How we write: method and sources
This page describes how an UpSkillLab article actually gets made. I wrote it because a huge amount of unsourced content circulates on work and career topics, and the only way to stand apart is to make the process verifiable.
1. Choosing the topic
An article only happens if there’s a real question to answer and if I have enough material to answer it better than what’s already online. If, after the research, I find I could only rewrite what others say, the article doesn’t get published. That’s why this site publishes few articles and updates them, rather than shipping one a day.
2. Sources I use
In order of priority:
- Primary institutional sources — World Economic Forum, OECD, Eurostat, national statistics offices, the World Health Organization, the European Commission.
- Peer-reviewed research — studies published in scientific journals, cited with authors and year.
- Official product documentation — for articles on software and AI tools, the vendor’s documentation rather than third-party reviews.
- Private-sector reports (LinkedIn, McKinsey, Gartner and similar) — used with caution and always flagged as research commissioned by a party with commercial interests in the field.
Every figure reported has a link to its source and a reference year next to it. If a figure isn’t verifiable, it doesn’t go in the article.
3. What we don’t do
- We don’t invent statistics. Phrases like “studies show” without a linked study don’t appear in the articles.
- We don’t copy or reshuffle other people’s articles. The texts are original.
- We don’t publish anything without human review. See the next point.
4. How we use AI
It would be hypocritical, on a site that covers AI at work, to pretend not to use it. So here it is, plainly: the texts on this site are drafted with the assistance of artificial intelligence tools, used for researching background material, for the writing, and for the English and Spanish versions.
What stays human is what determines the value of an article:
- the choice of topic and the angle it’s approached from;
- the selection of the sources cited, which are public and linked in the text;
- the review of every article before it’s published;
- editorial responsibility for what’s published, which is mine and nobody else’s.
This disclosure follows Article 50 of Regulation (EU) 2024/1689 (the AI Act), which asks anyone publishing AI-generated or AI-manipulated text on matters of public interest to make that clear to readers. The same article provides that the obligation doesn’t apply where the content has undergone human review and a person takes editorial responsibility for it: both are true here, but transparency still seems the right call.
What this does not mean. It doesn’t mean the text is infallible: language models can state wrong things with confidence, which is exactly why every figure has its source next to it, so you can check it without taking my word for it. If you find an error, report it through the Contact page: it gets corrected and the correction is noted.
5. Updates and corrections
Articles containing market data are revised when a new edition of the source is released (typically once a year). The last-updated date is visible at the top of every article.
If you report an error, the correction is made and noted openly at the bottom of the article with the date, rather than being changed silently. You can report errors through the Contact page.
6. How we’re funded
UpSkillLab is free and supported by display advertising and occasional affiliate links. Ad slots are handled by external networks and have no influence whatsoever on which topics we cover or what we write. Any affiliate links are explicitly flagged where they appear, and we don’t recommend products just because they pay a commission.
Who writes here is stated on the Author page.