A couple questions they tend to ask a lot is — “How do I show what the drivers are of this metric changing?” “How do I create the right suite of reports that link with each other to be used seamlessly?” Oftentimes, existing reporting tools like Tableau/Qlik/PowerBI don’t allow for such analytically insightful reporting.
Being able to integrate analytical tools like R into these environments for visualisation might be the key to solving this problem, but it then means a change of the skillset of your traditional BI analysts or developers.
One I’m still trying to crack — but at least have a number of good analysts to try to do so.
Data ScienceLastly, but surely not the least, the hottest job of the… I’ll stop there.
Apart from true AI/ML use cases, where you are building intelligent cars and robots, most end consumer products have their insightful POCs born out of data science teams and productionised in conjunction with the engineering function.
Data science teams focused on tasks tend first to focus on model accuracy and spending every last waking moment tweaking the model to get the best explanatory value.
Again, that’s great, if they’re one-off strategically aligned projects that need the most accurate answer, never to be used again.
Value-oriented data science functions tend to focus on the impact or change to the customer and what’s needed to hand-hold the customer through the change to improve likelihood of adoption, with initial sacrifice to the accuracy/complexity of the model, knowing they will be iterating it over time.
Emily Glassberg Sands gives a great overview of how data science teams assist in building great data products.
Her HBR article goes into the process of how to build them.
However, she takes a view from a digitally native business, one that already has an overarching product function.
For companies that do not have an overarching product function, the data science team is best suited to work in conjunction with data product teams and the business to bring the entire solution to bear.
The CustomerLast but most certainly not least, the most integral part of the data team that builds great products is the customer.
Now the customer here can come in internal and external forms, with the internal customer generally more forgiving of bugs/incorrect data.
They determine the features and help us understand the value of our product.
Without them, great products will not exist, data-powered or not.
The data team needs to develop and build solutions that make life easier for them, which in turn increases the adoption and ability to generate monetary value from them.
A product-centric mindset is a paradigm shift for most data teams today.
It changes people from being task oriented and perhaps even simple outcome oriented to being value driven.
It is just the beginning of the realisation that data is only the new oil if it is used to propel the right outcomes.
Oil in its rawest form existed for millennia and did not have any intrinsic value until we found the right uses (products) that benefitted mankind.
Data as a corollary is exactly the same.
The only way we truly monetise or create value from our data is if our customers/end users benefit from the products we create with data.
Only then does the data revolution truly take hold.
Thus, taking a product-centric view of data is the next step in creating a sustainable data revolution one that benefits both end users and companies alike.