How articles in the Full Stack Recruiter Newsletter are researched, tested, fact-checked, corrected, and disclosed, including how AI is used and who is accountable for what gets published.
Advice about recruiting, hiring, sourcing, and recruiting technology gets repeated far more often than it gets tested. Statistics move from presentation to presentation, LinkedIn post to LinkedIn post, and article to article until nobody remembers where they originally came from. Claims about hiring, AI, interviews, sourcing, and recruiting technology can become accepted as fact simply because enough people repeat them.
This page explains how I try not to contribute to that problem, how articles in the Full Stack Recruiter Newsletter are created, and what happens when I get something wrong.
Who writes this
Every article in this publication is written and edited by me, Jan Tegze.
There is no ghostwriting team, contributor network, or guest-post program. I have worked in talent acquisition and recruiting for more than 20 years and have written ten books on recruiting and job searching, including Full Stack Recruiter and the Job Search Guide.
That means the byline is also the accountability. I decide what gets published and I am responsible for the final content.
If something here is wrong, it is mine, and you can tell me directly.
How claims are supported
When an article relies on something I tested myself rather than something I read elsewhere, I try to make that distinction clear.
Where relevant, those articles include information explaining:
What kind of evidence it is: a first-hand test, professional experience, documented analysis, or another form of evidence.
The method: what was actually done, in enough detail for someone else to understand or potentially repeat it.
The scope: how many things were tested, reviewed, or analyzed, and against what.
The environment and date: the tools, platform, browser, AI model, conditions, or other relevant environment used, and when the test was conducted.
The purpose of publishing the method is to let you judge the result rather than simply take my word for it.
A test of 55 LinkedIn profiles is a test of 55 LinkedIn profiles. It is not automatically evidence for every recruiter, company, industry, country, profession, or platform. Where the scope matters, the article should make that clear.
Limitations are stated, not buried
Articles with meaningful limitations should state them where readers can reasonably see them rather than hiding them in a footnote.
A test conducted in one country, in one year, using one browser, one recruiting platform, one AI model, or one type of role does not necessarily generalize beyond those conditions.
Where that matters, the article explains the limitation.
The same applies to the age of the research or test.
Recruiting technology changes quickly. Search engines change, LinkedIn changes, AI models change, applicant tracking systems change, and products can disappear, merge, or be renamed.
Several articles in this publication document tests conducted at a particular point in time. Their results may still be useful as a historical snapshot, but they should not automatically be treated as a description of how the same system works today.
Where I know that a material change has occurred, I try to make that clear to the reader.
Sources and fact-checking
When a factual claim comes from someone else’s research or data, I try to link to the original source rather than to another article quoting it.
That means preferring, where available:
original research papers
official datasets
company documentation
government publications
regulatory documents
original surveys or reports
direct statements from the organization responsible for the information
If an original source has moved or disappeared, I may link to an archived version so readers can still inspect it.
Statistics that circulate online without a traceable origin are not treated as facts simply because they have been repeated many times.
If I cannot determine where a number came from, I will either avoid using it, clearly describe the uncertainty, or investigate the claim itself.
AI-generated answers are not treated as sources. If an AI system provides a statistic, quotation, research claim, or factual assertion, the underlying source must be checked independently before that information is relied upon.
Corrections
I correct meaningful errors publicly on the page where they appeared rather than silently rewriting history.
One example is the widely repeated claim that visual information can supposedly be processed “60,000 times faster” than text, often attributed to 3M.
I had repeated the claim myself. While fact-checking my second book, I went looking for the underlying research and could not find evidence supporting the number. Attempts to trace the claim back to its supposed source did not produce the study behind it.
The claim was not supportable, so I corrected it rather than continuing to repeat it.
If you spot an error, send it to me.
What happens next depends on the type of change:
A factual error is corrected on the page. Where the error materially affected the article, the correction is disclosed rather than made invisible.
A substantial revision, such as new testing, new data, or a changed conclusion, receives a visible update date and an explanation of what changed.
Small fixes, such as spelling mistakes, broken links, formatting problems, or other changes that do not alter the meaning of the article, may be corrected without a formal correction notice.
An article is not backdated or given a new publication date simply to make old work appear new.
If an article displays an updated date because of a meaningful editorial revision, readers should be able to understand what was updated.
Sponsored content and partnerships
I do run brand partnerships, including newsletter sponsorships, sponsored content, partnerships, and paid product reviews.
When content is paid for or produced as part of a commercial partnership, that relationship is disclosed.
Payment buys my time, attention, testing, or access to my audience. It does not buy a positive conclusion.
If a paid review identifies meaningful problems with a product, those problems can still be included. I may also decline partnerships when I do not believe the product or service is relevant or useful to my audience.
Recommendations in unpaid editorial articles are not automatically commercial placements. When I recommend or discuss a tool independently, it is because I have used it, tested it, researched it, or believe it is relevant to the subject being discussed.
Where an affiliate or other commercial relationship exists, it should be disclosed.
AI use
Because English is not my first language, I use AI to help correct grammar, improve flow, and catch typos, which means AI-detection tools such as Pangram may sometimes label my writing as AI-generated even when the ideas, arguments, and original text are mine. I created this page in part to be transparent about that use.
I am the author and editor of every article published in the Full Stack Recruiter Newsletter.
I use AI tools to assist with parts of the editorial process, including research assistance, brainstorming, outlining, drafting, editing, summarizing source material, identifying questions worth investigating, or challenging my own reasoning.
AI does not have editorial control over what is published.
I review, revise, and approve the final article before publication. I decide which arguments, conclusions, opinions, examples, and recommendations appear in it, and I take responsibility for the final published content.
AI-generated claims are not considered evidence simply because an AI system produced them. Important factual claims should be checked against the underlying research, documentation, data, testing, or another appropriate source.
AI systems are capable of inventing sources, misrepresenting research, confusing correlation with causation, and confidently producing incorrect information. Their output is therefore treated as something to evaluate, not as an authority.
Nothing is automatically published in the Full Stack Recruiter Newsletter by an AI system without human editorial review.
AI-generated media
AI tools may be used to create or modify illustrations, graphics, images, audio, or other media used in this publication.
Where AI-generated or AI-manipulated media could reasonably be mistaken for documentary evidence, a real event, an authentic photograph, a genuine recording, or something that actually occurred, it should be clearly identified as generated, manipulated, illustrative, or otherwise synthetic.
AI-generated illustrations used purely as visual representations of an idea are treated as illustrations and are not presented as evidence that the depicted event occurred.
AI-generated media is not used to fabricate evidence, research results, screenshots, candidate experiences, quotations, documents, recordings, or events and present them as authentic.
Where an article relies on an actual screenshot, test result, document, dataset, recording, or other piece of evidence, it should be distinguishable from illustrative AI-generated material.
Editorial independence
The existence of a commercial relationship, access to a product, advance information, free access, or sponsorship does not transfer editorial control to another company.
Unless explicitly stated otherwise, companies discussed in the Full Stack Recruiter Newsletter do not approve editorial conclusions before publication.
Where a company is given an opportunity to clarify a factual question before publication, that does not give it control over the final article.
What this policy does not cover
This policy explains how editorial content is created, reviewed, corrected, and disclosed.
Questions or corrections
If something in the Full Stack Recruiter Newsletter looks wrong, unsupported, misleading, or out of date, contact me.
I would rather investigate a credible challenge and correct an error than defend something simply because I published it first.