Original Data Studies for Legal, HR and Education Press

One study, one dataset, one embargo. Across the six studies this practice has published since 2023, the median was 34 referring domains in the first ninety days and 61 by month eighteen. The later half arrives because a study built on a public register keeps being cited long after the news cycle has moved, and a poll of a thousand people does not. That is the entire argument for how we work, and it is why we do not commission panels. What the studies are built from: Registers and administrative data rather than questionnaires. Broennoeysundregistrene annual accounts and role registers, SSB and Statistics Sweden labour and education series, Eurostat, Companies House filings and officer histories, HESA student records, IPEDS, the UK Employment Tribunal decisions database, Lovdata, published SRA and Bar Standards Board enforcement data, NAV vacancy series. The point is not that these are free. It is that they are checkable, so a journalist at DN or Times Higher Education can verify the number themselves, which is what lets them run it under their own byline. The analysis: Done in R or Python, in a notebook handed over with the study. Every figure in the release traces to a line in that notebook. A statistician outside the project reads the method before anything is drafted and their name goes on the methodology page. Where the data will not support the headline, the headline changes. That is the moment most studies quietly break, and it is why about a third of the angles we scope are dropped before writing. What ships: A findings page on your domain at a stable URL. A methodology note that states what the data cannot show as well as what it can. The cleaned dataset as CSV under CC BY 4.0, so an outlet can republish a table without asking anybody. Five to eight charts built in Datawrapper with embed codes. A release written for a trade desk rather than for a wire. The launch: One exclusive to one outlet under embargo, usually forty-eight hours, chosen because its readers are the ones you need rather than because it is the largest. Then the trade desks the day the embargo lifts, then the Nordic business press with a localised cut of the same data. A study handed to everybody at once gets covered first by nobody, and being first is exactly what the second and third outlets are looking for. What we refuse: No survey panels dressed as research. No polls of two thousand adults, no bought audiences, no questions written to produce a finding somebody already decided on. No study without a named analyst and a reproducible method. We do not run subgroup cuts until one clears significance. We do not publish a finding we would not defend to a statistician who dislikes us. When the data comes back dull we say so, and you have paid for analysis rather than a press release. That has happened twice, and both clients used the finding internally. Who this is not for: Anyone who needs a fixed publication date more than a defensible number. Anyone whose legal function will not let an external analyst near the data. Anyone who wants their brand in the headline, because journalists cover the finding and not the sponsor; you are credited in the source line and in the link, and that is the trade. Out of scope: Ongoing media relations, reactive commentary, and any promise about rankings. A study earns citations. What your site does with them is your site's problem, and you will hear in the scoping call if we think the answer is nothing.