South Africa’s municipal election will redistribute political power. The harder question is whether the next councils can turn that power into reliable services, credible budgets and repairs that last. Johannesburg’s next governing party could take office with far less than half the vote. The eNCA MarkData survey puts the Democratic Alliance (DA) on 35%, ahead of the African National Congress (ANC) on 21.8% . That is a substantial lead. It is also a warning against treating an election-night victory speech as a mandate to govern alone. The distinction matters beyond party arithmetic. Residents need a government capable of approving a budget, keeping qualified people in critical jobs and repairing infrastructure. A party can win the argument against its predecessor without having secured the council support or administrative capacity to do those things. South Africans vote on 4 November 2026, with special voting on 2 and 3 November . The practical question is what happens after the votes have been counted: who can assemble an administration that works, and how will residents judge it? A lead is not control The polling describes markedly different contests. In Cape Town, the DA has 52.6% in the published survey . In eThekwini, uMkhonto weSizwe (MK) is on 48%; in Tshwane, the DA is on 33.9% and the ANC on 24.3% . These are estimates of voting intention, not declared results or allocated council seats. MarkData’s methodology records 6,113 interviews across six metros, conducted from 10 August to 14 September. It models turnout at 48.8% and reports metro margins of error of approximately three percentage points at a 95% confidence level. The interviews therefore predate the closing weeks of campaigning. Turnout assumptions, sampling error and subsequent changes in preference all limit what the numbers can establish. The survey also does not cover Buffalo City or Mangaung. A six-metro poll cannot settle the electoral position in all eight metros, still less the country’s entire municipal landscape. Under the Municipal Structures Act, councillors elect the executive mayor from among their members . The party finishing first is therefore not automatically entitled to the mayoralty. Where it lacks sufficient council support, the negotiations become part of the governing contest. That makes coalition terms a legitimate concern before election day. Voters should be able to ask prospective partners which budget priorities they share, how they would resolve disputes and what would make them withdraw support. A coalition agreement that names office-holders but cannot explain the maintenance programme offers residents little basis for judging its usefulness. The evidence must outlast the campaign A service-delivery election should reward a demonstrated record. But a credible comparison must distinguish popularity, financial controls and the experience of residents. They answer different questions. The Auditor-General’s 2024/25 local-government report records that none of the eight metros achieved a clean audit. Three received unqualified opinions with findings, and five received qualified opinions. Cape Town’s result was unqualified with findings, a qualification to any suggestion that its governance record is beyond scrutiny. The report also records formal disputes over findings by Cape Town, Ekurhuleni, Johannesburg and Buffalo City. These categories should not be flattened into a verdict that every city is equally badly administered. An unqualified financial opinion and a qualified opinion are different outcomes. Nor is an audit a complete service-delivery score. The Auditor-General explicitly distinguishes its audit work from auditing metros’ performance , while using financial statements, performance reports and infrastructure work to illuminate service risks. For voters, the useful demand is for evidence that connects money to results: funded repair programmes, credible completion dates, fewer repeated faults and reports that explain missed targets. Those standards should apply to a familiar incumbent and an appealing challenger alike. What the water figures actually mean Johannesburg illustrates why definitions matter. In a 4 September parliamentary reply , the water minister reported unaudited non-revenue water of 47.1% for 2025/26. That category includes physical losses, commercial losses and authorised consumption that is not billed. It does not mean that 47.1% leaked from pipes. The distinction changes the remedy. Physical losses require infrastructure attention; the other categories also require scrutiny of measurement, billing and authorised use. A campaign promise to reduce “losses” should specify which measure will improve and how progress will be checked. Competence includes deciding who benefits There is a powerful appeal in treating municipal government as a practical business: collect refuse, maintain roads, provide water and electricity. Yet deciding which neighbourhood receives investment first, who q