Where the information in your report comes from

A confident-looking score means nothing if the data behind it is shaky. A precise-sounding number built on a single unverified record is more dangerous than no number at all, because it invites trust it has not earned. Because trust is the whole product, we open up where our numbers come from: the sources, the cleaning, and how we treat the inevitable gaps.

The sources we draw on

No single feed is complete, so we combine several and let them corroborate each other. The strength of any individual record is in how many independent sources agree on it; a fact confirmed by three feeds is a different thing from one asserted by a single listing.

• Official registration and inspection records.
• Aggregated listing data across millions of vehicles.
• Service and maintenance event histories.
• Reported damage, theft and write-off registers.

Each source has its own blind spots, which is precisely why we lean on the overlap between them. Where one feed is thin, another often fills the gap, and where two disagree, the conflict itself becomes a signal worth surfacing rather than quietly resolving in favour of the cheerier reading.

Cleaning before scoring

Raw data is messy: duplicate records, transposed digits, inconsistent units. Before anything is scored, it passes through normalisation and reconciliation steps that resolve conflicts and discard records that cannot be trusted. A bad input never quietly becomes a confident output, because the damage from a clean-looking wrong number is worse than from an obvious gap.

Garbage in, garbage out is not a slogan to us; it is the failure mode we spend most of our engineering effort preventing.

How we handle missing data

Gaps are unavoidable, and pretending otherwise is the real danger. Where a record is incomplete we widen the confidence band rather than guessing a value, so a thin history reads as uncertain instead of masquerading as solid fact. An honest range beats a confident invention every time.

When records disagree

Sources do not always agree, and how a system handles that says everything about it. Rather than silently picking a winner, we weigh each record by source reliability and corroboration, and where the conflict is genuinely unresolved we let the uncertainty show in the confidence rather than papering over it.

Why fresh data beats more data

A pipeline that hoards stale records is not richer, it is slower to notice when the world changes. Listing prices move, regulations shift, and a record that was accurate two years ago can quietly mislead today. We weight recency alongside corroboration, so a current reading from a reliable source carries more than an old one, and the score keeps pace with the market rather than describing a market that no longer exists.

Keeping the pipeline honest

We back-test the pipeline against vehicles whose outcomes we later learned, then feed the errors back into calibration. The aim is not a pipeline that never errs, but one whose stated confidence matches its real accuracy, so that when we say we are sure, you can be too.

That is why every figure on a report travels with a confidence indicator. The number is only half the information; how much to trust it is the other half, and we refuse to hide it. Access to all of this runs on a simple plan: a 2-day trial for €3.99, then €49.99/month, auto-renewing and cancellable anytime, so the data does the convincing rather than a lock-in.