Question: When you use data to help people make better decisions, are there any problems that don't categorize into the data that you use to help people and how would you configure the data to help solve those problems?

Andrew Harrison answered on 11 Jan 2020:

All decisions are about more than just the data and the best decision depends on how you measure how good a decision is.

If you measure the wrong thing, you might not get what you want. Also the implementation of the decision depends on things that can’t be measured in the data – an example is people and their behaviour.

Think of a call centre where you want the maximum number of calls answered. How many calls you answer depends on, amongst other things, the length of calls. So to answer more calls you can ‘call bounce’, which means to answer the call then end the call straightaway without speaking. This does happen in some call centres where people are struggling to meet their targets.

You got what you asked for but not what you wanted. When you said ‘answer the call’, you meant ‘answer the call, resolve the callers problem and end the call with a happy customer’.

The data can help you answer the question but not necessarily ask the right question, although it can help you understand the impact of asking the wrong question.