Methodology

How the site is made

We start with public grant records, bring them into one consistent form, add a small number of clearly labelled interpretations, and check the result before publishing it.

This page explains that process in plain English. The original funder record always remains the authoritative source.

Current site snapshot

Published 16 July 2026. Source collection dates vary.

Not a live feed
269,841searchable records
227,739funded awards
42,102NIHR-supported projects
29research topics
The short version

From a public record to this website

  1. 1Collect

    Take dated copies of public grant records from each source.

  2. 2Bring together

    Put different titles, dates, funders, organisations and amounts into a shared format.

  3. 3Organise

    Group records by topic and connect them to organisations and places where the evidence allows.

  4. 4Explain

    Add clearly labelled plain-English summaries to selected records.

  5. 5Check and publish

    Run automated checks, review important changes and build a new static snapshot.

Where the information comes from

The site brings together four public sources. They cover a large part of the UK research landscape, but they do not include every funder or every award.

How the records are prepared

Each source describes research in a different way. We keep a record of where every entry came from, then place the information into a common structure so that it can be searched and compared.

  • Clear duplicate versions are brought together, while uncertain overlaps are left separate.
  • Organisation names are matched only when there is enough evidence that they refer to the same organisation.
  • Status is worked out from the published start and end dates.
  • Organisations are placed on the map only when a reliable location can be found.

Topics, summaries and AI

Research topics

Every record is placed in one of 29 topics grouped into six broad fields. We compare the meaning of its title and available abstract with reviewed examples for each topic. The closest match becomes the main topic, with a second topic retained where useful.

This makes a very large collection easier to browse, but it is not an official or perfect classification. Interdisciplinary work and records with very little text are especially difficult to place.

Plain-English summaries

Some grants with a useful source abstract have an AI-generated plain-English summary. These summaries are prepared before publication, checked for common problems and clearly labelled. They can still contain mistakes, so readers can also see the original abstract and source link.

The Semantic Map

The Semantic Map places grants with similar language near one another. It is a tool for exploration, not an official classification, an assessment of research quality, or evidence that a particular award caused a later outcome.

Search

Keyword search runs against files stored on this website. It matches the words a visitor enters and supports optional Boolean searches using AND, OR and NOT. Small, fixed spelling rules can include common singular and plural forms; this is not AI-generated rewriting.

Only if a visitor chooses semantic search is their search text sent to OpenAI. That option compares the meaning of the question with the meaning of the grant records. The grant collection itself is prepared in advance.

How the site is checked

Checks happen at several points because a single accuracy score cannot tell us whether the data, topics, summaries, maps and website all work as intended.

Automated checks

Look for broken records, unexpected changes, missing files and totals that no longer agree.

Targeted review

Focuses on important or unusual awards, uncertain organisations, topic changes and risky summaries.

Publication checks

Can stop a new version when key data, location, topic or website checks do not pass.

These checks reduce errors; they do not guarantee that every public record or every interpretation is correct. We keep reviewed corrections so they are not silently lost in the next update.

What to keep in mind

  • This is a snapshot.Source collection dates differ, and funders can change older records after we collect them.
  • Coverage is broad, not complete.A grant missing from this site may sit outside the sources we use.
  • Published information is incomplete.In this snapshot, about 28.4% of funded awards have no published amount and about 15.7% have no abstract. Missing values are not estimated.
  • Funding totals need care.They add positive amounts published by the sources. They are not inflation-adjusted spending figures, and unresolved overlap between sources can remain.
  • Place can be uncertain.A researcher’s affiliation is not always the grant recipient, and a map point does not prove where every part of the research happened.
  • European figures can be partial.For CORDIS projects, the site shows known contributions to UK participants. If some contributions are not published, the displayed amount is a lower bound.
  • Topics and AI-written text can be wrong.They help people explore the data, but the original record remains the authority.

Updates, reuse and corrections

We aim to refresh the main data sources quarterly. A new snapshot can add records and can also change older ones when a source has been updated.

The public data can be downloaded and reused under the licences attached to the source records. The downloads follow the same record types, labels and corrections as the website.

Download the dataReport a problem

Attribution

Contains data from UK Research and Innovation (Gateway to Research), published under theOpen Government Licence v3.0. The site also contains data from NIHR Open Data, Europe PMC GRIST and European Commission CORDIS. Topics, summaries, status, recipient decisions and geography added by this project are not claims made by the source publishers.