Editorial policy
How we earn trust.
A plain-language guide to how EconToMarks researches, verifies, publishes and corrects its work.
Sources first
We prefer primary sources such as national statistics agencies, central banks, government departments and international organisations. Secondary reporting can add context, but it is labelled as such.
Original summaries
We write our own summaries and do not copy source articles. Links are provided so readers can check the underlying evidence themselves.
Story dates
The date shown publicly on a Story is the date of the economic release, policy decision or development it covers. If EconToMarks adds verified coverage later, the Story is placed in the feed by that relevance date rather than the day it was added. The actual EconToMarks publication and update timestamps are retained separately for editorial audit and technical metadata.
Mandatory publication standard
Every Story must pass all 12 checks.
The final test is simple: if a student saved only the Story, would they understand what happened, know exactly what each important number refers to and be able to use the evidence accurately? If the answer is not clearly yes, it should not publish.
- 1
Purpose
The story must cover a meaningful economic development. Article count is never a reason to publish. A student should be able to identify what changed, why it matters and how the evidence could be used.
- 2
Primary-source verification
Core claims and numerical evidence must be checked against an approved authoritative source. Forecasts, expectations, estimates and realised data must never be presented as if they were the same thing.
- 3
Time and comparison clarity
Every important statistic must identify the relevant month, quarter, year, survey period or forecast horizon and, where needed, the comparison basis such as month-on-month, year-on-year or previous quarter.
- 4
Evidence quality
Key data must be student-readable evidence rather than database fields or release metadata. Internal labels, snake_case, vague headings and unexplained values cannot be published.
- 5
What happened
The story must explain the actual outcome, the most important figures and the relevant comparison instead of simply reproducing institutional wording.
- 6
Why it matters
The article must connect the development to an economic mechanism such as aggregate demand and supply, inflation expectations, labour markets, trade, productivity, real incomes, investment or policy transmission.
- 7
Context and limitations
Where relevant, the story must explain revisions, volatility, survey uncertainty, base effects, methodology, nominal-versus-real distinctions, alternative measures or other limits on interpretation.
- 8
Exam usefulness
Evidence, explanation and evaluation must each make sense on their own. A student should be able to copy the evidence into revision notes without having to decode the page first.
- 9
Student language
Public wording must be clear economics English with explicit units, definitions where needed and no internal database or pipeline terminology.
- 10
Duplicate control
One canonical story should represent an economic event unless a later estimate or revision materially changes the evidence. Redundant coverage is held rather than published for volume.
- 11
Image integrity
A published story must have one relevant, active and unique editorial image. The institution, geography and subject must be correct, and used images are not recycled into unrelated stories.
- 12
Fail-closed confirmation
If any mandatory check fails, publication fails. Automatic publication does not receive a weaker standard than manually reviewed coverage.
Database-level publication gate
The objective parts of the standard are enforced at the publication boundary. Published stories must have approved primary-source evidence, substantive contextualised key data, sufficient explanatory sections, valid event timing and exactly one active editorial image. Machine-facing key-data labels, placeholder wording and missing temporal context cause publication to fail.
Governed automatic publication
Only a small allow-list of highly structured official releases can be published automatically. A release must come from an approved primary source and pass source-domain, freshness, classification, score, duplicate, completeness, evidence and release-specific parsing checks. Automatic publication must also leave a governed publication record. If any required check fails or the source format is uncertain, the story stays private for review.
AI-assisted workflows
AI may assist discovery, classification or private drafting, but AI is never treated as a source. The zero-click publication path for structured releases uses deterministic source-specific checks rather than asking a general AI model to decide whether a story is true. Broader or uncertain coverage remains subject to editorial review.
Updates, corrections and retractions
Recently automated stories can be rechecked against their official source. If structured official values change, the existing story can be updated with visible correction metadata rather than silently replacing history. If a publication should no longer be relied upon, we retract or return it to review rather than keeping it public to preserve article count.