Methodology

What counts as an entry

An entry is a specific, public statement by Donald Trump that something would happen, be delivered, be announced, or be decided, which then did not happen as described. Vague aspirations ("we're going to win so much") are excluded. Policy positions he simply changed his mind about are excluded unless he had promised a concrete deliverable.

Each entry records:

The promise, quoted. Where a verbatim quote was not available in the reporting we cite, the text is marked [paraphrase].
When and where it was said.
The stated deadline, if any, exactly as spoken ("within two weeks", "day one", "24 hours"), converted to a number of days.
Status: never delivered, broken (he did the opposite or explicitly abandoned it), partial, late, or still waiting (deadline has passed and he still says it is coming).
What happened, with dates and figures.
Sources: links to the reporting. We prefer primary transcripts and major fact-checking organizations (PolitiFact, FactCheck.org, the Washington Post Fact Checker, KFF Health News) and mainstream outlets.

The two-week file

The spotlight is reserved for promises with a stated deadline of 14 days or less. "Soon" and "shortly" do not qualify. "Two weeks", "a couple of weeks", "next week", "in a few days", "24 hours", "day one" and "immediately" do. Where the same promise was re-issued with a fresh deadline, we keep one entry and record the number of times the clock was restarted in repeat_count.

"Overdue" is computed from the date of the promise plus the stated deadline, through the evidence date shown in the footer. It is not updated live; the site is rebuilt as the ledger changes.

"Only I can fix it" and "Better than anyone"

Two companion files sit beside the promise ledger. They are not promises with deadlines, so they carry no status and no overdue clock.

Only I can fix it collects statements that only he could fix, save, stop, solve or prevent something ("I alone can fix it", "only I can stop World War III", "nobody else could have done it"). Each entry records the subject, the quote, where it was said, and the state of that thing afterward.
Better than anyone collects claims of unique expertise ("I know more about ISIS than the generals", "nobody knows more about taxes than me"). Each entry records the subject, the quote, where it was said, and what the public record showed on that subject.

China

A third companion file, China, in his words, collects statements in which he cast China as an enemy, an adversary, a cheat or a threat. Each entry is tagged by the kind of threat he described:

Economic: trade deficits, currency, tariffs, jobs, factories, intellectual property.
Military: military buildup, the South China Sea, Taiwan, nuclear weapons, espionage and cyberattacks.
Economic and military: a single statement that invokes both. These count toward both totals.
Other: COVID-19, fentanyl, elections and everything else.

The tag describes what he said, not a judgment about China. Each entry then records what the public record showed afterward on that same subject, including his own later statements and actions.

Russia and Putin

Russia and Putin collects occasions on which he conceded to, sided with or deferred to Vladimir Putin or the Russian government. An occasion qualifies if the public record shows at least one of the following:

Ukraine: he adopted Russia's position or granted a Russian demand on military aid, territory, NATO membership or the terms of a peace deal.
Sanctions and pressure: he lifted, delayed or declined to impose sanctions, or let his own stated deadline pass without the stated consequence.
Intelligence and security: he publicly sided with Putin over U.S. intelligence agencies or officials, or shared classified information with Russian officials.
NATO and allies: he acted to Russia's benefit on NATO, the G7, troop deployments or arms control.
Praise and deference: he praised or deferred to Putin personally in a way that was news at the time.

Each entry is anchored by his own words. Where he later reversed course or acted against Russia on the same subject, the entry says so, with the date. The tags describe what happened. They are not a claim about motive.

The family business

The family business collects transactions in which Donald Trump, a member of his family, or a business they own was paid by, invested in by, or granted something of value by an authoritarian government, its ruler, or its state-owned companies.

Which governments. Only countries rated "Not Free" in Freedom House's Freedom in the World report for the year of the transaction. Each entry shows the rating and the report year. Countries rated "Free" or "Partly Free" are left out, whatever is said about their leaders.
Who counts as family. Donald Trump, the Trump Organization, his wife and children, Jared Kushner, and ventures they own or are paid by.
Private buyers. A private businessperson counts only where reporting or an official record documents that they acted with or for that government or its ruler.
What counts as a benefit. Investments, management fees, loans, licensing and branding deals, property sales, gifts, payments to Trump properties, trademarks and approvals granted, and purchases of family-owned crypto tokens.
Policy context. Where the record shows what the U.S. government did regarding that country around the same time, the entry states it, with dates. Placing two facts side by side is not a claim that one caused the other. Where a source makes that claim, it is attributed.
Responses. Denials and responses from the Trump Organization, the White House or the family member are included where they are on the record, as is the outcome of any investigation.

Amounts are given as reported and are not added up, because they measure different things: a fund's size, a sale price, a project's projected value.

One entry per distinct subject and occasion. Where the same claim was repeated about the same subject, we keep one entry and note the repeats. The same sourcing rules apply: every entry links to reporting that documents the quote, and paraphrases are marked.

What he called them

What he called them collects remarks in which Donald Trump insulted or demeaned one of four groups:

Women: remarks about a woman's looks, body, intelligence or worth, and insults aimed at women who challenged or criticized him.
Immigrants: remarks describing immigrants, refugees or asylum seekers, or the communities they come from, as criminals, animals, diseased, an invasion, or a threat to the nation's "blood".
Other countries: remarks disparaging a nation, its people or its leader as a stand-in for the country.
American cities: remarks describing U.S. cities and their residents as infested, filthy, war zones or worse than foreign conflict zones.

Each entry is anchored by his own words, from a transcript, recording, archived post or reporting that quotes him directly. Paraphrases, such as remarks reported from private meetings, are marked and attributed. Acts are included only where they are bound to the words: a policy announced in the same remarks, or a court or jury finding. Allegations without such a finding are left out.

Our own text does not characterize the remarks. Where a public figure, organization or outlet called a remark racist, sexist or otherwise, the entry attributes it. Denials, apologies and repeats are recorded with dates. One entry per subject and occasion; when he repeated a line, we keep the first or best-documented occasion and note the repeats.

What this is not

This is not a scorecard of promises kept, and it does not claim to be balanced against them. Organizations such as PolitiFact's MAGA-Meter and its earlier Trump-O-Meter track the full picture, including promises kept. This site documents one thing: the gap between what was announced and what arrived.

Corrections and contributions

The ledger is a set of plain JSON files, published at /data/. If an entry is wrong, the quote is misattributed, or a source link is dead, the fix is a one-line edit. Missing entries need a date, a quote, a venue, an outcome and at least one source link.

Seed sources

This project started from these published lists, then expanded them with further reporting: