Tesseract Foundational Research: Public Engagement
Every government consultation that reached an outcome, as one corpus
Government runs thousands of public consultations and publishes their outcomes, but the estate has never been assembled as a single, queryable corpus: who consults, on what, and whether they publish what they heard back. This builds that corpus from open data, and finds that closing the loop is less universal than the outcome pages suggest.
The direction of travel
Consultation is a legal and democratic obligation, not a courtesy. The Cabinet Office Consultation Principles and the common-law Gunning principles require that consultation responses are conscientiously taken into account, and public bodies increasingly want to analyse large volumes of responses quickly and defensibly, with AI-assisted coding now on the table. Every one of those ambitions needs a structured view of the consultation estate to sample, benchmark and evaluate against. That view did not exist as open data.
The method: harvest, then use government's own coding
We harvested every GOV.UK consultation that has reached a published outcome, then pulled structured detail for each from the Content API: opening and closing dates, department, whether response documents are attached, and GOV.UK's own policy-area taxonomy. That last point matters: 98% of consultations are already coded by policy area by government itself, so the domain layer needs no guessing. The corpus is the map; the response texts stay on the linked GOV.UK pages.
Most consulted-on policy areas across the corpus.
Closing the loop, or not
Every consultation here has a published outcome page, yet only 77% attach response documents to it. Nearly a quarter close with an outcome that carries no published response analysis at all. Defra, the Department for Transport and MHCLG are the most prolific consulting bodies; the busiest subjects are the UK economy, access to the countryside, and low-carbon energy. Whether a consultation publishes its response is now a measurable property, not an impression.
Where this goes
As a corpus frame it is the sampling and benchmarking base for consultation-response coding and automated summarisation: pick a policy area, draw a stratified sample, and test a coding approach against a known population rather than an ad-hoc handful. It sits alongside our applied public-engagement work and our nature-governance graph, both built on the same conviction that the machinery of government should be legible as open, structured data.
Independent, self-initiated open research. Contains public sector information licensed under the Open Government Licence v3.0.
Explore the corpus
6,260 consultations as CSV with policy-area coding, plus the reproducible harvester.
