Ten Types of Seat

British politics treats constituencies as interchangeable units. A seat is safe or marginal, urban or rural, Labour or Conservative. Polis ran K-means clustering across nine composite scores — engagement gap, reality gap, petition momentum, digital presence, service pressure, media exposure, alignment, accountability, and heat level — for all 650 constituencies. The algorithm found ten distinct types of seat, each with its own political metabolism. Two of those types are moving in opposite directions under identical economic pressure, split almost entirely by education levels.

The ten archetypes are not a taxonomy imposed from above. They are emergent groupings — constituencies that resemble each other more than they resemble anyone else, across nine simultaneous dimensions. What they reveal about the coming realignment is more specific, and more uncomfortable, than any single poll can capture.

One Seat

Poole. A constituency on the Dorset coast held by Labour's Neil Duncan-Jordan with a majority of 18 votes — the slimmest in Britain. It sits in Cluster 8, the "Quiet Understaffed" archetype: high engagement, low media exposure, high accountability. An MP doing solid constituency work that nobody outside the area ever hears about.

Poole is the archetype in miniature. A constituency where the feedback loop between representative and represented appears to function — high engagement, decent accountability — but where that functioning is invisible to the national conversation. It is one of 100 seats in this cluster. 78 of them are projected to change hands. The archetype with the best-working local democracy is the one most likely to be swept away.

The Pattern

The ten clusters separate along two primary axes. Service pressure — the composite of unemployment, food bank usage, crime, insolvency, and NHS waits — runs left to right. Accountability — the alignment between what MPs do, say, and what their constituents want — runs bottom to top. But the real structure is nine-dimensional: clusters that sit near each other on this plot differ sharply on engagement, media exposure, or digital presence.

The algorithm does not know about parties, regions, or demographics. It knows only about nine scores. The political meaning of the clusters is emergent — and that is what makes the patterns credible.

The Numbers

650 constituencies. 10 clusters. The smallest contains 12 seats (the Accountability Void — 7 Sinn Fein MPs, the Speaker, and a handful of others who functionally opt out of Westminster). The largest contains 106 (the Silent Backbenchers — low-profile MPs in low-engagement seats, the single most electorally volatile grouping in Britain). 57 constituencies sit far enough from any cluster centroid that they are genuinely sui generis.

The two clusters that matter most for the next election contain 195 seats between them: the Responsive Performers (95) and the Pressured Activists (76). Both are moving. In opposite directions. For entirely different reasons.

The Ten Archetypes

The Contrast

The two archetypes that tell the story of British politics in 2026 are Responsive Performers and Pressured Activists. Both are overwhelmingly Labour. Both face existential electoral threat. The threat comes from opposite directions.

Responsive Performers are suburban, moderately educated (25% degree-holders), with above-average home ownership (65%). Their MPs have the highest accountability scores in the dataset. They are well-governed. MRP polling projects 73 of the 95 to flip to Reform — a 77% casualty rate. Good governance is no protection.

Pressured Activists are urban, more educated (32% average, but with enormous internal variation), with the lowest home ownership of any cluster (49%). Service pressure is the highest in Britain. These seats are not moving uniformly anywhere. They are fragmenting: 34 projected to Reform, 20 projected to Green. The direction of travel depends almost entirely on one variable.

The Education Split

Within the Pressured Activists cluster — the 76 seats with the highest service pressure — education is the single strongest predictor of political direction.

Constituencies where 40% or more of residents hold degrees — Brighton Pavilion, Bristol Central, Hackney, Camberwell, Islington — are trending Green. Their petition portfolios concentrate on climate, housing justice, and foreign policy. Their engagement is high, ideologically coherent, and leftward.

Constituencies where around 21% hold degrees — Barking, Dagenham, Thurrock, Stoke-on-Trent, Tipton and Wednesbury — are trending Reform. Their petition portfolios concentrate on immigration, crime, and democratic reform. Their engagement is lower but intensifying.

Both sets of constituencies face the same material reality: high unemployment, high crime, stretched NHS services, low home ownership. The pressure is identical. The response is not.

The Invisible 100

Cluster 8 — the Quiet Understaffed — is the single most consequential grouping in the dataset for the next election. 100 seats. 85 held by Labour. Engagement gap of 67.5, meaning residents are politically active online. Accountability of 60.4, meaning MPs are reasonably well-aligned with their constituents. Media exposure of 41.9, meaning the national press has no idea any of this is happening.

These are the seats where the democratic feedback loop between constituent and representative appears to function. Residents sign petitions. MPs respond with parliamentary questions and votes that track local priorities. The accountability score is among the highest in the dataset.

MRP polling projects 78 of the 100 to flip to Reform. That is a 78% casualty rate — the highest of any archetype. The best-functioning local democracies in Britain are the ones most exposed to the populist surge.

The demographics explain why. Average degree-holding: 22%. Average home ownership: 64%. These are the suburban and semi-urban seats where the education split bites hardest. Not affluent enough to feel insulated. Not deprived enough to have developed the institutional resilience of inner-city communities. Precisely the constituencies where dissatisfaction converts most efficiently into votes for something new.

The Stress Test

If the clustering is meaningful, the outliers should be instructive — constituencies that sit far from their assigned cluster centroid and break the archetype's pattern. 57 seats have a centroid distance above 2.5. The five most extreme test the framework.

Chorley (distance 4.30) sits in the Accountability Void because the Speaker barely votes, but its engagement and media scores are off the charts for that cluster. Richmond and Northallerton (distance 4.21) is Rishi Sunak's seat — an Expectation Mismatch constituency whose former PM status inflates media exposure beyond anything the cluster accommodates. Birmingham Ladywood (distance 3.88) generates a service pressure of 81 inside a Misaligned Advocates cluster that averages 34.8. Brighton Pavilion (distance 3.43) is too well-governed for the Media Darlings cluster it was assigned to. And Clacton (distance 2.55) — Farage's seat — sits in the Silent Backbenchers cluster despite being the noisiest constituency in Britain.

Each outlier confirms the same thing: the clustering captures a real structural pattern, and the exceptions are exceptions for identifiable, individual reasons. None of the outliers undermine the archetype; they prove that a nine-dimensional average can be overwhelmed by a single extraordinary feature.

Clacton is classified as a Silent Backbencher. Its engagement gap is 98 and its media exposure is 71. The archetype measures the constituency's structural position. Farage measures something else entirely.

What This Means

Good governance does not save seats. The Responsive Performers and Quiet Understaffed clusters have the highest accountability scores in Britain. Combined, they account for 195 seats. MRP polling projects 151 of them — 77% — to change hands, overwhelmingly to Reform. Whatever is driving the populist insurgency, it is not dissatisfaction with local representation. MPs in these seats are doing their jobs. Their constituents are engaging. The disconnect is between local governance quality and national political mood.

Education is the new class divide. Within the Pressured Activists cluster, constituencies facing identical service pressure diverge entirely based on educational attainment. High-degree seats trend Green. Low-degree seats trend Reform. This is not an income split or an urban-rural split. It is an education split operating within urban seats that share the same material conditions. The implication for Labour is severe: the party cannot hold both ends of this cluster simultaneously with a single message.

The largest cluster is the most volatile. The 106 Silent Backbenchers contain 55 marginal seats with majorities under 5,000. They are the swing bloc of British politics — not because they share a political identity, but because they share a structural one: low visibility, low engagement, and an absence of the institutional infrastructure that makes seats resistant to national swings. Where the ground game is weakest, the wave hits hardest.

The Expectation Mismatch is uniquely stable. The 13 seats containing the PM, Chancellor, and Opposition Leader are the least likely to change hands (36.4% projected turnover vs 55-78% in other clusters). Senior politicians' name recognition and media exposure create an electoral moat that has nothing to do with constituency service. These are personal brands, not party seats.

Cluster fragmentation is the story of Scottish politics. The 69 Settled Suburbs — 25 of them in Scotland — are projected to shatter in all directions: 21 to Reform, 18 to SNP, 8 to Conservative. The highest-accountability, lowest-pressure seats in Britain have no stable partisan home. This is not volatility driven by dissatisfaction. It is volatility driven by the absence of any strong signal to hold these seats in place.

What's Missing

The MRP projections used to assess electoral vulnerability are a single polling snapshot. Seat-level polling has wide confidence intervals, and the projection of 73 Reform gains in Responsive Performer seats could easily be 50 or 90. The directional finding — that these clusters are moving towards Reform — is more robust than the specific numbers.

K-means clustering forces every constituency into exactly one cluster. Real constituencies blend characteristics. Birmingham Ladywood has the service pressure of a Pressured Activist, the media profile of a Media Darling, and the accountability of a Misaligned Advocate. Assigning it to a single archetype loses information. The centroid distance metric partially addresses this by flagging poor fits, but it does not solve the fundamental issue.

The education data is from Census 2021, now five years old. University expansion and demographic shifts may have moved some constituencies across the 40% threshold. The education-to-political-direction relationship is the strongest finding in the data, but its precision depends on the currency of the underlying measure.

Nine scores is a methodological choice. Adding a tenth — say, campaign spending or social media following — would likely redistribute some seats between clusters. The archetypes are stable across reasonable parameterisations (k=8 through k=12 produce similar macro-structures) but the boundaries between adjacent clusters are not hard lines.

Finally, the analysis treats each constituency as independent. In practice, constituencies cluster geographically and influence each other. A Reform surge in one Quiet Understaffed seat in the West Midlands may catalyse neighbouring seats in ways that a purely score-based clustering cannot model.

How We Measured It

This analysis was produced by Polis Intelligence using public data sources. The methodology and data are available for scrutiny. Constituency-level data can be explored on the Polis dashboard.