How we work

Editorial Policy

Who is responsible, how stories are sourced and reviewed, and how we correct mistakes.

Last updated: 4 October 2026. The sourcing bar below was tightened the same day: a piece now needs two linked primary sources, and a chart may not combine separate surveys into one series.

Who is responsible

Data Today is published by Datastudy. The responsible editor is Lars Cornelissen, reachable at [email protected]. The responsible editor has final say over everything published here.

Lars Cornelissen works as a data engineer on Snowflake, Matillion, Python, and AWS platforms. That day job is where the Engineering Desk's first-hand notes come from; when a piece draws on it, the article says so in an editor's note.

How we work

Every story starts with a primary source: an announcement, a filing, a paper, a dataset, or vendor documentation. We read the original, extract the actual numbers, and chart them when the data supports a chart. We name and link every source.

Some stories break through another outlet's reporting, such as an exclusive built on unnamed sources. Then we name that outlet in the text and link it, link the underlying report, filing, or company statement as soon as it is public, and say which claims rest only on the outlet's reporting.

We label analysis and opinion distinctly from reporting. When a chart is built on a sample or on illustrative values rather than measured data, the caption says so. A chart shows one measurement. Figures from different surveys, vendors, or years are not plotted as one series or one funnel; that comparison goes in a table that names each source and who it measured.

Our charts redraw figures that the named source published; the caption names that source, and drawing a chart does not make the numbers ours. We call something a benchmark or a test only when we ran it ourselves, and an article reporting one describes the setup so the result can be reproduced. Ratings or scores we assign ourselves are labelled as editorial judgements.

If a piece cannot stand on specific, sourced facts, we do not publish it.

Review before publication

Every published piece is reviewed and edited by the named editor before it goes live. The editor checks that sources are cited and reachable, that numbers match the underlying document, that the framing is defensible, and that the piece adds interpretation rather than restating a press release. Pieces that do not pass are not published.

Before it runs, every piece needs at least two linked primary sources, at least one number specific enough to be wrong, and at least one thing no other outlet has said. A primary source is the document the claim comes from: the announcement, the filing, the paper, the dataset, or the documentation, on the site of the organisation that produced it. A news story, a newsletter, or a blog post about that document is not one, including a consultancy's recap of a survey it did not publish. The full list is under what we will not publish.

The byline on each article names the desk that produced it and links to the responsible editor, and every article links to its sources.

We do not publish content primarily to attract search traffic, and we publish no more than we can review properly: at most 6 articles a week and 25 a month.

The desks

Desk Covers
Data Today AI Desk AI & Research
Data Today Markets Desk Business & Markets
Data Today Engineering Desk Engineering
Data Today Security Desk Security
Data Today Snowflake Desk Snowflake
Data Today Databricks Desk Databricks
Data Today Matillion Desk Matillion
Data Today FinOps Desk Cloud cost & FinOps
Data Today Explainers Desk Explainers
Data Today Guides Desk Guides
Data Today Opinion Desk Opinion

What we will not publish

  • A piece that restates a press release or another outlet's reporting without adding original analysis
  • A piece whose only sourcing is other news coverage
  • A piece we cannot find a first-hand or primary source for
  • A piece that repeats a story we have already covered, unless it changes the picture; in that case we update or link the earlier piece

Corrections

We correct errors openly rather than editing them away quietly.

  • Substantive corrections are noted at the foot of the article with the date and what changed.
  • Corrections and clarifications are logged at /corrections/.
  • To report an error: [email protected].

When we withdraw a piece or fold it into a stronger one on the same story, the old address redirects to the article that replaced it.

Advertising and independence

Advertising on this site is served programmatically. Advertisers have no input into what we cover, how we cover it, or which stories run. Editorial decisions are made without reference to advertising.

We do not publish sponsored content. If we ever do, it will be labelled as such at the top of the piece and will not carry a Data Today byline.

Sources and attribution

We prefer primary sources over coverage of primary sources. Where we rely on reporting by others, we credit and link it. We do not scrape or rewrite other outlets' work.