Data science teams frequently struggle to deliver business value because the results of their work are not embedded in the organization. Often data science teams work somewhat isolated and try to push their output into the organization. As a Product Owner (PO) you are uniquely positioned to pull this work into the organization, thereby increasing the odds of creating business value. To do this effectively, you may need to better understand data science and how you can leverage it to build better products, regardless of the product you own. Let’s explore.According to a recent survey by Big Data Expo and GoDataDriven (2019), 79% of organizations see data as an essential part of their strategy. No surprise therefore that executives have put big data exploitation on their agenda over the last decade (NewVantage Partners LLC, 2019) and have deployed data science teams to generate value. As a result, one of the leading global data trends for companies at the moment is putting predictive models into production.
"The biggest challenge in any organization’s analytics journey is turning insights from data into valuable outcomes".
So, lots of new business value to look forward to then! Well, unfortunately not. According to McKinsey (2018), the biggest challenge in any organization’s analytics journey is turning insights from data into valuable outcomes. And indeed, many data executives report to struggle with achieving measurable results from their data science investments (NewVantage Partners LLC, 2019).
Embedding Data Science in the Organization
How then can companies overcome this struggle and unlock precious value from data science? That is exactly what this article is about. You do this by embedding analytics into the decision-making processes that are part of the ‘insight-to-outcome-journey’ (McKinsey & Company, 2018). Or, in other words, enable decision-makers in all levels of the organization to regularly and naturally make decisions that are driven by insights as the outcomes of these decisions create the value.
Product Owners, by virtue of their role, are such decision-makers and therefore the key to unlocking business value from data science. Let’s un-pack this a little further. One of your fundamental characteristics as Product Owner is to be the ‘Product Value Maximizer’. This means that in addition to having a solid vision for the product, you weigh stakeholder input together with knowledge about the marketplace and prioritize all this with maximum value creation in mind. This process requires constant decision-making by you. As such, you have the opportunity, arguably more than anyone else in the organization, to be informed and steered by insights from data science and identify where it can create value. A similar point was made by Hilary Mason in a recent podcast (CosmiQ Works, 2019).
Some companies have a data science team that is steered by a ‘Data Science Product Owner’ to make data science part of the Agile product management process. For example, Booking.com, CarNext.com and Sony PlayStation have recently advertised such roles (Linkedin, sd). These POs often have a strong background in data science (maybe they have even been a data scientist in the past). A Data Science PO may well be needed to make the data science team operate effectively but it’s not enough to fully embed data science into the organization and maximize value creation out of data. To achieve that, ‘analytics translators’ are needed. McKinsey coined this term and they explicitly highlight that extensive data science experience is not required for this role. Rather, the profile is more all-round, a Product Owner profile, I would say. That’s you!
Often companies employ multiple POs. When these POs all actively have data science in their toolbelt, the value-from-data-science-tentacles can reach further and wider into the organization thereby embedding it into the decision-making processes.
Imagine a Product Owner for a consumer-facing product that’s sold online. Naturally, this PO regularly looks at web analytics data and uses different ways of collecting user feedback to learn what to improve. But what about input on the optimal discount level for each consumer archetype? Or what about the best products or services to cross- and upsell each individual customer? Matching demand with the right supply is also important to keep internal stakeholders happy. What about demand forecasting that can be linked to manufacturing? Or finding optimal delivery routes to make a premium delivery service more efficient? A ‘value translating PO’ not only knows how to ask the right research question to the data science team but also has the ability to oversee the model’s limitations, data caveats and alternatives to reach the end goal more efficiently as well as effectively cooperate with the data scientists.
So, how to become a ‘value translating PO’ that leverages data science to build even more valuable products? Demystify the data science ‘magic black box’ by learning more about it. One way to learn is to follow training in data science fundamentals, processes and pitfalls. This ought to be part of the standard PO learning journey. Imagine the value that you can unlock in your organization by being data science literate and incorporating it in your day-to-day decision making. Finally: lots of new business value to look forward to. All thanks to you!