A mid-size national daily: cutting fact-check time by 60%
Illustrative scenario: a 120-person newsroom uses assessment to focus second reads on the drafts that need them.
This page describes a design target. No newsroom has been named, quoted, or contracted. Contact us to become a design partner.
The challenge
A national daily with 120 editorial staff publishes roughly 180 articles a day across politics, business, culture, and sport. Every piece receives a second read, but the depth of that read varies with the shift, the desk, and the time available.
Senior editors reported spending most of their review time on copy that was already clean, while evidence and sourcing problems in a small number of drafts went unnoticed until after publication. The newsroom had style guidance but no shared way to measure whether a piece met its own standards.
The solution
The proposed pilot assesses every draft against seven dimensions, weighted for its content type. Editors see a short list of exceptions rather than a full report: a missing primary source, a headline that overstates the body, a claim that conflicts with an earlier correction.
High-stakes findings route to a second reader with the evidence attached. Low-risk copy is auto-cleared only above a configured confidence threshold, and the threshold is visible to the desk. Weekly reporting shows which dimensions are improving and where the same problem keeps recurring.
The results
These are design targets, not measured outcomes. A pilot would publish its method and results, including cases where the system was wrong.
What’s next
The next step is a single-desk pilot with a pre-registered evaluation: which findings editors accepted, which they rejected, and where the assessment disagreed with the desk’s own standards.
“The point is not that a model replaces a read. It is that we finally know which drafts need one.”
Not a real customer quote