A week from now, Roberto De Zerbi starts his first full season at Tottenham. The wreckage he inherited is well documented: back-to-back 17th-place finishes, three head coaches in a single season, survival secured on the final day. He arrived in March as a rescue hire and did the rescuing. Now comes the actual job.
So what, exactly, went wrong? Some will say it's because Harry Kane left.
That answer is half right.
The Kane myth
Kane left in August 2023. Here is what happened to the attack:
| Goals | Shots per game (for / against) | Possession | |
|---|---|---|---|
| 2021-22, with Kane | 69 | 12.9 / 12.9 | 51.8% |
| 2022-23, Kane's 30-goal season | 70 | 13.6 / 13.7 | 49.9% |
| 2023-24, first season without him | 74 | 15.4 / 12.8 | 62.1% |
First, the half that's right. With Kane, Spurs' shot counts were dead level with their opponents — 12.9 against 12.9. That team never out-created anyone; it out-converted them, because the man taking the equal share of shots was Harry Kane. What he took with him was the team's conversion.
Now the half that's wrong: the loss was absorbed within a year. Postecoglou arrived, Maddison came from Leicester for £40m to run the creation, Son moved from the left wing to centre-forward — possession jumped from 49.9% to 62.1%, shots from 13.6 to 15.4, and the first post-Kane season produced 74 goals, the best attacking year in this sample. Efficiency lost, volume found. For one season, the trade worked.
What dragged Tottenham into the relegation zone wasn't Kane's departure. It was a different collapse, two years later.
The real collapse: a supply line dismantled in two years
xG measures the end of the chain — the chances created. The harder question sits upstream: who moves the ball into the positions where chances exist? Watch that population disappear.
In 2024-25, Tottenham had six attackers with 900+ minutes and real output — three of them above two key passes per 90: Kulusevski (2.48), Son (2.13), Maddison (2.07), with Porro just behind (1.96).
A year later: Son sold. Maddison's ACL in August. Johnson sold without replacement. Kulusevski and Solanke gone from the list entirely. Which produced the season's most telling result: the most creative player left at Tottenham was Mathys Tel — a young forward with 1,358 minutes all season, most of them from the bench, four goals and one assist — top of the list at 1.99 key passes per 90. The top scorer, Richarlison, created almost nothing for anyone else (0.92). The supply line had thinned to a part-time forward.
Team xG followed the roster: 1.81 in 2023-24, 1.57 in 2024-25, 1.07 in 2025-26 — halved in two years, third-worst chance creation in the league. The defence, meanwhile, actually improved (1.70 → 1.37 xGA). Spurs didn't decline evenly. They became a team with nobody to pass to.
What De Zerbi sells
This is the logic of the appointment. His Marseille is the cleanest before-and-after in our data:
| Marseille | xG per game | Goals |
|---|---|---|
| 2023-24, before him | 1.64 | 52 |
| 2024-25, his first season | 1.91 | 74 |
A 16% lift in chance creation, twenty-two more goals, a second-place finish — the thing Tottenham has lacked for two years is the thing he has actually built.
It has never come free. Marseille conceded a third of a goal more per game in expected terms under him; his Brighton were the league's most vulnerable side to the fast break. In both of his last jobs, the attacking gain and the defensive cost arrived as a pair.
His seven rescue games at Spurs, curiously, showed the opposite face: 0.78 xGA — the meanest defensive stretch of Tottenham's season — with the attack barely moved. To stay up, the great attacking coach bolted the door shut first. When the football he actually wants to play arrives this season, the Marseille ledger suggests both sides of it arrive together.
The arithmetic of hope
Suppose it all works. Suppose Spurs keep last season's improved defence (1.37 xGA) and De Zerbi lifts creation by 30% — nearly double his Marseille effect. That takes xG from 1.07 to 1.39, and the balance to +0.02 per game.
Where does +0.02 sit in last season's Premier League? Eleventh — between Bournemouth and Leeds. Reaching Brighton's level (+0.23, sixth-best) would take a 50% lift. Top-four territory (+0.35 and up) takes 60% — which is simply a return to Tottenham's own creation of 2023-24. The hole is that deep: a genuinely exceptional coaching effect, delivered in full, buys back mid-table arithmetic.
The rest can only come from rebuilding the dismantled supply line. This summer, Tonali and Fernandes cost nearly £190m between them, van Hecke another £50m-plus — and De Zerbi has said publicly the recruitment isn't finished: he wants two more attackers. The midfield money is spent. The players who create chances — and the ones who turn them into goals — are still to come.
The Premier League opens in a week. The answers start arriving weekly.
Related reading
- Two titles, two different teams — PSG's back-to-back Champions Leagues, read the same way.
- What is xG? — the 24-second explainer behind every number here.
Data
- Premier League team and player data 2021-26 (xG from 2023-24; earlier seasons from shot and possession statistics).
- Ligue 1: 102 matches with xG, 2023-26.
