The son of a friend of mine recently chose his GCSE* courses. At his school, each child is given a selection of choices based on past performance. Analysis of their data places them in one of a number of categories, each of which leaves specific GCSE choices available to them (these categories are mainly used to decide whether they’ll study general or specific sciences, whether a language choice will be open to them, that kind of thing).
IMy friends were surprised to learn that their son – a pretty bright kid - had been placed in a category that precluded him taking a language as one of his GCSEs. Although he’d struggled a little with German (new to him at high school), his performance in French was good.
So my friends contacted school and asked for a meeting, at which it was explained to them that the decision had been taken based on predictive analysis of student data. Students with their son’s data profile generally didn’t perform that well at languages by the time they reached GCSE. They’d based this analysis on his early indicators (UK Year 7, i.e. 11 years old) in foreign languages and English.
Crucially, it turned out the data they had on their son was inaccurate.
His parents were only able to suggest this may be the case by re-contacting his primary school and accessing their data, which contradicted the secondary school data. This resulted in head-scratching and an eventual re-categorisation of their son for GCSE choices, and he’s now happily studying for French.
Interestingly, the process for child categorisation didn’t include speaking to either his French or German teacher.
The crucial qualitative insight was lacking; the decision taken was wrong.
This story struck me as important, as more and more I hear – in a business context – people talk about data as the only way in which to understand and assess human (customer) behaviour. Data has become unarguable.
Of course data is an essential component of evidence that drives good decision-making. It’s the ‘what’ – what’s happening, what does behaviour look like, how does behaviour change if we do X, Y or Z? But without qualitative insight to provide depth, the ‘why’, to test hypotheses and develop solutions based on real understanding, data risks dehumanising.
In the case of my friends’ son, you can see how incorrect data and lack of qualitative insight dehumanised in an individual case.
More broadly in business, there is a risk that data-only decisions could dehumanise an entire customer base.
*Aged 16 exams in the UK, courses are chosen at age 13 or 14