How a Tradello record is built

The distinction everything else on this page follows from

A Tradello record contains two different kinds of statement side by side, and they are not equally reliable.

• What the publisher published. The contracting authority or the council said it, in a notice or an application. If we are wrong about it, we have misread a document and the linked source settles the matter.
• What Tradello worked out. The trade breakdown, an estimated value or quantity, the role an organisation is playing. Nobody published these. We produced them.

The breakdown of a project into trade packages, and any estimated value or quantity attached to one, are Tradello’s work. The contracting authority did not publish them, has not seen them, has not agreed them and is not answerable for them. Where the authority published a figure, the record says the authority published it and links to where.

This is not buried in a footnote because it is not a footnote. Every graded figure in the product carries its grade where the figure is, so you can tell the two apart at the moment you are looking at them rather than by coming here.

The three grades, and who is answerable for each

Published fact (SOURCE_FACT)

Who says so: The body that published the notice. The publisher stated this. Tradello copied it across and recorded which notice it came from and when. If what you see here disagrees with the published notice, the published notice is right — which is why every package carries a link to it. For example: The deadline, the contract value the authority published, the authority’s own name, the name of an organisation the award notice names as main contractor.

Derived by Tradello (DERIVED_FACT)

Who says so: Tradello, from something the publisher did state. Not published as such, but following closely from something that was. The derivation is recorded with the thing it was derived from, so you can see the step. It is more reliable than an inference and it is still ours rather than the authority’s. For example: Days remaining, calculated from a published deadline. A role concluded from where an organisation sat in a published document rather than from the document naming the role.

Tradello’s inference (INFERENCE)

Who says so: Tradello, and nobody else. Our reading of the evidence, labelled as a reading, with the basis named. The publisher did not say it. It is worth checking against the source before you rely on it, and the product says so in those words rather than presenting it flat. For example: That a scheme probably involves a particular trade, where the documents describe the work without naming the package.

Where a field is absent from the source, the record says the source did not publish it. It is never filled in with a plausible value, and an absence is never quietly rendered as a zero.

Counts and estimates are two different claims

Counts are counted

How many projects, how many organisations are on one, how many contact routes are in a package — these are counted from published records. Nothing in that category is estimated, sampled or modelled.

Anything read out of a document is approximate, and says so

Quantities, areas and scopes read out of drawings and statements a council published are estimates produced by reading those documents. They carry this caveat wherever they appear, in these words:

Approximate and estimated, read from the documents the council published. Figures are not a measured take-off and no site has been inspected.

That sentence is held as a single value in the code and rendered from it, so it reads identically on the first screen and the fourth and cannot be softened on one of them.

Trade packages are our reading of a project, not the authority’s

A public body publishes a scheme. It does not usually publish a list of the trade packages inside it, which is the thing a roofing firm or a cladding contractor actually wants. Working out what those packages probably are is what we do, and it is the reason the product exists.

It also means the trade breakdown is ours. The authority did not publish it, has not seen it, has not agreed it and is not answerable for it. Where the documents name a package, the record says so and grades it as a published fact. Where they describe work without naming the package, the record says that this is our reading and names what it was read from.

The same is true of an estimated value attached to a package. It is an estimate produced by reading documents. It is not a measured take-off, no site has been inspected, and it is not a price.

Every package links to the published record

Each package carries a link to the notice or application it was built from, so you can check anything you intend to rely on against the original. If a fact in Tradello and a fact in the published notice disagree, the published notice is right.

The published record is the publisher’s, not ours. We captured it, matched it and broke it into packages; we did not publish it and we do not own it. The licences it is used under are at Tritium 3.1 and licences, and what that does and does not permit is at documents, copyright and database right.

A source page may itself print a case officer’s name. That is the publisher’s page, and using anything you find on it is your own processing, on your own basis.

Records change, because published records change

A notice can be amended, withdrawn or superseded, and a planning application moves through stages. We re-read our sources, so a record here can change after you have bought a package built from it. That is the product tracking a moving public record rather than a defect in it.

The rules that decide what may be shown are applied at the moment a record is read rather than when it was captured, so an improvement to those rules applies to everything already held the next time anybody looks at it, rather than only to what arrives afterwards.