The problem
Most finance sites don't say how a claim got checked
A figure on a finance site is either right or it isn't, and the reader usually has no way to tell which. Most sites don't say where a number came from, whether it's current, or who checked it before it went live. This page isn't a general commitment to "quality". It's a specific account of the checks a claim has to pass.
Why it happens
A confident sentence and a checked sentence read identically
Nothing about how a claim is phrased tells you whether it was verified. "Rates are competitive" and "the base rate is 3.75%, held at the 30 July 2026 decision" can sit in the same paragraph with the same tone. One is empty; one is checkable. The only way to tell the difference is to show the source, which is why every specific figure on this site carries a link to where it actually came from, not just a claim that it's accurate.
Sourcing
What "sourced" actually means here
A primary source: gov.uk, HMRC, Companies House, the Bank of England, the FCA, or an equivalent regulator or official body, cited with a real, working link. A named industry body's own published research (UK Finance, the British Business Bank, The Data City) counts, cited the same way. A secondary source restating someone else's number (a marketing blog, a competitor's page, an aggregator) does not count on its own, and where a figure could only be traced that far, it either gets verified against the real primary source first or it doesn't get published as a hard fact. See Data Sources for the working list this site actually cites from.
What doesn't get published as fact: a specific number, date, or claim that can't be traced to a real source gets rewritten as a general, honestly-hedged statement instead, or left out. This has happened while the site was being written: drafts included figures that turned out to be unverifiable, and they were cut or rewritten before publishing rather than corrected after a reader flagged them.
Checks
The checks a page runs before it's allowed to publish
- Readability. Every page is checked for text that is genuinely readable against its background, on a phone as well as a desktop.
- A direct answer. The core answer on each page has to stand on its own as a clear statement, not a claim that only makes sense buried in context.
- Substance. A page is checked for real depth, not just polished prose sitting over nothing underneath.
- A sceptical read. Every page is deliberately read looking for fabricated figures, hollow padding, and unsupported claims, treating the page as guilty until the sourcing checks out, not the other way round.
- A named person. After the mechanical checks, a real person reads the page and asks whether it's actually true and actually useful, not just technically compliant. That's not a rubber stamp: pages have been rewritten after this step, not just approved.
If something looks wrong
What backs each kind of claim
Scope
Every page, not just the data-heavy ones
This process applies to every page on the site, not just the ones that look like they carry heavy factual weight. A short facility page and a data-heavy reference page go through the same checks. The difference is only in how much there is to verify.
Limits
What this process can't promise
This page describes the process, not a guarantee against every possible error. Rules change, sources get superseded, and a check run today can go stale. What it does commit to: every figure carries a link to the source it came from, and a genuine correction process exists rather than a static page nobody revisits. See Data Sources for the sources themselves, and tell us directly if something looks out of date.
