South Africa has historically had far too little data in its political analysis. Trying to advise on the politics without poll data is the equivalent of an economic or market analyst trying to advise on growth, debt, the deficit, the currency, stock markets, or the bond market without seeing any of the data. It is for that reason that this newspaper takes a data and fact-based approach to its reporting. So, how did that go for it over the course of the 2026 local government elections?

Start with the African National Congress (ANC) and the Democratic Alliance (DA).

Across the three rounds of polling conducted during 2026, the Social Research Foundation (SRF) tracked the ANC’s national support falling from 39.0% in the first quarter to 34.0% in the second and 32.0% in the third. The party ultimately secured 30.9% of the national proportional representation vote, in line with that falling trend.

The DA was similarly accurately measured. Its support moved from 28.0% in the first quarter to 27.0% in the second and 26.0% in the third, against a final result of 24.7%; again in line with the trend.

The uMkhonto weSizwe Party (MKP) was polled at 12.0% in the final round against an election result of 9.8%, while the Economic Freedom Fighters (EFF) was measured at 8.0% against a final result of 6.2%.

That is all set out in the table below, together with some of the data for the smaller parties.

Remember, these were not forecasts, as the SRF produced for the 2024 national election. These were measurements taken of the parties at the time they were measured. As is true for the currency and the stock markets when similar measurements are taken, the figures are at best indicative of longer-term trends if those trends hold, which they did in the case of this election. For the 2024 national election, the SRF ran a 30-day national tracking poll which polled every day for the month ahead of the vote, similar again to tracking the currency or the stock market daily, this allowed it to forecast the election day result.

The metro polling was also broadly accurate.

In Johannesburg, the SRF’s final poll placed the DA at 36.0% against its eventual 34.8%, and the ANC at 22.0% against 20.3%. The EFF was measured at 6.0% against 6.5%. The larger differences were for ActionSA, which secured 16.3% against a poll reading of 9.0%, and the MKP, which secured 5.6% against 11.0%.

In Tshwane, the SRF’s polling tracked a significant shift in political support over the course of the year. The ANC fell from 42.0% in the first quarter to 32.0% in the second and 26.0% in the third, before securing 20.2% in the election. The DA fell from 45.0% to 43.0% and then 32.0%, before recovering to 34.4%. The most important development in Tshwane was the rise of ActionSA. The party’s support increased from 5.0% in the first quarter to 7.0% in the second and 12.0% in the third, before reaching 22.1% in the election. That strong surge for ActionSA saw the party draw significant late-stage support away from the MKP, the EFF, and the ANC, which was also the case in Johannesburg.

In eThekwini, the SRF correctly identified the MKP as the largest party throughout the year. Its support was always volatile and measured at 44.0% in the first quarter, 33.0% in the second, and 43.0% in the third, against a final result of 38.3%. The DA was measured at 28.0% in the final poll against 25.5%, while the ANC secured 16.4% against a polling estimate of 13.0%. The Inkatha Freedom Party was measured almost exactly, polling at 7.0% against a final result of 6.8%.

Trying to advise on politics without poll data, and supporting economic and living standards data, is the equivalent of an economic or market analyst trying to advise on growth, debt, the deficit, the currency, stock markets, or the bond market without seeing any of that data. Then it just becomes a grift, which is true for nine out of ten political analysts and writers who mostly just make it up as they go along. However, as with economic and market data, political data moves from day to day and week to week, often sharply, and a figure taken at a moment in time is no more of a forecast of where it will be a month down the line than would be the case for the rand or the S&P 500, but where the trends revealed in the data hold, they may be usefully indicative of where a number is headed.