‘It’s noisy and ambiguous’: The dangers of being driven by data

27 Nov 2019

Image: © Björn Wylezich/Stock.adobe.com

Despite all the hype around data, Fathom’s Gareth Dunlop writes that data by itself rarely provides the nuance we need to solve real-world problems.

There are two kinds of people in the world – those who can extrapolate from incomplete data.

Data is called the new oil with good reason. It is an astoundingly potent force, increasing in value over time, and its power lies at the heart of the ‘contract’ we sign with the world’s digital giants when we use their platforms.

They give us their platform for free in return for our time, attention and behavioural data. We, the users, really have become the product. The internet undoubtedly knows more about the human soul than any deity.

Against the context of its power and ubiquity, we UX professionals do well to remember that hard quantitative data by itself rarely provides the nuance we need to solve real problems for real humans in the real world.

Data comes alive when it is married to the human stories behind the numbers.

Rail tales

A rail company in Britain was feeling very pleased with itself. All of its KPIs were on the up – traveller numbers, revenue and, crucially, customer satisfaction numbers in their key customer segments. So, its board wasn’t particularly enthusiastic when the head of customer experience said she had commissioned a service design project to explore what it was like to be a traveller on that network.

Most considered it waste of time, and likely to verify what they already knew – that they were fabulous.

The project commenced and focused on the human stories behind the numbers. The research uncovered extensive evidence that underneath the headline numbers, there were lots of travellers who were being let down by the network. One example in particular stunned the board into silence.

Janine, a single mum and regular traveller, agreed to be videoed for a week as she and her son travelled to and from work every other day. The train was regularly very full and, on most mornings, she only just made it on as other travellers squeezed onto the train in front of her, her bags and her buggy with her son strapped in.

On one particular morning, she was left behind, half on platform half on the train, as the conductor whistled for the train to leave. She explained that she has been shunted back off as other commuters pushed past, whereupon the conductor assertively told her to step onto the platform. The footage powerfully showed this young mother gently sobbing as the train pulled off into the distance.

The board’s collective sense of shame led to action and improvements for young parents travelling alone with dependent children.

Client collaboration

Closer to home, here at Fathom Towers, we had a very similar experience while delivering a piece of work for a banking client last year.

As we were moving from the research phase to analysis, one of the challenges that we faced was thinking about how we could most effectively involve our client fully in the analysis process. They had a number of different stakeholders (product owners, marketers, CX people, to name but a few) and we knew that we needed all of their assistance with synthesising the raw data in the right direction.

This was important from two perspectives: firstly, getting the most accurate solution and, secondly, ensuring that the client holistically felt invested in the outcome.

As we progressed through this process, there were a number of times where we felt our desire to collaborate might risk diluting some of our key observations from research. On one occasion, we felt the need to replay some primary user interviews from first-time mortgage applicants ahead of commencing a workshop to focus minds.

In the bundle of interviews we played were two short videos of first-time homeowners, couples, who were talking about the stress of trying to buy a home in a busy city, with more people wanting houses than were available and with costs rising all the time. Their weekends were dominated by house-viewing, blind auctions and general panic. Of the six individuals we interviewed as part of this work, twice we witnessed tears as mortgage applicants told the story of the stress that the whole process put on their relationships and family life.

This brought total clarity of focus to proceedings and a laser beam on the problem that the design needed to solve. It provided the entire project team with a fixed centre that guided us on the job to be done. It helped us avoid clichéd messages or falling into bad habits of lazy vanity marketing.

The human stories behind the mortgage sentiment analysis numbers gave the project team value that they couldn’t have received from the numbers themselves.

Data tells designers where to look

None of these improvements would have been possible without the research-driven pursuit of the human stories.

Data without story is like Barker without Corbett, Morecambe without Wise, Laurel without Hardy. Try one without the other and you’ll find yourself in another fine mess. Or answering each question before last. Or playing all the right notes but just in the wrong order.

Data tells designers where to look. And data lets us know if our design hypotheses are correct. But it can’t replace human thinking and it can’t think laterally.

Data is noisy and ambiguous and contradictory because humans are. Use your data to guide you to the human stories behind them and design for those.

By Gareth Dunlop

Gareth Dunlop owns and runs Fathom, a UX consultancy that helps organisations get the most from their digital products by viewing the world from the perspective of their customers. Specialist areas include UX strategy, usability testing, customer journey planning and accessibility. Clients include BBC, Bord Bia, Firmus Energy, Kingspan, AIB and Tesco Mobile. 

Gareth Dunlop runs Fathom, a UX consultancy that helps organisations get the most from their digital products.

editorial@siliconrepublic.com