Officiating over the data

x Sponsored content on Research Live and in Impact magazine is editorially independent.
Find out more about advertising and sponsorship.

Since 2012, there’s been a 450% increase in the number of Fortune 1000 companies with a chief data officer (CDO) position. While clearly there’s been a rapid rise in businesses seeing the value of employing a central custodian of data, what exactly that role entails, is still evolving.

“The role isn’t completely settled yet. Just like any other C-suite role, the position of CDO will grow and evolve as our ambitions with data become greater and greater,” says Caroline Carruthers, chief executive of consultancy Carruthers and Jackson, and author of Data Driven Business Transformation.

She worked with data science software business Dataiku on its white paper based on the insights from more than 50 CDOs around the world. The research looked at the different types of CDO, the challenges they face, what support they need and how they fit within their organisations.

Titles aren’t everything, it’s not just about CDOs. Vice-presidents of data, chief analytics officers or even CEOs might effectively be CDOs if they are responsible for data and its use within their organisation. But having someone specifically in the CDO role does demonstrate a commitment to data transformation.

What’s crucial is that there is one person who takes ownership of the data strategy. So, how have the requirements for a chief data officer altered as the role has become more established?

“With the widespread understanding of data science as strategically important to business growth, the role of CDO has become slightly less science and slightly more business-orientated. You still need someone who has intimate knowledge of IT systems, but the role is now much less back-end and more about helping the company understand the strategic value of data,” says Carruthers.

In the survey, the CDOs were asked what motivated their data evolutions and the top three things were: increase company revenue ( 76%); decrease operating costs ( 56%); and increase team and project efficiency ( 56%). Competition ( 44%) and risk management ( 41%) were lower priorities for CDOs.

One stark finding was that only 8% of CDOs are content with the quality of the data they have access to. But Carruthers wasn’t surprisedby this.

“The bottom line is you’re never going to get it perfect, so you need to get on with it. You shouldn’t spend all your time trying to get your data right – let’s look at ways to improve data quality but we need to target where we put our effort,” she says.

No discussion about data is complete these days without referring to artificial intelligence (AI) and machine learning (ML) and, while they are priorities for CDOs, perhaps not to the extent one might expect.

CDOs’ top three data priorities were: data process automation and operationalisation ( 23%); AI/ML model creation ( 18%); and data governance ( 16%).

Eight per cent of CDOs do not leverage ML/AI/deep learning and have no intention of doing so, with an additional 8% saying adoption is a low priority. Fewer than half ( 43%) use these technologies on a small scale that has not yet influenced decision-making.

But Carruthers thinks this reflects data officers wanting to make sure they have their data right, that it’s organised and cleaned, before moving on with automation.

“There are so many companies that haven’t got their basics right when it comes to data. If you don’t understand and secure the fundamentals, you’ll inevitably struggle to leverage the power of machine learning and AI,” she says.

As companies’ use of data has progressed, the role of the CDO has changed, as has the skill set of the data teams they lead.

While first-generation CDOs are the ground-breakers, working in companies that have never had that role before, second-generation CDOs can build on the work of those that have gone before them, and change the role somewhat.

The report points to a five-step plan for first-generation CDOs: align data collection with business goals; democratise so that everyone who needs access has it via self-service analytics programmes; establish trust in the data and those giving the insights from it; foster an environment where data is driving decisions (often involving ML); and iterate to improve the breadth and quality of the data, analysis and insights.

Second-generation CDOs have four ways to improve data maturity: build trust to avoid having to prove return on investment (RoI) with every initiative; align data strategy across lines of business; focus on education; and enrich the data community.

It also suggests that the skills for first and second-generation CDOs are quite different and that it’s ‘rare for a CDO to carry out both roles effectively’.

Understaffed data teams were CDOs’ biggest challenge, followed by outdated or outmoded data ( 12%) – extreme expectations, manual inefficiencies and lack of buy-in from the rest of the company each got 10%.

A data team should consist of both scientists and analysts to create the best models and gain the best insight, although the report could not determine a clear trend in data teams. Just over a third ( 35%) of data teams from those interviewed consisted mostly of data analysts, while 29% were half analysts and half data scientists.

So how can CDOs overcome the problem of under-staffing, whether it’s analysts or scientists they need?

“Simple – by describing the art of the possible and sharing our enthusiasm for bringing people on the data journey. Data leaders need to be outward-looking and help others understand how important this data revolution really is,” says Carruthers.

She points to two data revolutions happening concurrently – the internal and the external. “The fourth industrial revolution is the external, and lots has already been written on how data is transforming society and industry. But there is also an internal data revolution – to convince other people in organisations that they need to treat data as an asset for their business, rather than something to be afraid of.”

Business transformation of this degree is not always comfortable or easy. Carruthers was most surprised by how many data leaders in the study were inward focused rather than driving forward a wider agenda.

For business transformation to be achieved, those leading data strategies must look outward, to offer their organisations a vision of how data use can help better understand and reach their customers.