UK automotive intelligence
Promo(Beta)

Methodology

Last updated 24 August 2026

Every live figure and every article on Promo.co.uk is built on data we can name and a check we can describe. This page sets out exactly where the numbers come from, what they do and don't measure, and the gates a story has to clear before it goes live.

Where the figures come from

Fuel prices: GOV.UK Fuel Finder

UK mean unleaded and diesel prices are drawn from the official GOV.UK Fuel Finder service, accessed through fuel.co.uk (part of the same group), which itself ingests the government feed every 30 minutes. Nothing is estimated: the figures shown are the same official means fuel.co.uk publishes.

That data is published under the Open Government Licence v3.0, which requires it to be acknowledged wherever it is used. Any article built on it carries that acknowledgement, with a link to the source, in the line at the foot of the page. The same applies to the DVSA and DVLA records described below.

Private number plate activity: official DVLA new-reg data and newreg.co.uk

DVLA new-reg mark figures (sold-or-removed counts, initials, prefixes, numbers, area codes and word plates) come from a daily feed published by New Reg, a DVLA registrations dealer, covering the pool of marks available for private purchase. This is a transaction and pool-movement count, not a measure of buyer demand such as searches or enquiries, which we don't currently have data for.

One deliberate honesty note: a mark leaving the available pool means it was sold or withdrawn. We never claim it was assigned to a specific buyer, because the feed cannot tell the difference. We label this “sold or removed”, not “sold”, for exactly that reason. Similarly, a two-letter initials figure (e.g. “SM”) is only ever published when it can be summed correctly across every three-letter combination that reduces to it (SAM, SBM, SCM and so on) using the full underlying event feed, never estimated from a shortlist of top combinations, which would understate the true figure and mislabel it as exact.

Who writes these articles

Promo's articles are drafted by an automated editorial system, not typed by a reporter. It takes a figure or an announcement from one of the named sources above, writes it up, and puts it through the checks described below. Nothing is invented: the system is given the data and is instructed to ground every claim in it.

Most articles are then published automatically, once they clear those checks, without a person reading them first. The exception is anything the system flags as a potentially damaging claim about a named person or company, which is held for a human to decide on and is never published automatically.

A byline names the desk that owns that subject and is accountable for it. It does not mean that person wrote every sentence. If you think a specific article gets something wrong, the fastest route to a correction is to tell us, and the corrections log records what happened.

How an article is verified

Every Promo story must clear a truth gate before publishing: confirmed by an official source, corroborated by several independent sources, or announced first-party by the brand it concerns (a company's statement about itself is reported with attribution, and its claims about the wider market still need independent verification). Official government sources and first-party brand announcements are named and linked in the attribution line at the foot of the article. Commercial outlets that merely corroborate a story are counted toward that verification but never named, since their reporting is not itself an authoritative record.

Beyond sourcing, every draft is automatically checked for internal consistency before it can go live: a figure quoted in the headline or standfirst must also appear in the body, near-duplicate stories are blocked, and where an article names a specific vehicle registration, the claim is checked against DVSA and DVLA records. A separate screen holds back any draft that makes a potentially damaging claim about a named person or company for human review before publication, rather than assessing it automatically.

These same checks run again periodically against everything already live, so a story that later fails one of them is taken down, not left to stand. See our corrections log for what that has actually caught.

How old a story can be

A news story that has sat unpublished for more than fourteen days is archived rather than published, because putting it out under today's date would misstate how fresh it is. Reference material is exempt from that rule: what a dashboard warning light means does not expire, so it is not aged out.

Every figure on the site carries the date it was taken. Where a feed stops updating, the figure keeps its own as-of date rather than being presented as current, and a period label such as “today” is dropped once it is no longer true. A day still running is never compared against completed days.

Limits, honestly stated

A correction or an unpublish is recorded the moment it happens, but it may take a short time to propagate everywhere the story was shared or cached externally. We do not delete a corrected article outright. It is held back for review, and the record of why is public on the corrections log.

The automated drafting described above is the main limitation worth knowing about. The checks catch unsupported figures, duplicates and internal contradictions; they cannot catch a claim that is well-formed, consistent with its source and still wrong. That is what the corrections route is for.

Questions or something looks wrong

If a figure or a claim on Promo.co.uk looks wrong, tell us: email us. We would rather correct it publicly than leave it standing.

See also our corrections log and about page.