Within a few minutes of the final whistle of every Manchester United game, Ruben Amorim and his coaches receive a detailed but digestible report of all the key data points from the match direct to their mobile phones.
It helps to inform the manager and his staff about where the team have been good and bad, weaker or stronger, how hard they have worked and provides an array of metrics that present an analytical snapshot of the game. Perhaps some suspicions will be confirmed by the data or maybe it flags things that were more deceiving to the naked eye.
Amorim’s predecessor, Erik ten Hag, had been the initial beneficiary of such data reports, which were not available when Ole Gunnar Solskjaer, Jose Mourinho, Louis van Gaal or David Moyes had been in charge at Old Trafford.
Those reports were just one of the many platforms built and designed by United’s data-science team, which had not existed until 2022 when Old Trafford’s powerbrokers finally woke up to football’s data revolution – an area that Sir Jim Ratcliffe bluntly stated this month the club had “completely missed” over the previous decade.
It is clearly a source of irritation for Ratcliffe, particularly in the context of recruitment and the money frittered away on transfer blunders and inflated contracts that have contributed to the financial challenges now facing the club.
The Ineos chairman first made reference to it last December, 10 months after securing a minority stake and taking over day-to-day operations at Old Trafford, when he told the United We Stand fanzine that the club was “still in the last century of data analysis” and that it “doesn’t really exist here”.
Ratcliffe expanded on that theme in interviews this month, talking as enviously as he did admirably about the transformations at Liverpool, Brentford and Brighton while bemoaning that United are “still missing out because we still don’t have data analysis” at the club.
He even went as far as to suggest that “all we’ve got is Jason’s eyes”, a reference to Jason Wilcox, who was appointed by Ineos as United’s technical director in April last year and has ended up taking on more responsibilities since Dan Ashworth was ousted in December after just five months as sporting director.
In both instances, Ratcliffe’s overarching message was clear: United have to get much better at recruiting players if things are to improve on the pitch and, to do so, need the best data analysis to underpin their processes.
So are United really still in “the last century” when it comes to data analysis? Is it true to say it still barely exists at the club and that, when it comes to signing players and with another transfer window beginning to loom into view, there remains an over-dependence on “the eyes” of individuals and old-school scouting?
No one would dispute United were alarmingly slow to react to the innovations in that field, just as it is no great secret they are still playing catch-up with the Premier League’s pacesetters on that front.
But the idea that data analysis remains largely non-existent at the club does not stand up to any real scrutiny and is at odds with the belated but concerted efforts of the previous regime led by Richard Arnold and John Murtough, after being appointed chief executive and football director respectively, to bridge that void and modernise.
Telegraph Sport has spoken to numerous sources at United to establish a picture of the data operation at the club: what is in place, who is involved, how it has been developed and utilised before and after Ratcliffe’s minority investment in February last year and what is still missing.
Indeed, some of the bigger, more pertinent questions centre around how Ineos intend to use and grow data analytics going forward to help better inform the decision-making process amid accusations that they are actually unpicking some of the building blocks that were being put in place.
Nine months have passed since United’s first director of data science was sacked in an initial round of redundancies last summer, for example, and he has still to be replaced. Dominic Jordan practically built United’s data science arm from scratch after leaving Manchester-based online retailer N Brown Group to formally start work with the club in March 2022 after initially serving a six-month notice period at his former employer.
Sources have told Telegraph Sport that Sir Dave Brailsford, the Ineos sporting director and Ratcliffe’s eyes and ears at United, never actually sat down with Jordan to understand his role and what he was implementing as part of his reputed audit of football operations.
There was a certain irony then when a recent audit of United’s data division by Gary Hemingway, the Ineos group projects director, was said to have provided a broadly positive appraisal of the infrastructure and systems built so far.
Jordan’s exit after two and a half years certainly seems to have left a department that should be central to United’s short and long-term plans in need of new direction, leadership and investment.
The data team that Jordan built
The modern, airy surrounds of Manchester’s Bridgewater Hall were an appropriate venue to host the inaugural Field of Play convention just under a fortnight ago. The sports-data conference attracted a large crowd and plenty there were excited to hear the opening guest speaker kick off proceedings with an illuminating 30-minute address entitled “Clean data, clear decisions”.
Chris Shumba is head of data operations at United. He was born in Zimbabwe and worked as a cleaner, warehouse operative and support worker while a student at university in Sheffield where he studied mathematics. When he joined United in March 2023, he would make a point of walking the floors to familiarise himself with as many people as possible and was one of four key data appointments by Jordan, with whom he had briefly worked at N Brown Group.
Andrew Davies, who has a PhD in pure mathematics, had arrived the previous month as United’s first machine-learning scientist from cybersecurity specialist Netacea and who also knew Jordan from their time together at data-solutions firm INRIX.
Around the same time, Max Adema was recruited from sports data-and-analytics company StatsBomb to become the first of two so-called decision scientists, who are responsible for translating the data to the scouting and coaching teams into a language those football experts can readily understand. The second of those football-facing roles would be filled three months later by Alex Kleyn, who joined from Southampton having previously worked at Norwich City.
Southampton had launched a B team tasked with mirroring the style and system of Ralph Hasenhüttl’s high-pressing first team and one of Kleyn’s mandates was to use sophisticated data modelling to help illustrate where and how the second team needed to be more tactically aligned with the seniors. Additionally, Nick Grimshaw would also join that year as a senior data scientist but working on the commercial side of the business.
It had been an exhaustive, 12-month process to assemble such a team to the point where most were on the ground working. Jordan, a United fan who lives in Stretford and had been able to walk to his own job interview with the club, had received around 500 job applications.
Those were whittled down to around 50 before the process narrowed again. In between times, Jordan was working around the clock to first assess the quality of the data United needed before he could acquire it and then start building the scalable infrastructure and pipelines required to ingest and unify it all, work Shumba and Davies would later accelerate as they developed the data warehousing and software engineering.
At that stage, United had no subscriptions with private data providers, such as Opta, StatsBomb and Impect, who supply the location events and relevant metrics for every on-ball action such as shots, passes, dribbles and tackles. Equally, United had neither the platforms nor expertise to integrate data into their decision-making processes.
Steve Brown, United’s head of scouting operations, would have access to some summary data but the club were not buying the raw information that is the cornerstone of any data department’s work. That would change.
One of Jordan’s earliest tasks was to crunch the numbers on a shortlist of candidates to replace Ralf Rangnick, who was in interim charge following the sacking of Solskjaer in November 2021. Ten Hag came out on top and that data analysis was used to support Murtough’s decision to appoint the Dutchman. United’s first meaningful foray into the data sphere was underway.
As Liverpool’s director of research, Ian Graham was the brains behind the data operations that would help turbo-charge the club’s renaissance and eventual march to Premier League and Champions League glory under Jürgen Klopp. But in his fascinating book, How To Win The Premier League, Graham talks candidly about how his work “would have had little impact” but for the buy-in of the Liverpool owners Fenway Sports Group (FSG), sporting director Michael Edwards and Klopp.
FSG had observed the success of Billy Beane’s so-called Moneyball approach with the Oakland Athletics baseball team and applied a similar data-driven philosophy with their own franchise, the Boston Red Sox, in a bid to drive much greater value for money.
When FSG then bought Liverpool in 2010, John W Henry and Tom Werner felt similar methods could work in football and so began the data revolution at Anfield under Graham.
Similarly, Arnold and Murtough recognised there would have to be alignment from the top of the club down if data was to really cut through at United and, for a good while at least, Jordan’s team were pushing at an open door. They wanted to refocus football operations and ensure greater collaboration across departments. Now the management and coaching staff, analysts, sports-science, scouting and data teams would all be working across two parallel corridors at the club’s Carrington training base.
Arnold recognised the data department would require time and investment and offered both. A three-year plan was devised and Jordan ended up having his unit supplemented by a core of staff from DXC Technology, the IT services company who had signed a commercial deal as United’s shirt-sleeve partner and would help build “a distinctive architecture tailored for player data”.
One of the critical areas of focus was recruitment. For years, there had been no discernible football strategy, from style of play to transfers to wage structure. Key decisions were often made in isolation, sometimes on the hoof and invariably from London. As much as United wanted to get good players in, a significant a part of the data team’s job was to keep the bad ones away.
United had an internal scouting database called TrackerMan, a bank of reports collected by their sprawling network of global talent spotters. Yet a data-led approach would allow United to assess all the games and, in theory, afford scouts more time to then drill down on the players who rank highly in the metrics as part of an intended shift towards signing younger talents.
As such, Jordan’s team were tasked with building the statistical models covering physical and technical performance, injury history and financial profiling as well as providing the benchmarking to adequately compensate for the relative strengths or weaknesses of a competition against the Premier League.
Market misjudgements hard to comprehend
In that regard, fans will justifiably wonder how United then continued to make so many errors of judgement in the market. Some of those frustrations were also felt internally. Ultimately, the work of a data team is undermined if the red flags they raise – and recommendations they make – are overlooked as decision-makers act out of desperation or indulge a manager’s fancy.
In April 2023, United opted to award Luke Shaw a new four-year contract on highly lucrative terms that did not align with the data on the England left-back’s worrying injury record. Shaw had averaged just 17 Premier League starts over nine seasons and, in the two years since being handed that deal, he has made just 20 league starts, his last coming 13 months ago.
Rasmus Hojlund’s transfer from Atalanta is also an interesting case in point. United had been looking at a number of strikers that summer, including Victor Osimhen and Randal Kolo Muani after resolving not to get into a protracted battle with Tottenham for Harry Kane on the grounds of his age and price.
The data on Hojlund was encouraging and, with the right training, environment and support, there was a feeling a transfer presented significant upsides for the right price. The reports from the scouting team about Hojlund’s character and personality were also very favourable.
Nonetheless, the club’s financial modelling had suggested United should consider walking away if the price began to get above £50 million. United ended up paying Atalanta an initial £64 million rising to £72 million.
In reality, the warning signs had been there the previous summer when United, who had been determined not to be held to ransom, panicked after a bad start under Ten Hag and ended up spending wildly over the odds on Antony and Casemiro, who was awarded a four-year contract worth £350,000 a week despite already being 30 years old.
Not that data-driven insights are a guarantee of success, of course. The modelling on Mason Mount, for example, had not thrown up any undue concerns over injuries, which have since come to plague him at Old Trafford.
United, equally, can point to the determination of Liverpool and Arsenal to sign the England midfielder. Given how strong Liverpool and Arsenal’s data divisions are, neither club would have pursued the player if they had felt he was physically incapable of withstanding the rigours of an intense schedule at a top-six club but injury risk is hard to predict.
Why data was overlooked by Ineos
By late 2023, cracks had begun to appear. Arnold was on his way out as Ineos closed in on a deal with the Glazers for a minority stake in the club and Murtough knew he would soon be following suit. The data team assembled on their watch had suddenly lost two strong advocates even if there was a strong sense of optimism that Ineos, given their involvement in the data-rich sports of Formula One and cycling, would prove welcome bedfellows and there would be an opportunity to tap into that expertise.
The reality would prove different. Ineos had a bulging in-tray and the feeling grew that data did not figure prominently on their list of priorities.
Dan Ashworth’s arrival on July 1 as sporting director was considered encouraging given his experience of working at Brighton, where owner Tony Bloom had utilised the services of his data company, Starlizard, to take the club from the third tier to the top half of the Premier League.
Yet Jordan was sacked soon after Ashworth’s arrival, leaving a void when the sporting director was then reportedly asked to present data on potential managerial replacements for Ten Hag. Jordan has yet to be replaced but wrote on LinkedIn that the team he assembled at United were “the best I have ever put together”.
In fact, there have been no permanent data appointments since the summer of 2023. The Cambridge-educated son of a diplomat, Jordan is now chief data officer for Swedish software company Twelve Football where he has just helped to launch Earpiece, a new AI scouting product providing instant player data from anywhere in the world directly to your mobile device.
Strangely, Brailsford is not thought to have consulted with Jordan during his audit process. Ratcliffe’s remarks have left some wondering what kind of messages were actually being fed back to the co-owner about the club’s data operations.
In Jordan’s absence, deputy football director Andy O’Boyle was briefly enlisted as the data team’s point of contact before he left as part of the swingeing job cuts, after which Richard Hawkins, the well-regarded director of football insights and innovation, began lending his support.
In December, a fortnight after Ashworth’s shock departure, Hemingway was brought in from Ineos to begin reviewing United’s data practices and start overseeing the department. His findings are believed to have been positive.
Despite the current vacuum and loss of expertise, there has been data input into some of United’s signings under Ineos, most recently Patrick Dorgu from Lecce in January.
How highly United’s purchases last summer featured on the transfer lists provided by the data teams and to what extent they were asked to model potential replacements for Ten Hag is unclear. But the £42 million signing of Matthijs de Ligt from Bayern Munich involved a series of data checks that allayed some of the concerns about his injury record, pace and mobility.
United’s data approach still a mystery
Asked this month about the hierarchy’s plans around data analysis, United’s chief executive Omar Berrada said it was “at the top of our list of areas that we want to invest in and get better at”. “You need very good technical people that are complemented by very good data-analytics insights,” he added.
Interestingly, Berrada was pressed on whether United intended to do all that “in-house” or appoint the services of an external company. “It’s a great question but I don’t have an answer to that yet,” he said.
Given the substantial investments already made, it would be a surprise if United opted to rip that up, just as the prospect of effectively running the data-science team as a consultancy might risk devaluing its currency and weight in the decision-making process.
Shumba, whose primary task is to ensure the right data is available at the right time all of the time, has spoken publicly about being “on a mission to build the best data platform in sport, not just football” and turning the club from a “data-reactive” to a “data-informed” organisation.
Brentford, who have maximised the expertise of owner Matthew Benham’s Smartodds firm, ring-fence millions each year for data innovation in a bid to stay ahead of bigger-budget clubs, and Brighton continue to unearth gem after gem in the transfer window.
At Liverpool, Graham and his team were able to successfully concentrate all their metrics and innovative algorithms into answering one question: ‘How much does this player increase his team’s chance of scoring or not conceding?’
United are not at that point. Tracking data, which tracks every player’s location at a rate of 25 frames per second such as that provided by companies like SkillCorner, has only more recently been incorporated into their recruitment processes.
United also have access to tracking data from the official Premier League supplier Second Spectrum and use it for internal assessments of their players and opponents. There has been an element of United learning to walk before they can run but technology advancements in raw data continue to move at a pace and present new challenges around how clubs ingest, store, analyse and extract meaningful insights from such information.
Pose data, which contains 29 locations – including eyes and ears – per player every 40 milliseconds, for example, and artificial intelligence are considered the next two frontiers in tracking data.
Where United land with it all remains to be seen. Ratcliffe’s remarks have given the impression, wrongly, that the club are still obsolete in the data age. But they are still chasing rivals who have had a 10-year head start and risk losing the ground they have made up if a clear plan for the future is not established.
