting with organised material prevents research from becoming a miscellaneous drawer.
After the search, I read the material and began diving into the data. We developed an analysis and conclusions, then paused to reread them. That checkpoint changed the scope: something was taking shape that could help us align as a team on our strategic approach.
The challenge became turning the information into something clear and usable while preserving the context needed to interpret each recommendation.
Direction
Research. Learn. Teach.
“Knowledge gains value when it circulates.»Jmillans
Those three actions sum up how I work. I see leadership as developing the talent and potential of the people I work with, rather than being the person who knows the most. Knowledge gains value when it circulates.
Being able to rely on your team, knowing you are not alone and sharing what you discover is immensely valuable.
That is why I decided to turn the analysis into a best-practice guide for the team. I wanted a balance: recommendations people could look up, and short paragraphs explaining why. Something understandable and manageable, with enough depth to support decisions.
AI helped me gather and compare information. Selection, reading and editorial judgement remained part of my work. The direction was to build a shared foundation everyone could understand, discuss and use.
System
Cross-check, distinguish and reread.
One source was not enough. I expanded the research with material from Sprout Social, Hootsuite and LinkedIn itself. I also considered sources such as HubSpot; for this guide I chose to focus on studies and documentation from platforms specialising in social media. As specialists, they should be best placed to answer our questions.
The process involved cross-checking sources, understanding their samples and separating study findings, platform explanations and our editorial recommendations. Then reading again. The aim was to distinguish what we could claim, what needed context and what we should test with our audience.
One example was the discrepancy between 11× and 17× in different Metricool materials about carousels. We retained the 11× supported by the full report and recorded the difference. We also clarified engagement formulas and distinguished cross-platform data from LinkedIn-specific findings.
The guide was organised around voices, content, formats, links, frequency and measurement, with visible sources and dates. I will review potential changes every six months and consolidate new editions when the evidence justifies them. Keeping it up to date is essential. The first edition’s reference date is September 2026.
The research journey
01
Collect
AI-generated illustrative setting · Actual sources and resources appear below.
Newsletters, reports and documentation in an organised foundation.
Understand each source’s contribution and how its sample was built.
Cross-check figures, definitions and scope before drawing conclusions.
Turn research into criteria people can apply.
Reread conclusions and keep editions up to date.
Applications
Letting knowledge support everyday work.
I was already reading and rereading the guide when the next idea emerged: reading it is essential, but what if I also turn it into a skill that helps plan and create content?
The skill brings those criteria into an agent: reviewing a draft, proposing a calendar or comparing results. It is an application of the research we can share with the team to work from a common foundation, adding each brand’s context and each professional’s judgement.
A Saturday that began with a question about work ended with a guide and an open tool. By that point, yes, I was incredibly excited.
Each resource has its place: the Lab research article explains the findings and sources; the skill article shows how to install, use and update it. This case traces the professional journey connecting the initial question with both applications.
SKILL.md v1.0.0 · Jmillans · MIT
# LinkedIn Best Practices Guide
Help the user make editorial decisions and produce useful, credible LinkedIn content. Adapt the depth of the work to the request: a quick post review does not require a full strategy.
## Decide what to publish
- Establish the audience, intended outcome, topic, and author or account. Infer routine details from the user's context; ask only when a missing choice materially changes the result.
- Prefer original knowledge, examples, evidence, and a recognizable point of view. Choose a format because it helps explain the idea, not because a study found a higher average for that format.The skill.
The same criteria, available in the agent you work with every day.
Outcome
Sharing it beyond the team.
When I started in photography and design, I met people who hid their knowledge for fear that others would improve. That way of guarding the master’s secrets always saddened me.
Sharing what I know and discover makes me better. Explaining it also reveals things about myself; your perspective opens angles and conclusions I might never have found alone, deep in the cave, watching the shadows.
Once it was ready for the team, the next step was to make it available to any professional who could benefit from it. The guide was finalised on 29 September 2026. We published the skill on GitHub, authored by Jmillans under the MIT licence, and verified its installation and availability on skills.sh.
The outcome is a resource that can be consulted, used and improved. There are no measured results yet on its impact on teams or audiences. My intention is to keep updating and sharing it: together we can make our profession stronger, more respected and enjoy the journey.
Published to be shared.
The work behind
the resource.
Sharing what I know and discover makes me better.
Documentary research and critical data reading; content strategy and information architecture; using AI to turn knowledge into a reusable tool. All in service of a way of leading: sharing judgement and helping the team develop its own.
I hope you find it useful and can put it to work. If you use it, I would love to hear about your experience, the questions it raises and what you would improve. Continuing to learn is part of the project too.