Reflecting on this and some of the replies...
Timpson's can do this in part because their business model is simple. There is little need to use data for "discovery", i.e., understanding the underlying system. Other companies are not so fortunate.
This leads back to the question that all analytic projects should start with - "What are you actually trying to do/understand?"
And it's amazing how often that is an afterthought compared to just implementing vast and complex solutions on the recommendation of a consultant.
For example, I reckon that if I lent Jeremy Corbyn £20 I would be more likely to get it back than from Boris. If I needed a politician who stuck to his principles doggedly then Corbyn is a great shout. But that's not necessarily what I want.
Would I prefer a PM who inflicted billions of bounds of damage to the country's economy but would never pocket £20k himself (or herself)? Would I prefer someone who was openly communist or fascist but an honest one to a crook who wouldn't wreck the country? No.
I have a theory about the PM's resilience in the polls despite current "difficulties"...
An awful lot of voters like him personally. They don't kid themselves about his honesty. They've known his flaws for some time but don't care. They voted for him despite those flaws. 1/n
They're also aware that many other people detest him and are willing to use any available method to depose him, just as they tried to use any available method to prevent Brexit from happening. 2/n
But there is really only one way to depose a populist/popular politician - the ballot box. It worked with Trump in a way that impeachment simply wouldn't have. Voters like being the ones to choose their leaders. 3/n