Press Release Builder and Fact Checker

A press release is an official statement handed to the news media so that they can report something new. It is also a primary source: the first written account, issued by the people the news is about. That double role explains nearly everything about the form. A reporter may copy a sentence straight into a story, quote it under a byline, or check it against a filing. Every figure and every date in it is therefore a promise made in public.

The modern form is usually traced to Ivy Lee, who issued a statement in October 1906 about a railroad accident near Atlantic City that killed some fifty people. What made it notable was candour: The New York Times printed his statement as he wrote it. Since then the release has become the routine way for companies, agencies and courts to announce decisions, results and events.

The parts, and why each one exists

A release is a fixed sequence of parts, and the sequence exists for the reader on deadline. The headline summarises the news so that an editor can decide in a second whether to read on. An optional subheadline adds one supporting fact. The dateline names the city the release comes from and the date it is issued, placed in front of the first sentence, the same way a wire story begins. Then comes the lead, the first paragraph, which should answer who, what, when, where and why on its own. The body supplies explanation, statistics and background, often in four or five paragraphs.

After the body sits the boilerplate: a short “about” section describing the issuer. The word comes from printing. Newspapers once received ready-cast metal plates of syndicated text, and the plates were called boilerplate by analogy with rolled boiler steel. Today it means text reused unchanged, which is exactly what this paragraph is. It is written once, approved once, and pasted under every release.

The close marks the end. In North America the traditional marks are -30- or ###, placed after the boilerplate and before the media contact; elsewhere the word “ends” does the same job. The media contact gives a name, a phone number and an email address, so a reporter with a question has somewhere to send it. Each of these parts is small. Missing one still costs a newsroom time.

Length varies with the news. A common guide is four to five body paragraphs and 400 to 500 words, with real releases ranging from about 300 to 800. Length is not the hard part.

Embargoes, and the clocks they depend on

Some releases go out early under an embargo: a request that the news not be published until a stated date and time. Embargoes give reporters time to research and prepare an accurate story, and they let a source release complicated material to everyone at once. A release marked “For immediate release” says plainly that no embargo applies. One whose dateline is later than the day it was sent is asking for an embargo, whether or not it says so in words.

An embargo is usually a gentlemen’s agreement. Unless a journalist has signed a binding agreement, nothing in law stops publication. What enforces it is the relationship: an outlet that breaks an embargo can be cut off from advance information for a long time. That makes the stated time the heart of the arrangement, and a vague time is an invitation to argue about it later.

Time is where releases quietly go wrong. Newsrooms sit in different zones, and the offsets between them change during the year. Briefings from the International Monetary Fund, for example, are typically embargoed until 10:30 a.m. Washington time, or 14:30 GMT, so that markets everywhere see the news at one instant. Writing both forms is the safe habit. A local time alone is weaker, because the offset behind a name such as Eastern Time shifts twice a year.

The shifts are not synchronised. Members of the European Union change their clocks together, at 01:00 UTC on the same Sunday, which keeps European offsets constant. North America changes each zone at 02:00 local time, and on different Sundays from Europe. Since 2007 most of the United States and Canada have kept daylight saving time from the second Sunday in March to the first Sunday in November, while Europe has used the last Sundays of March and October. For a few weeks each spring and autumn, Paris is five hours ahead of New York instead of six.

The authority for all of this is the time zone database maintained under the Internet Assigned Numbers Authority, often called tz or zoneinfo. It records the history of local time for representative places, names zones by region and city, as in America/New_York or Europe/Paris, and records when daylight saving time was in use. Its code and data are in the public domain.

Governments change their rules, sometimes with little notice, so new editions appear several times a year; the 2026e edition, released on 29 September 2026, notes that Manitoba moves to a permanent offset of five hours behind UTC on 31 October 2026. The standard date formatting interface in web browsers accepts any of its zone names, so one page can show a single instant in many zones. The answer is only as current as the zone data the browser holds.

Two kinds of local time need care. When clocks spring forward, an hour of local time never happens; when they fall back, an hour happens twice. The iCalendar standard settles both cases for calendar files: a repeated time means its first occurrence, and a skipped time is read with the offset in force before the gap. An embargo set for 02:30 on the night clocks change is a time someone will read differently, so the safest course is to avoid such hours entirely.

Figures and dates that must agree with something

The numbers in a release travel further than the prose. A reporter may lift “38 percent” into a headline without the sentence around it. That is why every figure should trace to a source the issuer can produce. For a listed company the stakes are formal: a current report on Form 8-K is how companies tell the securities regulator about major events that shareholders should know about, generally within four business days. A release that rounds a number one way while a filing rounds it another way invites questions. At the far end, the regulator’s investor guidance describes schemes in which promoters push a stock price up with false or misleading statements about a company.

Rounding is where honest releases slip. A fact sheet may say 4,213,507 trips, and a release may say 4.2 million. Those agree if the writer rounded to the nearest 100,000, which is what “4.2 million” signals. “More than 900 riders” agrees with 912 only if the writer rounded down on purpose. “38 percent” does not agree with 36.4 percent under any ordinary rounding, though it is the sort of figure that survives three rounds of editing because it looks plausible. A useful check therefore compares each written number with its source under the rounding the writer meant, and names the rule.

Dates fail in a different way. A release that says “Wednesday, November 12, 2026” contains two claims, and only one of them can be right, because 12 November 2026 is a Thursday. Weekday errors come from copying an old release, moving an event, or counting on fingers. Computing the weekday is mechanical. Most methods start from an anchor date whose weekday is known, such as 1 January 1800, a Wednesday, count the days between, and take the remainder after dividing by seven. The count must include every leap day in between, or the answer drifts by one day for each one missed.

Relative words are the quieter hazard. “Today” and “tomorrow” mean one thing on the release date and another on the day an embargo lifts or a reader finds the release a week later. Replacing them with dates costs nothing. The ISO 8601 form, such as 2026-11-03, cannot be misread for day and month order, and it also fixes Monday as the first day of a week, which settles what “this week” covers.

A worked example, and what checking cannot do

Take a release embargoed until 09:00 in Toronto on Tuesday, 3 November 2026. Clocks in Toronto fell back on Sunday, 1 November, so the offset is five hours behind UTC and the embargo lifts at 14:00 UTC. In Berlin it is 15:00, because European clocks changed a week earlier. In Sydney it is already 01:00 on Wednesday, under daylight saving time eleven hours ahead of UTC. A newsroom in Sydney reading “Tuesday” in the release line would be a day out.

Writing the zone name, the offset and the UTC time together removes the doubt. The release’s figures can be checked the same way: 4.2 million against 4,213,507 to the nearest 100,000, which agrees, and 38 percent against 36.4, which does not.

The reminder that travels with an embargo deserves the same care. A calendar file following the iCalendar standard ends each line with a carriage return and line feed, folds any line longer than 75 octets, gives each event a unique identifier and a timestamp, and can state the start in UTC so that every calendar shows the right local hour. The contact address deserves a check too: the internet message format defines an address as a local part, an @ sign and a domain, and a typing error there strands every reporter’s question.

None of this makes a release true. A check can show that every figure matches a fact sheet, but not that the fact sheet is right. It can confirm that a weekday matches its date, not that the event really happens then. It can show an embargo in eight zones, but only with the rules the browser’s time zone data knows, and a government that changes its clocks next month is not yet in that copy. What checking does is narrower and still worth doing: it removes the errors that are purely mechanical, so the attention left over goes to whether the news itself is right.