PollCheck

PollCheck is a UK polling aggregator that publishes rolling moving averages of voting intention across UK general elections, Scottish Parliament, and Senedd elections. It aggregates the major pollsters — YouGov, Opinium, Savanta, Survation, Ipsos, More in Common — into a smoothed series less noisy than individual polls. Its rolling averages are widely cited in political coverage as the most defensible single source for vote-share claims in multi-party election cycles.

In the 2026 election cycle, PollCheck's most prominent output was a projection — produced jointly with YouGov's MRP — that Reform UK would win 2,342 England council seats. The actual result on 7 May was 1,448 — an 894-seat, 38% undershoot, the worst projection failure in modern UK polling. PollCheck's pre-election English council projections had also pointed to Reform winning control of Essex, Norfolk, and Suffolk county councils outright. Essex and Suffolk fell to Reform; Norfolk produced a hung council where Reform won 40 of 84 seats, three short of a majority. PollCheck's 1 April 2026 five-poll Senedd average placed Plaid Cymru at 28.4% and Reform at 27.6% — a 0.8-point gap accurately characterising the closeness of the pre-election race.

The 7 May England results put the PollCheck/YouGov MRP methodology under scrutiny for the first time after a run of accurate UK general election forecasts. The failure reflects a structural challenge common to all polling aggregators: rolling averages are robust against noise but cannot correct for systematic model assumptions — in this case, uniform swing within demographic cells — that diverge from actual vote concentration patterns.