Hansard Parliamentary Debates

Extract full text UK parliamentary debates from the official Hansard record: one row per contribution, with the speaker, party, constituency, debate title, timestamp and the text itself.

🔍 What does UK Hansard Parliamentary Debates do?

It walks the official Hansard record for a date range and returns every contribution as a clean row of data, ready for a corpus, a RAG index, a database or your own product.

It covers:

• House of Commons debates
• House of Lords debates
• Westminster Hall debates
• Written ministerial statements

Weekends are skipped automatically, and days when the House did not sit return nothing.

📊 What data can I extract from Hansard?

One row per contribution:

Field	Description
date	Sitting date, YYYY-MM-DD
house	Commons or Lords
section	Debate, WestHall or WMS
debateTitle	Title of the debate or question
debateId	Hansard debate identifier
speaker	Member name, null for unattributed text
role	Ministerial role or Chair, where the attribution carries one
constituency	Seat, Commons members only
party	Party abbreviation, for example Lab or Con
memberId	Parliament member id, useful for joining to other datasets
time	Timestamp of the contribution where Hansard records one
text	Full contribution text, markup stripped
wordCount	Word count
uin	Unique identifying number for written questions
contributionId	Hansard contribution identifier
url	Link to the debate on Hansard

💡 Why use Hansard data?

LLM training and fine-tuning. Millions of words of formal political discourse, already attributed to a named speaker and party.

RAG and search. Build a parliamentary knowledge base you can cite, with a source URL on every row.

Policy and legislative research. Track how a bill or a topic was argued over time, and by whom.

Fact checking and media. Verify a quote against the official record rather than a press report of it.

🚀 How do I use UK Hansard Parliamentary Debates?

1. Click Try for free.
2. Pick a chamber: commons, lords or both.
3. Set startDate and endDate to bound the range, YYYY-MM-DD.
4. Add a searchTerm to keep only contributions containing a word or phrase, and set maxSpeeches to cap the run.
5. Click Start, then download the results as JSON, CSV or Excel, or pull them from the API.

⬇️ Input

{
  "chamber": "commons",
  "startDate": "2024-01-01",
  "searchTerm": "housing",
  "maxSpeeches": 5000
}

Field	Type	Default	What it does
chamber	string	commons	Which house to collect, commons, lords or both
startDate	string	2024-01-01	First sitting date to collect, YYYY-MM-DD
endDate	string		Last sitting date, leave empty for today
searchTerm	string		Only contributions containing this term
includeWrittenAnswers	boolean	true	Include written ministerial statements, the WMS section
includeWestminsterHall	boolean	true	Include Westminster Hall debates
maxSpeeches	integer	5000	Hard cap on the number of results

Two more options are worth knowing. maxDays limits the run to N weekday dates, where 0 means every weekday in the range. includeProcedural adds short non-speech entries such as Prayers and the sitting date, and it is off by default because each one is a billable result.

⬆️ Output

Table view

Results arrive as a Contributions table you can sort and filter in the Console, with the date, house, debate title, speaker and word count lined up for scanning.

JSON

A typical row:

{
  "date": "2026-06-03",
  "house": "Commons",
  "section": "Debate",
  "debateTitle": "Supreme Court Dillon Judgment: Policy Implications",
  "debateId": "C66E1F5D-9FA8-409F-9D3B-87F3A026A61A",
  "speaker": "Peter Swallow",
  "role": null,
  "constituency": "Bracknell",
  "party": "Lab",
  "memberId": 5182,
  "time": "11:32:00",
  "text": "What assessment he has made of the potential implications...",
  "wordCount": 26,
  "uin": null,
  "contributionId": "26060368000025",
  "url": "https://hansard.parliament.uk/commons/2026-06-03/debates/C66E1F5D-9FA8-409F-9D3B-87F3A026A61A"
}

Download it from the run as JSON, CSV or Excel, or read it straight from the API.

Dataset size examples

Typical volume per time period:

Period	Contributions	Approximate words
1 day	50 to 200	50,000 to 200,000
1 week	200 to 800	200k to 800k
1 month	1,000 to 3,000	1M to 3M
1 year	12,000 to 35,000	12M to 35M

And for larger corpora:

Range	Contributions	Approximate words
Full 2024	around 25,000	around 25M
Last 5 years	around 125,000	around 125M
Modern era, 2000 onwards	600,000 plus	600M plus

Three runs to size against: 30 days of Commons returns around 5,000 contributions; climate change across both houses since 2020 returns around 2,000; a full calendar year across both houses returns around 30,000.

Recommended update schedule

For an ongoing corpus, run daily at 2am UTC to capture the previous day's debates. Hansard publishes the next working day after a sitting.

For a historical corpus, run once across the full date range and store the output in your own pipeline rather than re-fetching it.

Proxies and reliability

The Actor uses Apify's residential proxy service to avoid rate limiting and keep access to Hansard reliable. Proxy cost is included in the per-result price, so there is nothing to configure.

⏱️ How long does a run take?

Measured on real runs, so you know what normal looks like and can tell it apart from a run that has stalled.

Contributions returned	Typical run time
20	5 to 6 seconds
200	about 20 seconds
500, roughly one sitting day	about 25 seconds
1,000, roughly five sitting days	90 to 100 seconds

What makes a run long is the number of sitting days in the date range, not the number of contributions that come back, because every debate on every day in range has to be opened before it can be filtered. A searchTerm over a whole month is the slow case: a June search for "artificial intelligence" returned 34 matching contributions and took 3 to 6 minutes, because it read the month and kept almost none of it. Narrow the date range first and the search term second. The first few seconds of any run are the container starting rather than the work.

A run is never silently stuck. Each sitting day is logged as it is opened, and contributions are written to the dataset as they are parsed rather than held back to the end, so a run that hits its time limit still leaves everything it had already collected. A range with no sittings in it ends successfully with an empty dataset and the reason in its status message.

Set the run timeout to suit the size of the ask. This Actor's default is 3600 seconds, which is comfortably more than the largest run in the table above (30 sitting days takes about 6 minutes). You are charged per delivered result rather than per minute, so a generous timeout costs you nothing and a tight one risks losing the run's work. Lower it only if you want a hard ceiling on how long a scheduled run may sit.

💰 How much does it cost?

You pay per contribution returned. Set maxSpeeches to cap what a run can cost, which matters most when a wide date range would otherwise produce far more rows than you expected.

Days when the House did not sit return nothing and are not charged. Leaving includeProcedural off keeps short procedural entries such as Prayers out of the bill. Paid Apify plans get a lower rate per contribution.

🔌 Integrations

Send results straight to Google Sheets, Slack, Airtable, Zapier, Make or your own webhook using Apify integrations. You can also trigger a run whenever something happens in another tool, and the Actor is callable from AI agents and MCP clients through the standard Apify API.

🔗 Using UK Hansard Parliamentary Debates with the Apify API

curl -X POST "https://api.apify.com/v2/acts/spookyweb~uk-hansard-parliamentary-debates/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"chamber": "commons", "startDate": "2024-01-01", "searchTerm": "housing", "maxSpeeches": 5000}'

Or with the Apify client:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_TOKEN' });
const run = await client.actor('spookyweb/uk-hansard-parliamentary-debates').call({
  chamber: 'commons',
  startDate: '2024-01-01',
  searchTerm: 'housing',
  maxSpeeches: 5000,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();

Full detail is in the Apify API reference, and every run is also callable from the Python and JavaScript clients.

❓ FAQ

Does it cover the Lords as well as the Commons?

Yes. Set chamber to lords for the upper house only, or both to collect the two together. The house field on every row tells you which one a contribution came from, so a both run is easy to split afterwards.

Does it include written questions and Westminster Hall?

Yes, and both are on by default. includeWrittenAnswers covers written ministerial statements, includeWestminsterHall covers Westminster Hall debates. The section field labels each row as Debate, WestHall or WMS, so you can filter them apart later. Written parliamentary questions and answers live in a separate Parliament API and are not part of the Hansard debate record.

How far back does it go?

The official Hansard record runs back to 1803, and structured coverage through the Hansard API is reliable from 2005 onwards. Set startDate to whatever you need and the Actor walks forward from there, skipping weekends and non-sitting days.

Can I search for a term across debates?

Yes, searchTerm keeps only contributions containing that term. It is the cheapest way to build a topic corpus, because contributions that do not match are never returned and never charged.

What counts as one item?

One contribution, meaning a single speech, intervention or written statement, exactly as Hansard segments it. That is also the billing unit. Procedural entries such as Prayers are contributions too, which is why includeProcedural is off by default.

Do I need a Hansard API key?

No. There is nothing to register for and no quota to manage. Give the Actor a date range and it handles pagination, non-sitting days and rate limiting for you.

⚖️ Is it legal to scrape Hansard?

Hansard is the official report of proceedings in Parliament, published under the Open Parliament Licence, which expressly permits reuse, including commercial reuse, subject to acknowledgement. Contributions are made by members in their public role. This Actor uses the official Hansard API.

Source: hansard.parliament.uk. Apify's ethical scraping guide covers the wider picture.

👍 Your feedback

Found a bug, or want a field that is not here yet? Open an issue on the Actor's Issues tab. Requests that make the data more useful get built, and problems get fixed quickly.

🔎 You might also like

Actor	What it does
UK Case Law Search and Monitor	Court judgments and tribunal decisions from Find Case Law, with full text and change monitoring
UK ICO Enforcement Actions	ICO fines, reprimands, enforcement notices and prosecutions with fine amounts and sectors
UK Planning Applications	Planning applications from every UK council, tagged by the type of work proposed