How much is enough when it comes to data?
Data is everywhere. It is the new gold, or oil, or whatever. Organisations collect our data and then use it to persuade us to buy or vote or behave in a certain way.
But data isn’t certainty. Data is probability a best. Data can point or show but no amount of data can factor in every possibility or eliminate all uncertainty.
The fact that I have behaved in a certain way in the past doesn’t mean that I will behave that way in the future. Whilst it may show what I am LIKELY to do it cannot deal with outliers and it cannot predict the future. Data isn’t a crystal ball. Data can make a guess at one possible future and even take a number of possible futures and rank them in order of likelihood or probability, but it cannot be certain.
At some point, even after collecting all the data in the world a decision has to be made. And that decision carries an element of risk. It is impossible to eliminate all uncertainty. It is impossible to negate the chance that something will affect the outcome in a way that the data didn’t predict.
So that being the case when do we have enough data? At what point to we stop collecting and press the button on the decision? When we are over 50% certain? When we are 75% certain? When we are 90% certain. Or 99%? Or 99.9%? When?
Sometimes less is more. And more is less.
Sometimes collecting more data makes us need to collect even more. To try to minimise the risk or the guesswork or the element of uncertainty we collect more and more and more. This behaviour inevitably leads to what his commonly called ‘analysis paralysis’ – the continuous delaying of a decision until more data is available. Like a rabbit in the headlights of an oncoming car we stand stupefied and unable to decide which way to act.
Often the stakes dictate the amount of risk. If it is a genuine matter of life and death then we want the highest possibility of success and the lowest risk. But even that doesn’t necessarily mean that we need to collect more and more data. For the rabbit the key is to move. The longer it waits trying to analyse which is the best way to jump the lower it’s chances of getting out of the way of the car and the more likely it is to be reduced to a lump of furry roadkill.
Whatever the decision and whatever the stakes we ultimately need to make a call and accept that there is an element of uncertainty that we have to live with. We should collect what data we can to try to reduce the uncertainty as much as is reasonably possible but at the end of the day we need to call time on collecting data and accept that we have enough.
When you have what you think is enough data, just ask yourself: Are you sure?
😉