Valuing Clarity

When it comes to data, there are few abilities more coveted than being able to explain what the heck it all means, why it’s important and how it can be taken out of the confines of your excel spreadsheet and into the real world.

Here at Conjura we fondly refer to this superpower as Clarity.

In truth, some are born with clarity, some achieve clarity and some have clarity thrust upon them. Without it the complex topics (Data – Analytics – Machine Learning – Data Science) that we deal with on a day-to-day basis would get lost in translation and left to gather dust in the corner.

This is why Clarity is one of our core values at Conjura.

In fact, after more than doubling our team over the past year, we recently created the Conjura Values Awards to recognise and reward team members who live and breathe those values. So it seemed fitting that our very first award go to our newly appointed Head of Data Science Delivery, Sean O’Driscoll for demonstrating outstanding clarity to everyone he works with.

If you’re wondering why Sean – we’ll tell you why!

Sean knows just how to seamlessly break down super complex topics without breaking a sweat. In fact, it’s been said that he can explain data science to absolutely anybody. Which is quite a feat. So we thought we’d let him do just that and stick it on our blog, which we feel is the equivalent of putting it in a frame on our wall!

Sean breaks down Data Science!

A baby learns by building up a library of experiences and interactions that give him context as to how the world works and the effect of each decision he makes. For example if he cries, he learns that he will get the attention he needs from caregivers to sort out his food, nappy etc. A successful outcome!

A data science model is somewhat like a baby, but instead of food, it cries for historical and outcome data. The more you feed it, the better it understands the relationship between customer behaviours and, for example, their likelihood to convert, repurchase or churn. That is how it gets better and better at making predictions.

With your healthy bouncing data science model all grown up, providing rewarding and statistically robust insights into your business – there is no limit to what you can do.

Do bear in mind that the data used for feeding your algorithm (e.g. customer behaviours) must be representative of ‘business as usual’. Because a model trained on pre-pandemic behaviours trying to perform well under “the new normal”, will be unable to discern what realistic expectations look like.

If all you have is atypical data, then the goalposts will constantly be moving and you’ll never know if you’re even aiming in the right direction!