South Africa have lost eight Tests since the start of 2023 and, laid end to end, those defeats read like a coaching manual for beating the world champions. This is what the manual says, and what happens to it on the pitch.
The top line
Mike Tyson’s line about everybody having a plan until they get punched in the mouth has been quoted so often it’s lost most of its teeth, which is a pity, because it was built for this exact question. Ask any coach in the game how you beat South Africa and you’ll get a plan, usually a decent one, and usually one containing the words “move their big men around”; ask the same coach at the hour mark in Cape Town, with a brand new Springbok front row arriving and Cheslin Kolbe slotting everything he’s offered, and the plan has gone wherever Michael Spinks went.
The question has a date on it. The Springboks arrive in Europe in a fortnight for the November leg of the Nations Championship: Italy in Turin on 7 November, France at the Stade de France on the 13th and Ireland in Dublin on the 21st, having won all three of the July leg at home with 130 points scored and 49 conceded, and the finals weekend at Twickenham follows straight after, where the top side in the southern pool meets the top side in the northern one on the Sunday. South Africa and New Zealand sit level on 15 points, the Springboks ahead on points difference; France lead the north on 12, with Scotland, England and Ireland within two points of them. Four sides, in other words, are spending the next five weeks working on a plan.
This, then, is a piece in two halves: the plan as the evidence writes it and what happens to that plan on contact, and the plan itself is clear enough once you’ve been through the official match reports and the SANZAAR and World Rugby statistical packs on the eight defeats. Survive their set piece without letting it run the match; attack before their defence has rebuilt around the next collision; keep the ball with something close to fanatical care; kick with variety and from more than one pair of boots; keep enough on the bench to still be an attacking side at 55 minutes; and when the chance comes, take the points on offer rather than the points you’d prefer. “Move the big forwards around” doesn’t get close. What the evidence actually says is to make their defenders change jobs, then attack them in the second or two between those jobs.
The proof points are solid, with one caveat. Across the eight defeats, the winning side averaged 29.9 points, six of them scored 25 or more and not one of them scored fewer tries than South Africa, so scoring capacity is the first thing the evidence asks for. It doesn’t follow that the restrictive approach can’t work. Ireland’s two wins, 13-8 in Paris in 2023 and 25-24 in Durban a year later, were both done with a single try, exceptional defence and accurate kicking, and Ireland absorbed 31 dominant tackles in Paris to get the first of them. The caveat is the size of the sample. Eight defeats give you the recurring features of the performances that beat them, which is a different thing from a list of conditions proven necessary, and it’s why the counter-examples later in this piece matter as much as the plan.
The eight Springbok defeats since the start of 2023
|
Opponent |
When |
Score |
Tries (v SA) |
What decided it |
|
New Zealand |
July 2023, Mt Smart |
35-20 |
4-3 |
Fast ball and attacking kicks; five first-half clean breaks to none |
|
Ireland |
September 2023, Paris |
13-8 |
1-1 |
Absorbed 31 dominant tackles and 4.83-second ruck ball, and still won |
|
Ireland |
July 2024, Durban |
25-24 |
1-0 |
No Springbok try; two drop goals from the replacement fly-half |
|
Argentina |
September 2024, Santiago del Estero |
29-28 |
4-3 |
17-0 down after 15 minutes, 26-22 up at the break; survived the second half |
|
Australia |
August 2025, Ellis Park |
38-22 |
6-3 |
22-0 down; counter-attack, breakdown work and the bench |
|
New Zealand |
September 2025, Eden Park |
24-17 |
3-2 |
48% possession, 40% territory; lineout variations scored |
|
New Zealand |
August 2026, Ellis Park |
33-16 |
5-1 |
Turnovers won 14-5; replacement front row turned the scrum |
|
Australia |
September 2026, Perth |
42-38 |
5-5* |
42-17 up with a quarter to go before four yellow cards invited the comeback |
*South Africa’s five in Perth included a penalty try. Figures from official union and competition match reports and statistics.
The reality is harder to tabulate. Eight defeats in some 50 Tests is one in six, and in Cape Town in August New Zealand ran the plan almost to the letter, won every attacking number on the sheet and lost 33-26. We’ll come to that.
The plan on paper
You need physical credibility before any of the clever stuff works. If your carriers are repeatedly stopped behind the gain line, your scrum-half is digging for the ball and your support is arriving backwards, then their defence is making your next decision for you, and the whole Springbok system is built to produce that moment. But winning the collision count isn’t winning the match. Paris 2023 is the proof: World Rugby’s analysis gave South Africa 31 dominant tackles that night, Ireland’s ruck ball slowed to 4.83 seconds, and Ireland won. What the collision has to buy you is the next thing: clean ball, forward momentum and an option still open. The coaching prescription for that is short enough. Carry at shoulders with footwork before contact; use short passes between forwards so the Bok who’s lined up his hit gets a different carrier; keep an inside runner for the moment their outside men accelerate; and sit a second distributor behind the first carrying threat.
Against the rush, the temptation is width. A long pass to the touchline can work, but it also hands their outside defender the intercept or lets him hit the receiver before the support arrives, so the better route is to make him decide. Run the same picture with four endings: a genuine short carry, a pull-back, an inside return and a kick in behind, and he has to read before he commits; SA Rugby’s own account of Durban in 2024 identified Ireland’s inside passing as the adjustment that hurt them, and it was an inside line that produced Conor Murray’s try.
Then the ball has to leave your hands, and often, because the numbers kill the possession answer stone dead.
Australia’s win at Ellis Park, August 2025
|
Ellis Park, August 2025 |
Australia |
South Africa |
|
Score |
38 |
22 |
|
Tries |
6 |
3 |
|
Possession |
48% |
52% |
|
Territory |
39% |
61% |
|
Carries |
97 |
139 |
|
Kicks in play |
29 |
18 |
|
Offloads |
5 |
14 |
|
Rucks and mauls won |
77 |
98 |
|
Rucks and mauls lost |
1 |
5 |
|
Ruck and maul retention |
98.7% |
95.1% |
Figures from SANZAAR match statistics; retention calculated from rucks and mauls won against lost.
Australia scored six tries at Ellis Park in 2025 with 42 fewer carries and nine fewer offloads than the side they beat. They kicked 29 times to 18 and, out of SANZAAR’s match pack, kept the ball at roughly 98.7% of their rucks and mauls against South Africa’s 95.1, and that last number is the one to tattoo on the inside of the fly-half’s wrist. Three weeks later at Eden Park, New Zealand won with 40% of the territory, kicked 34 times and scored their first two tries off an attacking kick from a lineout variation and a movement after a 50:22.
The kicking plan writes itself from there. Contestable kicks with a connected chase and a man assigned to the second ball; long kicks that turn the back three; the cross-field ball only when the edge has advanced and your receiver has a genuine mismatch; and all of it from more than one position, so their chase can’t organise around your scrum-half. Mind you, receiving matters as much as launching, because winning the catch achieves little if the support arrives late and the next action is a turnover.
Which brings us to the ball itself. At Ellis Park this August, New Zealand won the turnover count 14-5, South Africa gave the ball away 23 times and the match went 33-16. The crucial part, though, is what happens in the seconds after the ball changes hands, because a Springbok side spread between carrying positions, support lines and the backfield is about the only Springbok side you’ll ever see out of shape. The transition calls want agreeing before the bus leaves: immediate width when their outside defence is short, an inside support line behind the first break, a kick into space when their wings have come up, and if neither the support nor the space exists, keep the ball. The reverse applies when you lose it. The nearest two have to contain the first carry whilst everyone else gets back into the line and the backfield, and three unnecessary bodies chasing a lost ball is usually how a knock-on turns into seven points.
The punch in the mouth
The punch comes in three shapes. In Dublin last November it was the scrum: six scrum penalties, one of them a penalty try, four yellow cards and a 20-minute red, and Ireland, who’d beaten South Africa three times in three years, finished the night 24-13 down with the referee holding a five-for. Set-piece pressure eats territory, points and bodies, in that order, and penalties are bodies.
In Perth in September it was the last quarter. Australia had built 42-17 by the 63rd minute and made nine line breaks to the Springboks‘ three before the cards started; then came four yellows, Harry Wilson, Ryan Lonergan, Joseph-Aukuso Suaalii and Hunter Paisami all in the bin and eleven Wallabies on the field at one stage, and a Springbok side carrying 13 changes and a first-time captain in André Esterhuizen scored three converted tries and finished four points short. By the end, the plan consisted of hanging onto the ball and watching the clock, and 42-38 reads like a T20 scorecard.
And in Cape Town in August it was the scoreboard, which is why that one should be framed and hung in every analysis room in the game.
New Zealand’s defeat in Cape Town, August 2026
|
Cape Town, August 2026 |
New Zealand |
South Africa |
|
Possession |
56% |
44% |
|
Metres gained |
481 |
350 |
|
Defenders beaten |
40 |
25 |
|
Tries |
4 |
3 |
|
Score |
26 |
33 |
Figures from official match statistics.
New Zealand had 56% of the ball, out-gained South Africa by 131 metres, beat 40 defenders to 25 and scored four tries to three, and lost by seven. South Africa converted all three of their tries and kicked four penalties, Kolbe’s boot doing the damage, and the Springbok blitz spent most of the eighty pushing New Zealand sideways rather than backwards, so 481 metres produced four tries and 350 produced 33 points.
Three of the eight wins took the punch first, which is worth remembering before anyone sells you a blueprint. Argentina were 17-0 down after 15 minutes in Santiago del Estero in 2024, turned it into 26-22 at the break with four tries in 20 minutes and still needed Manie Libbok to miss a 78th-minute penalty; Australia were 22-0 down at Ellis Park a year later; New Zealand conceded half a dozen scrum penalties in the first 45 minutes on the same ground this August, with Ethan de Groot’s knee going, and trailed 13-12 at half-time.
What survives the punch
Start with the scrum, because nothing survives if that doesn’t. New Zealand’s answer at Ellis Park was George Bower for De Groot late in the first half, then Asafo Aumua and Fletcher Newell at the break; Newell popped Gerhard Steenekamp open for the penalty that got them out of trouble and, late on, a five-metre Springbok scrum was penalised for wheeling against that replacement front row. The second front row has to be picked for the contest it will meet rather than the one the starters met, and Durban 2024 is the other half of the lesson: Ireland conceded scrum penalty after scrum penalty and won anyway.
The lineout is a selective contest, decided by field position and the men you’ve got: New Zealand stole three first-half Springbok lineouts at Eden Park in 2025 and scored off their own variations, with forwards receiving where the scrum-half normally stands. The maul is the second half of the same job, because in Perth South Africa scored off it through Johan Grobbelaar and then used the same drive to draw defenders before releasing Edwill van der Merwe, so somebody has to be responsible for what comes off the back of it whilst the rest are stopping the drive.
The bench has to keep the attack alive as well as the scrum. Angus Bell’s handling straight after half-time made Harry Wilson’s first try in the 2025 comeback; Ciarán Frawley’s two drop goals, off the bench, decided Durban. The window between 50 and 70 minutes is where the Springbok bench has settled most arguments since 2019, and the sides who’ve beaten them kept scoring through it: New Zealand’s burst of 21 unanswered points at Ellis Park this August came after the break, and Tomás Albornoz’s winning penalty in Santiago del Estero came in the 68th minute. Fresh Springbok forwards still have to defend, and the winners made those forwards fold, turn and reconnect rather than settling into one static collision after another, which means the support runners keep working after the first pass instead of watching the ball go sideways.
And the chances. Take the three. Rehearse the drop goal, because Ireland needed two of them after being held up over the Durban line and it’s the cheapest way of scoring you have when the attempt is on; a miss hands them the ball, a counter or a 22 restart, so it’s a decision rather than a free hit. Whether you go to the corner depends on the score, the clock, the weather, the lineout’s reliability and what the previous entries have produced, and going there repeatedly whilst your platform deteriorates is about the most expensive habit in Test rugby; South Africa’s 33 in Cape Town came from three tries, three conversions and four penalties.
Two routes through
There are two ways to do it, in the end: an exceptional defensive performance with accurate scoring, which is Paris 2023 and Durban 2024, or enough attacking production to withstand the Springbok response, which is the other six. Both need the same selection, and it’s a demanding one: a pack strong enough to secure the ball and skilful enough to use it, outside backs who can catch, more than one decision-maker, and a bench that sustains all of that when the match is at its hardest.
The November leg tests both routes inside a fortnight. Ireland in Dublin have the first one in their own recent history, Paris and Durban, and a 24-13 scoreboard from last November to settle; France have the attacking production for the second and a front-row question we’ve put before, a new French front row against Ox Nché, Gerhard Steenekamp, Wilco Louw and Thomas du Toit; Italy in Turin, with no points from three in July, have the plan every underdog has carried since 2023 and the same problem with it. Whoever tops the northern table, France as it stands, gets the Springboks again on the last Sunday in November at Twickenham, provided South Africa hold off New Zealand on points difference.
Put bluntly, the plan is whatever you’ve still got once the clever stuff has been knocked out of you, and the eight sides who’ve managed it since 2023 had, in one combination or another, a scrum that survived, a kicker, a bench and the ball at the hour mark. Tyson, as it happens, lost his title in Tokyo to a man who got up at nine.