A landslide built on a thin vote, a coalition pulling in opposite directions, and a run of named, datable decisions that tracked his ratings down. As reports swirl that the prime minister may set out his departure — unconfirmed by Downing Street — this is the anatomy of how his support drained away.10
Labour’s 172-seat majority in July 2024 disguised a fragile base. The party won it on roughly a third of the vote — about 34% — the lowest share of any majority government in modern record-keeping, on low turnout and lower enthusiasm.1 Much of the win was a verdict on fourteen years of Conservative rule rather than a positive embrace of Labour. Support that shallow is easy to lose, and it began draining at once: YouGov recorded a brief honeymoon — 44% favourable just after the election — before the rating turned net-negative within weeks.5
By mid-June 2026 Starmer’s net favourability sat near −45, having bottomed at roughly −57 in January — territory otherwise reached only by Liz Truss.5 The decline was not random noise. It tracked a sequence of concrete decisions.
The common story — that Reform UK is devouring Labour’s vote — is only half right, and arguably the smaller half. When YouGov traced where 2024 Labour voters went at the May 2026 local elections, just 46% stayed loyal; of those who left, far more moved left than right: 22% to the Greens and 16% to the Liberal Democrats, against 6% to Reform and 5% to the Conservatives.6
Nationwide estimates show the same direction, less steeply: Persuasion UK put Labour-to-Reform switching in England at about 5%, against 22% to the Greens and 10% to the Lib Dems, with nearly a quarter intending to stay home.8 Sir John Curtice has made the mechanical point: Labour’s vote falls further where the Greens are strong than where Reform is, and a leak to the Greens can hand Reform a seat on a split.9
Defectors’ own reasons make the trap explicit. In a YouGov study of more than 10,000 Britons, those who left for Reform cited immigration being too high (62%) and broken promises (44%); those who left for the Greens cited the opposite — that Labour had governed too far to the right (48%) — plus a values mismatch (36%) and the government’s stance on Gaza (25%).7
A thin mandate and a divided coalition set the conditions; specific, named decisions did the damage — and their timing shows up in the polling. The trajectory below maps YouGov’s net-favourability readings against the moments that drove them.
The fiscal squeeze. The first wave was money. Rachel Reeves’s 30 October 2024 Budget raised taxes by around £40bn to fill what the government called a £22bn fiscal hole: employer national insurance rose to 15% with the threshold cut from £9,100 to £5,000 (about £25bn a year), inheritance tax was extended to farms and family businesses worth over £1m, and 20% VAT was applied to private-school fees from January 2025.3 The measures spared headline taxes on “working people” but landed on employers, farmers — who drove tractors through Westminster — and the squeezed middle.
The winter fuel cut. The most toxic was earlier and smaller. On 29 July 2024 the government means-tested the winter fuel payment to pensioners on Pension Credit, cutting recipients in England and Wales from about 10.8 million to 1.5 million to save roughly £1.3–1.5bn.2 Starmer’s net rating slid from around −9 in mid-summer to −21 by September as it bit. After a sustained backlash the policy was reversed on 9 June 2025 — restored to all pensioners, clawed back above £35,000 of income — but the damage was done.2
The welfare revolt. On the other flank, the March 2025 Pathways to Work green paper sought about £5bn of savings by 2030 by tightening Personal Independence Payment eligibility (the “four-point” rule) and cutting the universal credit health element. The resulting Universal Credit and Personal Independence Payment Bill triggered the largest rebellion of the Parliament: 126 Labour MPs signed a reasoned amendment, and ministers were forced to strip the PIP cuts out two hours before the vote, passing a gutted bill on 9 July 2025.4
The smaller wounds. Others accumulated. The government kept the two-child benefit cap and suspended seven of its own MPs for voting to scrap it (July 2024).10 In December 2024 it rejected compensation for the “WASPI” women — about 3.5 million born in the 1950s — despite an ombudsman recommending up to £2,950 each, after Starmer had backed their campaign in opposition.10 From September 2024 the SDS40 scheme cut automatic prison release from 50% to 40% of sentence to ease overcrowding, fuelling “soft justice” attacks.10 And one month in mid-2025 brought three reversals — winter fuel (9 June), a national grooming-gangs inquiry Starmer had resisted (mid-June), and the welfare climbdown (late June).410
Scandal, and the slide to the trough. Then came the appointment, and September 2025 dismissal, of Peter Mandelson — a friend of the late convicted sex offender Jeffrey Epstein — as ambassador to Washington, with questions over what Starmer knew. He never met Epstein and is not accused of wrongdoing, but it sharpened a sense of poor judgement.10 Through this run the line kept falling to its −57 January 2026 low, recovering only slightly thereafter.5
Underneath the specifics sat the economy — weak growth and a grinding cost-of-living squeeze, with the chancellor’s ratings falling below her Conservative predecessor’s.5 The cumulative effect was a collapse in trust: by late 2025 around two-thirds of Britons saw Labour as out of touch, weak and untrustworthy, and some 77% did not trust it to keep its promises or ease the cost of living.9 The reach was geographic too — in 2026 Welsh Labour lost power for the first time in roughly a century.9
The trajectory shows correlation, not proof of cause: ratings move on many forces at once, and the timing of a policy and a dip cannot by itself establish which drove which. Polling also cannot cleanly isolate which grievance mattered most; the drivers overlap. Local-election defection figures over-represent urban, Green-leaning areas, so the steep Green-over-Reform ratio there overstates the nationwide picture.68 A large share of lost Labour voters did not switch at all but stayed home.8 And these are snapshots: YouGov’s online figures carry no classical margin of error, and the reported Monday resignation is unconfirmed by Downing Street.10
That Starmer is unpopular is not in dispute; why, and whether anyone could have done better, is — partly an empirical question, partly a judgement about counterfactuals.
The mandate was shallow, the coalition irreconcilable, and incumbents almost everywhere are being punished amid a long cost-of-living squeeze. A centre-left party trying to hold both Green-leaning and Reform-leaning voters was always likely to fracture, whoever led it.89
To keep this focused on causes, three strands were left for separate treatment.