Meet the data scientist: What does it take to hold down 'the sexiest job of the 21st century' and how can agencies attract talent?

With over 2.5 quintillion bytes of data created every day according to IBM, every industry vertical wants to hire data scientists in order to capitalise on their ability make sense of huge amounts of data, spot behavioural patterns, as well as inefficiencies, and forecast potential opportunities.

However, such talent is not easy to come by, and neither is it cheap, according to industry experts.

Given that most media companies (both on the buy- and sell-side of the industry) are going through a period of unparalleled economic disruption, paying top dollar for such talent, and then bedding them into an outfit, requires a massive leap of faith.

The role of the data scientist

Many of the ‘digital first’ – or ad tech – companies already have such teams of data scientists firmly embedded into their operations. For instance, Facebook employs teams of data analysts to work with its massive data set to produce media insights for specific industry verticals, such as telecoms operators, etc.

Likewise, media owners such as the BBC, the Economist, FT and Sky Media are all recruiting data scientist teams to both support their traditional editorial teams as well explore potential new business opportunities by providing further insights to their audiences.

Such functions ape the traditional role of the media agency, further blurring the lines of the traditional role(s) of just about every tier of the media industry. This also demonstrates the potential for disintermediation of media agencies in the era of big data.

The race for talent in the big data era

Speaking recently with The Drum, Arun Kumar, president of Dentsu Aegis’ ad tech hub Cadreon, explained the intensity of competition for such talent between ad tech and digital media firms – which have data science firmly embedded in their DNA and business models – and the traditional advertising network agencies.

In many cases, this appetite for talented data scientists involves wooing PhD-level academics to the media business, as well as hunting for graduates from mathematics, statistics, economics, finance, and science courses – a far cry from the days when the majority of graduate recruits at a media agency would have come from a social sciences academic background.

Indeed, many agency chief executives are keen to point out how their recruitment priorities have changed, with some undertaking in charm offensives to leading academic institutions in such fields.

Academic overtures

Kevin Fitzgibbon, business science manager at GroupM’s MediaCom, explains how the media agency is making direct overtures with universities across the country to drum up interest from such talent, many of whom usually apply their skills elsewhere, such as banking or other areas of high finance.

“We’re now building relationships with universities across the UK, doing everything from going in and talking to the students to running media data science projects as part of the courses,” he explains.

This appears to be paying dividends with his data science team – or business science as the media agency has labelled it – now numbering 35.

He further explains: “When I studied economics and econometrics six years ago, no one really knew this role existed within media agencies. It was only when I went on the hunt for a job elsewhere that I discovered media data science.”

However, as much progress as MediaCom is making in its wooing of data-literate talent and implementing their insights at the core of its efforts, there is more work to be done when it comes to attracting those with such skillsets to media according to Raluca Efford, head of digital and social engagement at Direct Line Group.

She also points out how marketing brands and agencies alike have to think about what they can offer data sciences to get them excited about marketing, adding that discussions of an industry-wide attempt to woo data science academics to the media sector are taking place, but yet to be realised.

“If the first struggle is finding the talent, the second is attracting this talent. Data scientists can model marketing data, but can equally work with data in so many different industries and sectors,” adds Efford.

So what do data scientists actually do?

So once you’ve attracted talent that can crunch numbers and statistics to produce insights to consumer behaviour, just how do you put their skills to work?

Speaking at a recent event in London, Catherine Williams, the chief data scientist at ad tech outfit AppNexus who was formerly a PhD candidate at Columbia University, explained how her team was able to detect “behavioural patterns” on its ad exchange to root-out an endemic level of fraudulent traffic.

MediaCom’s Fitzgibbon describes how such levels of expertise are necessary, especially as marketers are under growing pressure internally to demonstrate ROI for their media spend.

“During the recession media budgets were cut and businesses scrutinised what money was given to marketing teams. So now, what clients really want to know is when, what and how they should be spending their money. This is where keen data insight is key,” he adds.

Integrating the Math Men and the Mad Men

As mentioned, ad tech and digital media outfits are arguably the most natural habitat for such talent, given that data science is in their DNA. However, more traditional media players with legacy divisions require some deft manoeuvring when integrating such data scientist teams, as the ‘Math Men’ and the ‘Mad Men’ are often two very different breeds.

Direct Line’s Efford explains: “While data scientists have incredibly detailed expert knowledge in their field, not all are great communicators. The challenge is to find people that can talk about their work in a productive way that is inspiring rather than intimidating to those who don’t have the same eye for data.”

Fitzgibbon points out how this is something MediaCom tries to impress upon its young data scientist recruits at an early stage.

“However, the essence of being a good media data scientist comes from taking something really complicated and explaining it in a really easy way,” he says.

“This is a skill we develop with graduates as soon as they join our team. For example, if a slide takes more than 30 seconds to explain and you’re seeing a few blank expressions, you need to rethink how you’re communicating.”

All parties are in agreement, that while data may be the lifeblood of the media industry – or ‘the new oil’ – it still boils down to communication to making it impactful.

Fitzgibbons sums it up thus: “Data science can provide some incredibly useful insights, but if we can’t explain them in order for the client and the planning and buying teams to use, the insights become useless.”

This interview was first published in The Drum's 28 October issue.