Newsroom Correction and Accountability Pack
Three documents and two sheets that check a complaint against the evidence, draft the correction and reply, then categorize why it happened: sourcing, analysis or editing.
Free download · No account needed
Every correction policy guide agrees on the mechanics: correct promptly, match the correction's prominence to the size of the original error, and keep an internal log. The Editors' Code of Practice Committee's Clause 1(ii) requires exactly that, a significant error corrected "promptly and with due prominence." None of it asks what the log is actually for once a year of corrections has accumulated inside it. A log that records what changed and stops there cannot tell a newsroom whether its errors cluster anywhere in particular, or simply happen.
This pack adds the one field a conventional log skips: for each correction, was the root cause in the sourcing, the analysis, or the editing. On the worked file, a downtown parking garage contract was reported at $4.8 million, "tripling" the year's $1.6 million parking capital budget. The council's certified vote record puts the contract at $3.2 million, doubling the budget rather than tripling it. The 50 percent overstatement traced to a preliminary staff worksheet cited instead of the resolution that superseded it before the final vote.
Built for editors and reporters who own the correction line, especially anywhere complaints arrive often enough that memory alone cannot hold the pattern. Categorized against the rest of a year's corrections, that same specific failure, a superseded document cited instead of the final record, turns out to be the newsroom's single largest cause, a quarter of everything corrected all year. Pair it with finding verification before a piece runs, and the investigation evidence file for what the assessment note checks against.
What's in the pack
Assessment Note
Checks the complaint against the original evidence itself, not against the complaint's framing of it, with the arithmetic shown wherever the evidence involves a comparison, a ratio or a share of a total.
Correction Wording
An inline version and a standalone version, drafted from the Assessment Note's conclusion, stating the magnitude of the change and naming what does not change alongside it.
Response to Complainant
The direct reply to whoever raised it: what was checked, against what document, and what will run. Separate from the published correction and shorter than it.
Correction Register
One row per correction: date identified, the article, a neutral description of the error, severity, corrective action and date published. The conventional log, kept honestly.
Cause Analysis
The row a conventional log skips: sourcing, analysis or editing, plus the specific cause and the preventive fix, written concretely enough to check whether it actually got followed.
A space rule every prompt reads first
A correction doesn't close until Cause Analysis has a category for it, assigned from what the Assessment Note actually found rather than from how the story reads afterward or who it points at.
How to use it
- 1
Send the complaint and the evidence
The published piece, what the complaint alleges, and the document or record you're checking it against, not a description of it.
- 2
River builds the Assessment Note
What the story said, what the evidence shows, and the arithmetic behind any comparison the correction would change.
- 3
Wording and reply draft together
The Correction Wording and the Response to Complainant both draw from the same conclusion, so neither drifts from what was actually checked.
- 4
Log it, then run the cause
The Correction Register gets the entry; Cause Analysis gets the category and the specific, checkable fix.
Frequently asked questions
Is this template free?
Yes. Download the Assessment Note, Correction Wording and Response to Complainant as Word files and the Correction Register and Cause Analysis as CSV, no signup. "Edit with AI" is a separate, optional path for anyone who wants the agent to check a complaint against evidence and draft everything from it.
How is this different from a standard correction log?
A standard correction log records what changed and stops. This pack adds a category, sourcing, analysis or editing, to every closed correction, plus the specific cause and the fix it points to. On the worked file, that category shows one failure behind a quarter of a year's corrections: a superseded document cited instead of the record that replaced it.
What if the complaint is only partly right?
The Assessment Note separates them. On the worked file, the complaint was right that a contract's dollar figure and its budget comparison were wrong, and the correction fixes both. It was not disputing, and the correction does not touch, the story's separate finding that the contract was awarded without competitive bidding.
How do you pick the category when more than one thing went wrong?
Log the one that would have prevented the error, not every contributing factor. A wrong figure from a stale document that also got rounded oddly on the way to print is a sourcing failure. Fixing the sourcing step prevents the next one; fixing the rounding alone would not.
What should a correction log actually record?
Date identified, the article, a factual description of the error, a severity rating, the corrective action, and the date published, which is the same shape UK JournoHub's guide to building an internal corrections policy recommends. Cause Analysis adds the category and the fix on top of that log.
How prominent does a correction need to be?
As prominent as the original error, not as prominent as the newsroom would prefer. The Editors' Code of Practice Committee's Clause 1(ii) requires a significant inaccuracy corrected "promptly and with due prominence," and a front-page claim needs a correction an inside-page reader would actually see.
Where does this sit in the rest of the workflow?
After a piece runs and a complaint arrives. Before publication, the fact check pack verifies assertions against sources and finding verification puts findings to the subject. If the disputed number came from a dataset, the analysis pack is where to check the figure was computed correctly in the first place.
Give every correction a cause, not just a fix
Download the blank note, wording, reply and sheets as Word and CSV files, or open this exact pack in River and send it the complaint you're checking.
Edit with AI