Accountability

Corrections

This page lists corrections to figures and facts that were published on this site and turned out to be wrong. It is not a complete list of everything that has ever been fixed: layout problems, broken links and internal faults that never changed a number on screen are not here, because nobody could have relied on them. What is here is anything a reader could have believed and been wrong about.

When something is found to be wrong it is corrected as soon as it is confirmed, and the correction is written down here with the date it was made. Nothing is quietly edited away. Entries already on this page are never rewritten, because a corrections log that gets revised is worth less than no corrections log at all. Most of what follows was found by our own checks rather than reported by anyone, which is the point of running them - and where a check missed something, that is written down too.

The log

8 corrections, newest first

Corrected 8 August 2026

The site understated how current its own figures were, by five days

What the site showed: 'Figures as of 2 August 2026', printed directly above prices from the 7th, and the same date given to search engines.
What was actually the case: Prices were current to the 7th. There are genuinely two dates - one for prices, which update nightly, and one for company figures, which do not - and the page now gives both instead of the older one.
How it came to light: Checking whether the nightly price update had been running. It had, correctly, every night.
How far it reached: Every company page.

Corrected 6 August 2026

Earnings dates for 830 companies were being silently dropped

What the site showed: Company pages were missing the next reporting date for most companies, and the site could not rebuild the data it was publishing.
What was actually the case: The step that assembles each company's record copied across a fixed list of fields and discarded anything else, so earnings dates and fund holdings never survived the build. They now pass through, and a test renders a full record and fails if any field is lost.
How it came to light: Tracing why a full data refresh had long been treated as unsafe to run. The refresh was unsafe, and this was why.
How far it reached: 830 companies for the earnings date, 54 funds for holdings.

Corrected 6 August 2026

The home page search could only find 275 of 933 companies

What the site showed: Typing a company name into the search box returned nothing for most of the companies the site covers, including some of the largest.
What was actually the case: The search built its list from the rows visible in the page's table, and the table was capped at 100 rows, so everything below the cap was unreachable. It now searches the full list.
How it came to light: Testing the search against companies chosen from the far end of the alphabet rather than the top of the table.
How far it reached: 658 of 933 companies were unfindable by search.

Corrected 4 August 2026

Every page claimed the same price date, while about 3% were older

What the site showed: One 'prices as of' date across the whole site.
What was actually the case: The nightly price update fails on a handful of companies each run - 17 on one night, 31 on another - and those keep the previous day's price. Each company now carries its own price date, and a page whose price missed the latest run says so and names the real date.
How it came to light: Reading the nightly update's own log against what the pages were claiming.
How far it reached: Around 3% of company pages on any given day.

Corrected 3 August 2026

A company with no debt figure could read as a company with no debt

What the site showed: On company pages, a total-debt figure of zero was treated as a real, debt-free balance sheet, and produced a line saying the company held net cash and did not need to borrow.
What was actually the case: Zero in that field means the figure was missing from our data source, not that the company owes nothing. The claim is now shown only when both the cash and the debt figures are genuinely present.
How it came to light: An adversarial review of the company-page generator, checking what each claim would say when the underlying figure was absent rather than low.
How far it reached: Affected any company whose debt figure was missing from the feed.

Corrected 3 August 2026

Cash and debt were labelled in the wrong currency on 65 pages

What the site showed: Cash and debt were shown in the currency the share is quoted in.
What was actually the case: Companies report those figures in the currency of their accounts, which is often different. Shell and BP report in US dollars, Unilever in euros, and many London-listed companies with overseas operations do the same. The figures now carry the reporting currency, and no symbol at all where the currency cannot be confirmed.
How it came to light: The same review. The numbers were right; the label on them was not.
How far it reached: 65 company pages.

Corrected 3 August 2026

Profitable dividend payers were described as having reported no profit

What the site showed: Where the payout ratio came through as zero, the page said the company had reported no profit, or a loss.
What was actually the case: A payout ratio of zero for a company that is paying dividends is a gap in the data, not a loss. The line now appears only for a genuinely negative ratio, and the row is left out otherwise.
How it came to light: The same review, checking the claim against companies known to be paying dividends out of profits.
How far it reached: Affected dividend payers whose payout ratio was missing.

Corrected 2 August 2026

The ISA calculator was not comparing like with like

What the site showed: The taxed-account side subtracted every pound of tax, including five years of tax on dividend income, directly from the capital pot.
What was actually the case: Tax on dividends is paid out of the income drawn, not taken out of the capital, so the taxed account was being understated and the gap in the ISA's favour was too large. Both sides are now modelled the same way, with dividends reinvested and capital gains tax applied once at the end.
How it came to light: A check of the calculator's arithmetic against a hand-worked example.
How far it reached: The ISA versus taxed-account comparison.

Found something here that looks wrong? Email [email protected] and tell us - a correction that comes from a reader is recorded here the same way as one we find ourselves, and you do not need to be sure, only to have noticed something. How the figures are put together in the first place is on the methodology page.