Projections vs Results

About the baseline (and why our numbers differ from the BBC's)

The "Defending" column counts the seats each party held going into the 7 May 2026 elections according to our source-of-truth data (Open Council Data + our by-election and defection tracker). It includes every by-election result and every defection we have on record, but is not perfect - small councillor moves do go unrecorded, especially recent ones. Treat the defending numbers as accurate to within a handful of seats per party. Reform's 71 defended seats reflect by-election wins and known Conservative-to-Reform defections; their 2022 starting point would have been close to zero.

The Press Association (used by the Guardian and most papers) uses a similar baseline to ours - counting changes from who held the seat just before the election, factoring in by-elections and defections. Our seat-change numbers should be in roughly the same range as theirs.

The BBC uses a different baseline: it compares each result to the last time those exact seats were fought (typically 2022 or 2024). Because Reform/UKIP/Brexit barely existed in most of those contests, the BBC baseline starts at near-zero and every Reform win is logged as a gain. John Curtice's PNS team note the same issue: "whatever share of the vote Reform secures this time around will represent an improvement on zero." That makes for a bigger Reform "+" and a bigger Conservative "−" in BBC totals than in ours or the PA/Guardian's.

Other reasons our seat changes can differ from any of the broadcasters or papers:

• Boundary changes. Several councils were redrawn between elections (Sunderland, Bradford, Newcastle, South Tyneside, Birmingham went all-out under new ward maps; Essex/Suffolk had divisions added or merged; Surrey was abolished and split into two new unitaries). To work out how many seats each party was "defending", we recalculate what 2022's vote shares would have produced under the 2026 ward boundaries. The BBC does the same. The Press Association (Guardian) doesn't — they just compare raw seat totals before and after, which can make a real swing look like "no change" when the council shrinks.
• Independents and minor parties. "Others" combines independents, residents' associations, Workers Party, NIP, Aspire and the rest. Outlets group these differently and may break some out separately.

For ward-level methodology, see the 2026 locals methodology and the model changelog.