In the results below, performing well combines those who said very well or quite well. Performing poorly combines those who said quite poorly or very poorly. The figures reflect likely voters in each metro.

In Tshwane, 62% of likely voters said the ANC was performing well, while 37% said it was performing poorly. For the DA, 84% said it was performing well, while 15% said it was performing poorly.

In Johannesburg, 47% said the ANC was performing well, compared with 39% who said it was performing poorly. For the DA, 70% said it was performing well, while 17% said it was performing poorly.

In eThekwini, the ANC was rated more negatively than positively. 35% said it was performing well, while 63% said it was performing poorly. For the DA, 62% said it was performing well, compared with 20% who said it was performing poorly.

The finding suggests that the DA has built a stronger positive image in the GNU than the ANC across all three of South Africa’s biggest metros. The DA is rated positively in every city, and even in eThekwini — its weakest of the three — a clear majority still say it is performing well. The ANC’s standing, by contrast, swings sharply by metro: solid in Tshwane, split in Johannesburg, and clearly negative in eThekwini, where nearly two-thirds of likely voters say it is performing poorly.

These results were gathered in February 2025, meaning they do not yet reflect political developments in the months since.

*Figures are drawn from the Q1 2025 Market Survey produced by The Common Sense in partnership with the Social Research Foundation, conducted by Victory Research among 1 004 registered voters between 4 and 19 February 2025 using telephonic CATI interviews with a single-frame random digit-dialling design. The sample was weighted to match the national registered-voter population across language, age, race, gender, education, income, and geography. The survey carries a margin of error of ±4.0% at a 95% confidence level, with a design effect of 2.01. The figures in this piece are among likely voters within each metro; these subsamples (approximately 250–280 respondents each) carry a larger margin of error of roughly ±8 percentage points, so differences within a single metro should be read with that in mind.