36 Tests into this World Cup cycle, the Springboks’ results sort themselves by two names on the team sheet, Jasper Wiese at eight and Sacha Feinberg-Mngomezulu at fly-half, and the numbers behind that are starker than many Springboks supporters realise.

The top line

Since 1 January 2024 South Africa have played 36 Tests and won 30, which is the sort of record that tends to get waved through as the Boks being the Boks without anybody asking which of the 69 players Rassie Erasmus has used actually makes the difference. So we asked.

We took every one of those 36 team sheets, every result and every margin, and for each player with 10 or more matchday appearances we compared how the side did when he started with how it did when he didn’t, stripping out the year and the opponent so that a start against Portugal isn’t worth the same as a start in Auckland. The method and its limits are set out at the bottom, and, as with all these things, there’s limits of what we can interpret.

However, the two men at the top of that list are Wiese, whose starts the Boks have won 19 times in 21, and Feinberg-Mngomezulu, who has started 16 Tests at 10 and lost one, and neither Eben Etzebeth, Pieter-Steph du Toit nor Cheslin Kolbe gets near them. Damian de Allende, Ethan Hooker, Ruan Nortje and Siya Kolisi fill in behind, and the full 10 are in the first table. The method can’t measure a man who never misses, which is bad news for the biggest names, and we’ll come to why.

Plate 1: the 10 players the Springboks’ results move with, 2024 to date.

The fly-half question

The fly-half numbers are the ones that’ll interest, since they’re the most un-Bok like findings. Feinberg-Mngomezulu’s 16 starts at 10 have produced 15 wins at an average margin of nearly 21 points, and that run takes in Wellington 43-10, the 67-30 and 29-27 against Argentina, Paris, Dublin, the 73-0 in Cardiff and all four All Blacks Tests this year, three of them won. The one defeat was Ellis Park in August, 16-33, the first Test of that series.

Handre Pollard’s 11 starts have brought seven wins and four defeats at an average margin of under seven, and the four defeats are Durban against Ireland in 2024, Santiago del Estero, Auckland last September and Perth a month ago. Two of those were lost by a point, so this isn’t a tale of a fly-half being found out; but the gap in margin, 21 points against seven, is too wide for the quality of the opposition to explain on its own, because the opponent adjustment leaves it intact.

What the sequence shows is a selection migrating. Pollard was the 10 for the tight ones in 2024 and the first half of 2025, Feinberg-Mngomezulu started in Brisbane, Perth and the Ellis Park win over New Zealand in between, and since Wellington in September 2025 the younger man has started every Test against the All Blacks, France, Ireland and Argentina, whilst Pollard has been given Turin, Scotland at Loftus and the side with 13 changes that lost in Perth with Andre Esterhuizen captaining. Erasmus, in other words, has already made the call that the numbers now describe.

Plate 2: results by starting fly-half since January 2024.

The eight that changes the shape

Wiese is the forward equivalent and arguably the clearer case, because his absences have come through suspension and selection rather than age. The Boks have won 19 of the 21 Tests he’s started at an average margin of nearly 22; in the 15 he hasn’t started they’re 11-4 at under 14, and four of the six defeats in the cycle, Durban 2024, Ellis Park and Auckland in August and September 2025, and Perth, came with him either on the bench or out of the squad.

The mechanism is there on the pitch. Without Wiese the first carry off a lineout or a scrum goes to Kolisi or Kwagga Smith, both of whom are better over the ball than through it, so the gainline on first phase moves back, the second carrier arrives on the back foot, and the Bok pattern of two or three hard carries into a kick for territory starts behind the advantage line instead of over it. With him it starts in front of it, and the kick that follows goes 10 metres further.

The other eight

De Allende’s number is the one that’ll surprise many. The Boks are 21-3 when he starts and 9-3 when he doesn’t, and the adjusted swing of nearly eight points a game is the third largest in the squad, which rather suggests the inside centre who does the quiet carrying into the teeth of a blitz is still doing it. Hooker, 11 wins from 12 starts, is the wing Rassie now picks for the aerial game against the sides that kick; Nortje, 17-3, has become the lock he turns to when Etzebeth is rested; Kolisi is 20-3 as a starter, which at 35 is a number his critics should be asked to read aloud. Cobus Reinach’s 13 wins in 14 and Morne van den Berg’s five in five say the Boks have found a way of winning behind two different nines, and Damian Willemse’s 12 from 13 is the quietest argument on the list for the 15 shirt being settled.

Why the biggest names look bad

Now the caveat, and it’s a big one.

Run this method on Etzebeth and he comes out as a man the Boks are worse with, 18-5 as a starter against 12-1 without, and the same is true of Jesse Kriel, Ox Nche and Pollard. That’s the split squad talking.

The men who sit out Portugal, Georgia, Wales and Italy are the first-choice ones, so their “without” column is a run of 50-point wins they never played in, and their “with” column is every tight Test the side has had. The year and opponent adjustment takes the edge off that but can’t cure it, because a Rassie second string against Wales isn’t the same proposition as a Rassie first string against Ireland even if the fixture list says it is. Etzebeth started five of the six defeats; so did Kriel and Nche. That’s a list of who plays the hard games.

The players who do well here, then, are the ones who’ve played a genuine mix of both sides, and the reason Wiese, De Allende, Kolisi and Du Toit come out well whilst Etzebeth doesn’t is that Rassie has rested the first four more often, which gives the method something to compare. Make no mistake, 36 games is a small sample and an adjusted swing is a direction rather than a proof; the honest range on Wiese runs from a couple of points to 17, and on Feinberg-Mngomezulu from one to 17.

The longer view

Run the same test over the whole Erasmus era, 101 Tests from Washington in June 2018 to Perth last month, with the Jacques Nienaber years included because Rassie was director of rugby throughout, and the same two names sit on top, Wiese at plus six and Feinberg-Mngomezulu at plus six, with Canan Moodie, Kolisi, Jaden Hendrikse, Nortje, Tendai Mtawarira, Willie le Roux, Reinach and du Toit behind them, and De Allende level with Du Toit but outside the 10 on the second decimal. Etzebeth, Pollard, Nche and Kriel all come out negative over the eight years for the reason above, and over 101 Tests the pattern is the same as over 36: the men who show up are the ones Erasmus has rested, and the men who don’t are the ones he hasn’t dared to.

Plate 3: the same measure over the whole Erasmus era, 101 Tests.

What it means for the rebuild

So, in essence, you can chop and change, and Rassie does it more than any coach in the game; 69 players in 36 Tests says as much. But the numbers say some of those players add a long-term advantage the rest don’t, and the skill in the job now is working out which ones they are whilst he whittles the squad down to 33 come the World Cup.

Wiese is the clearest case the data has: when he plays he informs the whole forward game, the first carry, the gainline it sets and the territory the kick behind it buys, and the Boks have lost four Tests in two years without him and two with him. The flip side sits in the backline, where the most attacking fly-half has turned out to be the most successful one, 15 wins from 16 for Feinberg-Mngomezulu against a Pollard record that was never bad and is now, on the evidence, second. Rassie has drawn both conclusions already; the table shows his working.

How we did it

What was counted. Every capped Springbok Test from Wales in Washington on 2 June 2018 to Australia in Perth on 27 September 2026: 101 matches, 74 won, one drawn, 26 lost. The three Lions Tests of 2021 count as Tests; the Barbarians games of 2025 and the cancelled second Georgia Test of 2021 don’t. There were no Tests in 2020. The current cycle is the 36 Tests from 1 January 2024.

Want more from Planet Rugby? Add us as a preferred source on Google to your favourites list for world-class coverage you can trust.

Where it came from. Results and matchday 23s were taken match by match from SA Rugby’s team announcements and match reports, ESPN’s match pages, Planet Rugby and rugby365, with late changes applied so that each 23 is the one that took the field, and cross-checked against the published cap totals.

What “with” means. A player counts as “with” only when he started. Bench appearances aren’t credited, because a man who plays the last 20 minutes can’t fairly carry the result, and the record doesn’t reliably say who came on. The threshold is 10 matchday appearances in the cycle and 15 over the era, with at least five starts and five non-starts in the cycle and 10 starts over the era, so that nobody makes a table on the strength of two games.

The adjustment. For every match the margin was compared with what the Boks typically did in that year and against that opponent, both fitted across all 101 Tests, so that the swing is the difference between the adjusted margin when a player started and when he didn’t. A 50-point win over Portugal and a seven-point win over New Zealand are therefore scored on the same footing.

What it can’t do. It can’t measure a player who is rarely rested, because there’s nothing to compare him with; Etzebeth, Kriel, Nche and Pollard are the clearest cases, and their negative numbers describe the games they’re picked for rather than how they played. It can’t separate one player’s contribution from those he’s always picked alongside. It doesn’t weight a World Cup final above a July Test, doesn’t account for cards or injuries during a match, and takes no account of minutes played. And with 36 matches in the cycle, every number comes with a wide range around it; the direction is sound, the decimal places are not.