Filling the gap: boosting data skills

We are living through times of immense flux in the employment market. People are changing jobs at a frequency seldom seen before, with July to September 2021 seeing total job moves increase to a record high of 979,000, largely driven by resignations rather than dismissals. More than 47 million people in the US also quit their jobs last year, according to US government statistics. The pandemic has wrought catastrophe on numerous industries. But above all, artificial intelligence (AI) and data are transforming the world around us, with data skills increasingly a necessity for businesses.

Yet, there is a data skills gap in the UK. Research last year from the Department for Digital, Culture, Media and Sport on quantifying the data skills gap found there were 178,000 vacancies for specific data specialist roles and those that require ‘hard’ data skills, where the majority of the work is centred around data and requires more advanced data knowledge. Almost half ( 46%) of businesses said they struggled to recruit for roles requiring data skills. The results were based on surveys of 1,045 businesses, 5,000 workers and 1,000 students in higher education or training.

Part of the issue is that AI itself its rapidly developing. Analysis from McKinsey estimates that AI could boost the UK economy by 22% by 2030, and government figures estimate the UK is third in the world for private investment into AI companies in 2020, behind only the USA and China.

“AI is prone to the same change that computer science had 30 years ago – it will, in my opinion, become a horizontal skill,” says Kian Katanforoosh, chief executive at Worker.ai. “One example is that 30 years ago, not everyone knew how to code – it was a very specialised skill. Today, it is hard to find a mechanical engineer or electrical engineer or an analyst that doesn’t know how to code, as it became a horizontal skill.”

A shortfall in qualified workers is increasing competition. Ryan Howard, data science consultant and developer of an AI platform for qualitative research called ‘Big Qual’, says that a significant part of the problem is that other industries pay more than the insights sector. “Our variety of work and genuine growth opportunities would make our industry an easy sell, were it not so far out of step,” he argues. “We pay 20% less than other sectors, across all levels, for the same base skill-set. In addition, we expect increased specialisation in sampling and survey analysis, higher levels of commercial acuity and agility to attack new problems every few days.

“Until that disparity is corrected, there is nothing to be done. We can talk all we like, set up mentorships and training academies; folks simply don’t hang around long enough.”

Katanforoosh says solutions to the skills gap must be tailored to businesses, rather than seeking a silver bullet to re-skilling that can be replicated in numerous organisations. “What we are seeing in our engagements with enterprises is there is a one-size-fits-all solution to upskilling that doesn’t work,” he says. “Employees don’t know where to direct their efforts, they have limited time for learning, and it is critical that we guide them – they need mentorship that is provided at scale. Companies need to adopt a ‘build’ approach rather than a ‘buy’ approach to skills.”

He adds: “AI projects won’t be able to develop without the required skills. You need people who understand how models drift, how data drifts, how it will be maintained and monitored, and making sure it is operating ethically and responsibility. If there are not enough people with those skills, this will have a cost for businesses.”

Building skills
The trick is building specialisms in the organisation, as well as upskilling more broadly across the entire business, says Daniel Singer, managing director of analytics at Kantar UK and Ireland. “Companies shouldn’t try to upskill every person in their insights and marketing teams to make everyone a data scientist – having a broad base of expertise spread across the whole business is often not as effective as building targeted in-depth teams working together,” he says.

“Instead, businesses should concentrate on identifying their objectives, working out what talent they need to achieve them and then recruit appropriately or partner with a consultancy with those areas in mind. Once on board, it’s important that analytics colleagues work collaboratively with marketing and commercial teams so that data use is purposefully embedded and supports a wider strategy, rather than sitting in a silo.”

Howard says that the consequences for the industry of not filling the data gap could be profound, especially for research methodology. “It’s not really about data pipelines and analysis – those skills can be replaced overnight, no biggie,” he explains. “The danger is that the same individuals that busy themselves with these things long enough naturally grow into our repository for research design, methods and quality control. These skills are nearly impossible to replace, their absence leaves a vacuum.

“We desperately need those willing and able to check under the hood, and if we can’t recruit or train them, this story ends with us as glorified opinion peddlers. Let’s be clear, no one is going to pay us for that.”

Stian Westlake, chief executive at the Royal Statistical Society (RSS), says that businesses need to have statistical and data science leadership in their organisation. Organisations including the RSS, BCS, The Chartered Institute for IT and the Alan Turing Institute have launched an Alliance for Data Science Professionals to help govern the data science industry and will seek to create a single, searchable public register of certified data professionals.

“You want an awareness of data and a comfort in using and thinking about data to be not something that is just in a specialist function, but something that is infused throughout the organisation, particularly in leadership roles,” Westlake says. “That is a mixture of making sure people with those skills rise into leadership roles, and making sure people already in those leadership roles get comfortable with operating in a data-rich environment. The end state is where the organisation lives and breathes data-informed decision making.”

Singer adds that many firms recognise that data analytics can play a critical role in shaping their understanding of human behaviour but they often don’t have people who are fluent in different methods and can identify the right solution to a problem. “Without that, there’s a risk insights teams will waste resource or fail to achieve their objective by using the wrong techniques for the job in hand,” he explains. “There’s no broad ‘analytics solution’ that applies to every challenge.”