Data Ethics in Practice - a provocation by Diana Spehar


The hand of an elderly person holding a hot drink in a ceramic teacup
Sky's offices in London

Diana opened up the provocation by outlining what brought her to work at Sky. She was hired not only for her knowledge and experience in data ethics, but also her ability to navigate complex corporate landscapes. This is an important skill, because her role is to support all teams in using data in an ethical manner – her success criteria are not solely based on the quality of the frameworks she develops or adopts. A large part of the challenge is getting everyone on board. As such, she takes a three-pronged approach.


Firstly, in a role like this, you have to engage everyone to a point where they see ethics as a day-to-day priority, and not something ‘extra’ which drains time and resources. In Diana’s case, she’s making great strides in putting ethics firmly on the board’s agenda. There are also opportunities to plug ethics into existing practices and frameworks: e.g. the data protection team frequently roll out or update privacy-by-design frameworks, and Diana is involved in these conversations.


The second part of the approach is operationalising ethics. This entails helping everyone understand how embedding data ethics in their work will benefit both their day-to-day operations and the business as a whole. One method has been to distribute an employee handbook which clearly outlines what data ethics is. Another is to run workshops and use engaging tools such as moral IT-deck cards, which help teams identify ethical risks of new features.


The third prong concerns cultural change. For Diana, this has meant going beyond showing people why company-wide adoption is important, and focussing on the how. So teams need to be given clear guidance on what features are or aren’t allowed, or what to do when exposed to unintended ethical risks – according to whichever ethical frameworks are being followed. Involving people in conversations and workshops goes a long way in shifting cultural norms within a company. There’s also an employee-led body that was set up to raise awareness about ethical practices, and to support teams with testing out new ideas in safe environments.

Diana Spehar

Diana then drew out some observations on her experiences so far at Sky – the nature of the role gave her a kind of freedom that she hadn’t experienced elsewhere. Sky gave her a very broad scope in which to define data ethics, which was useful when just starting out, because she wasn’t restricted to pursuing one pathway. In addition to this, the sponsor for data ethics at Sky is the Chief Data Officer (CDO), which means she has direct access to someone on the executive team. This proximity brings her closer to the ideal scenario: where data ethics informs corporate strategy.


"In Diana’s case, she’s making great strides in putting ethics firmly on the board’s agenda.
There are also opportunities to plug ethics into existing practices and frameworks."

Of course, data ethics is a nascent field, and so taking on this role did present some tough and unique challenges. One was building a business case for data ethics: there is very little regulatory push for ethical practices outside finance and healthcare, so Diana found she was building her own narratives around Sky’s business model. It was very easy to fall into using the usual generic talking points around this – such as how data ethics ‘enhances customer value’ or how it might ‘prepare us for upcoming regulations’. So part of the challenge here was tweaking these narratives to fit in with Sky’s goals.


Another challenge was navigating around the use of KPIs. Through her time at Sky, Diana has seen a lot of pitches for the potential use of KPIs for data ethics. In practice, however, data ethics is not something that can be measured easily. The public are generally unaware of how companies will merge customer data sets together and run analytics on these – unless a massive scandal reveals these practices outright. Unfortunately, industry practices are not always keeping up with the latest ethical frameworks, so producing clear quantifiable benefits of data ethics is especially challenging. Instead of trying to tack-on additional KPI’s for teams, Diana has gone into individual teams and had a look at how ethics can fit into their existing objectives.



"...data ethics is not something that can be measured easily. The public are generally unaware of how companies will merge customer data sets together and run analytics on these – unless a massive scandal reveals these practices outright. Unfortunately, industry practices are not – for the most part – particularly ethical."

The final challenge that Diana shared was finding effective ways of working. Teams have an expectation that they will be provided with a plug-and-play solution, with a clear checklist. In reality, Diana works with teams to co-create tools and processes for data ethics. This is often seen as an extra drain on resources, so Diana often gives teams two to three months to build out a use-case and test it.


As is to be expected with this kind of work, there were also a few things that Diana learned the hard way, over time:


● There is great value in taking the time to articulate the risks of not doing the right thing, such as damaging the brand, or even alienating entire audience groups. Spelling it out like that is often the best way of getting the message across.

● People’s individual motivations for engaging in data ethics are often not what you expect, and understanding these is key to getting the ball rolling. Through internal surveys, she learned that most people just wanted to protect the interests of their customers – and things like keeping algorithms accountable were low on the list of priorities.

● Build your social capital before you spend it. Diana’s role is seen as a function of the data team; other organisations might have their data ethicist sit in compliance. So, it’s important to know who your allies are, and when it’s appropriate to call on them.


"Most importantly, data ethics is extremely new and as such there are limited practitioners."

To finish off, Diana shared a few things she struggled with personally at the beginning. At first, she thought she had to acquire expert knowledge in a range of complex areas – such as machine learning and data law – in order to effectively do her job. Over time it became apparent that her understanding of these subjects only needed to go so far, and where needed she could lean on experts. Data ethics is not exclusively a technical field – a lot of it is about understanding governance.


Most importantly, data ethics is extremely new and as such there are limited practitioners. Data ethicists are usually the sole ambassador for data ethics in an organisation. Therefore anyone working in this nascent field should strive to be kind to themselves: connect with others doing similar work and share learnings, and through the years collective knowledge will grow, which will hopefully centre data ethics as a typical stream of operations within technology companies and beyond.




@diana_spehar _________________________________________________________________________


Sign up to our mailing list

_________________________________________________________________________


Stay up to date with the world of tech ethics by signing up to our newsletter here