Every pollster that has polled the UK, rated on how close its final polls came to the result and how openly it publishes; 27 polls there in the last 30 days.
Ratings
Pollsters that have polled in the last year · as of 20 Nov 2026| Pollster | Rating | Accuracy | Lead | Evidence | Openness | What it publishes | Lean now | Polls | Latest |
|---|---|---|---|---|---|---|---|---|---|
| Survation | B 73 | 35 | 14 | 3 elections | 25 | ◐ ✓ ✓ ✓ ✓ | +2.1 | 27 | 12 Nov |
| Ipsos | B 71 | 34 | 12 | 3 elections | 26 | ✓ ✓ ◐ ✓ ✓ | – | 11 | 22 Oct |
| Opinium | B 62 | 34 | 13 | 3 elections | 14 | ✗ ◐ ✓ ✗ ✓ | +2.0 | 65 | 20 Nov |
| YouGov | B 62 | 30 | 12 | 2 elections | 20 | ✓ ✗ ✓ ✗ ✓ | −3.6 | 103 | 16 Nov |
| JL Partners | C 51* | 29 | 12 | 1 election | 10 | ◐ ◐ ◐ ◐ ◐ | – | 16 | 11 Aug |
* Provisional: fewer than 2 checked final polls, so its accuracy is still pulled towards a typical pollster's. Record only: graded on its final polls alone, as we haven't reviewed what it publishes yet. Lean now: +2 means its latest poll has Labour's lead over the Conservatives 2 points bigger than the average does, −2 smaller. A lean is not an error: we only find out who was right on election day.
New pollsters
Polling now, with no checked final poll at a past election yet to grade them on| Pollster | Accuracy | Lead | Evidence | Polls | With an England figure from its tables | Latest |
|---|---|---|---|---|---|---|
| More in Common | 30 | 12 | No checked final poll | 114 | 25 | 15 Nov 2026 |
| Find Out Now | 30 | 12 | No checked final poll | 111 | 102 | 19 Nov 2026 |
| Techne | 30 | 12 | No checked final poll | 43 | 0 | 15 May 2026 |
| BMG Research | 30 | 12 | No checked final poll | 24 | 0 | 29 Oct 2026 |
| Freshwater Strategy | 30 | 12 | No checked final poll | 23 | 16 | 1 Nov 2026 |
| Lord Ashcroft Polls | 30 | 12 | No checked final poll | 21 | 19 | 2 Nov 2026 |
| Deltapoll | 30 | 12 | No checked final poll | 20 | 2 | 16 Nov 2026 |
| Focaldata | 30 | 12 | No checked final poll | 9 | 9 | 19 Jan 2026 |
| Trajectory Partnership | 30 | 12 | No checked final poll | 9 | 0 | 12 Nov 2026 |
| Verian | 30 | 12 | No checked final poll | 4 | 0 | 23 Mar 2026 |
| Convergent Opinion | 30 | 12 | No checked final poll | 3 | 2 | 30 Jul 2026 |
| Good Growth Foundation | 30 | 12 | No checked final poll | 3 | 3 | 7 Jun 2026 |
Most of these have no England figure because we couldn't get their data tables: some aren't published, and some sites block automated access. They get a rating once the review is done.
Past pollsters
A checked final-poll record, but no polls in the last yearParty by party: who each pollster favours
red = too high for that party · blue = too lowAt past elections
Final poll minus the result in England, points, averaged over each pollster's checked final polls (2015–2024). Sorted by how much it has overstated Labour's lead.
| Pollster | Con | Lab | LD | Ref | Grn | Lab lead |
|---|---|---|---|---|---|---|
| YouGov 2 | −4.2 | +2.4 | +1.3 | +0.6 | +0.2 | +6.6 |
| JL Partners 1 | −2.0 | +4.0 | −0.1 | +2.6 | −1.9 | +5.9 |
| Savanta 3 | −2.8 | +1.5 | −0.4 | +2.4 | −0.6 | +4.2 |
| Norstat 1 | −0.8 | +3.2 | −1.4 | +1.8 | −1.2 | +4.0 |
| Survation 3 | −2.8 | +0.6 | −0.0 | +1.3 | +0.7 | +3.5 |
| Opinium 3 | −1.4 | +1.2 | −0.2 | +1.6 | −0.1 | +2.6 |
| Ipsos 3 | −1.4 | −0.5 | −0.1 | +1.0 | +0.7 | +0.9 |
| Qriously 1 | −1.9 | −2.6 | −0.4 | +1.3 | +1.2 | −0.7 |
Now
Left: each pollster's latest England figure minus the average of all pollsters. Right: who its sample over- or under-represents before weighting (its unweighted 2024 Labour and Conservative voters against the real 2024 result).
| Pollster | Con | Lab | LD | Ref | Grn | Sample: Lab | Sample: Con |
|---|---|---|---|---|---|---|---|
| Find Out Now | −2.2 | −0.0 | −1.3 | −0.4 | +1.7 | – | – |
| Survation | −0.6 | +1.4 | +1.1 | +2.4 | −1.4 | – | – |
| Opinium | −1.3 | +0.8 | −0.6 | +0.7 | +0.7 | +2.0 | −1.5 |
| More in Common | +3.2 | +0.5 | −1.0 | −2.6 | −1.5 | +3.2 | +2.4 |
| YouGov | +0.9 | −2.7 | +1.7 | −0.2 | +0.5 | +2.9 | −4.0 |
A pollster that leans one way today isn't necessarily wrong: we only find out on election day. The left-hand table shows who was wrong before, and in which direction. Weighting corrects for too many or too few of a party's voters, but only if the ones a pollster does reach are typical of the rest.
How the rating works
The rating rewards pollsters that get elections right and let people check how their headline was made, measured against the other British pollsters. Out of 100:
Accuracy: up to 50
Its average miss on the Conservatives, Labour, the Lib Dems, Reform (UKIP and the Brexit Party before it) and the Greens in England, at its final poll at each election from 2015 to 2024. Only polls we have checked against the pollster's own tables count.
Its miss is ranked against every checked final poll: the more of them it beats, the more points. A typical final poll (2.06 points) scores about 30; one in the best tenth about 46.
A short record is pulled towards the typical final poll as if it had 2 more elections at that level, so one lucky poll can't earn a top grade.
The lead: up to 20
How far it has missed Labour's lead over the Conservatives, in either direction: the number that decides who governs. A poll can be close on every party and still get the lead badly wrong.
Ranked and pulled towards the typical lead miss (6.2 points) the same way.
Openness: up to 30
What it publishes with its polls, so anyone can see how the headline was made:
- Raw stage (8): Voting intention for everyone asked, don't knows included, before any turnout step. It is what people actually said, and what our own reading starts from.
- Turnout step shown (6): A table after the turnout step, so you can see how much it moved the figures. Turnout steps caused the big 2017 miss.
- Past-vote table (6): Voting intention by 2024 vote, including people who did not vote in 2024. Shows who is switching and whether the sample looks right.
- Likelihood-to-vote table (4): How likely each party's supporters say they are to vote. Needed to judge a turnout step.
- England figure (6): An England column, or English regions with weighted bases, so England can be measured exactly. Polling error is measured in England.
Partly published scores a third. Scored on the items we have checked; a pollster we haven't reviewed yet is graded on its record alone.
Grades
| Grade | Score | Roughly means |
|---|---|---|
| A | 75–100 | Proven accurate and fully open |
| B | 60–74 | A good record, and checkable |
| C | 45–59 | Some evidence of accuracy, or very open |
| D | 30–44 | Little evidence it gets elections right |
| E | 0–29 | Unproven and hard to check |
Things to bear in mind: methods change between elections, so a good 2019 poll doesn't guarantee a good one now. Several of today's biggest pollsters had no standard final poll we could check in 2024, so they are rated as unproven until the next election. We don't use how often a pollster polls, who pays for it, or which way it leans now: none of those tells you whether it will be right.
Figures as of . Published by Polling Forecast.



