Connecting Investment in Data: Using Data Products to Drive Business Performance

Introduction

As a Chief Data Officer (CDO) you can make your CEO one of the top 20% performers versus their peers. They achieve this if they return 2.8 more total returns to shareholders. You can do this because as a CDO you have the power to do something no one else in the C-suite can do. You can align the organisation by measuring what matters and ensuring employees know how each action they take influences financial performance. This is one of the six capabilities the top 20% of CEOs have.*

Data has the promise of being able to inform decision-making right across the business. From reducing cognitive bias to allowing decisions to be made at speed, granularity, and levels of complexity that a human can’t simply operate at. Those businesses that execute the processes and decisions that underpin their business model more effectively, will generate more value than their competitors. Yet given the average tenure of a CDO is just two to two-and-a-half years**, the ability to deliver on this promise is evidently hugely challenging. How do you effectively enable employees to take the action that will drive the most performance?

CDOs are increasingly adopting data products to try and generate value and the return on investment that hasn’t always been present in data functions. But this is relatively new thinking that doesn’t have well-established practices. The risk is that it becomes badly adopted or seen as the new singular silver bullet – which it is not. Data products and data product methodology, if done well, can be an enabler to drive business performance.

What is Driving Data Products

Expectations of the promise of data, what it can deliver and how it can support business performance, continue to rise. We have progressed from data being primarily a back-office function to one that directly enables competitive advantage. Now, there is organisational anxiety that competitors are doing it better, and this is stoked by the abundance of data-related media we are now exposed to.

There are four components of data that are driving the promise of what it can deliver:

• Consumption variation. Data has moved from historically being consumed in a lagged fashion via PowerPoint in meetings to being embedded in workflows at the point it is needed through web, mobile, and even IoT devices.
• ML and AI breakthroughs. Various types of ML and AI are enabling new types of predictions to automate decisions and processes where a machine can now be more effective than a human.
• Data growth. The volume, variety, and velocity at which we can collect and analyse data from all sorts of sources mean that we have new and greater inputs to inform decision-making.
• Technology capabilities. With the cost of compute decreasing due to cloud technologies, it is now more economically viable to use data for decision-making. The speed and granularity at which these decisions can be computed mean that they can be augmented into workflows with no friction to the user.

Whilst each of these components are incredible enablers for employees taking the actions that best drive performance, individually they are all cost lines. They only create value when they are combined in the right way to enable an employee to take action. This is what is driving the change in data product adoption. CDOs increasingly recognise that ‘product thinking’ addresses two significant questions: are we solving a problem that is valuable? And will the product get used?

What Makes ‘Doing’ Data Products Hard

Data functions need to build the organisational time and capability to do data product discovery. Business employees might be experts in their roles, but they are not, nor are they expected to be, experts in translating problems into well-defined data products. This is the principal function of a Data Product Manager. The challenge is in finding people with the skill and the will to invest most of their time working with employees to identify opportunities for data products to support decision-making and actions. Around 25% of the data product lifecycle needs to be invested in robust discovery to ensure the right products are being built. But with data functions under pressure to deliver, this often drives teams to want to appear to be working at pace. This invariably leads to products being built that don’t get adopted and don’t drive value.

Data products only deliver value when they are used. The organisational changes necessary to sponsor and adopt data products can be a significant blocker to many data functions. Whilst leaders in organisations might claim to want to be data-driven, the inconvenient truth is that they may be resistant to data making their performance visible. They may also be resistant to data driving what actions they should be taking, preferring data’s role to be a validation of what they have determined is the right action in the first place. Data Product Managers must focus on only building data products where they have the organisational sponsorship and buy-in, deferring potentially valuable opportunities until they can build that support.

Data products are a team sport and require all the roles in a data function to come together as an aligned team to deliver through the full data product lifecycle. Data Product Managers typically don’t have line manager accountability over these roles, so a Data Product Manager needs to operate in an organisational design that sponsors them to lead without authority. All members of the team need to commit to the product strategy and roadmap which is the principal responsibility of the Data Product Manager.

How Data Products Drive Performance

Data products ensure employees know how the actions they take directly improve business performance. In doing so, data products deliver value when:

• A human or machine uses it to take action.
• The action it drives can be directly measured by the impact it has on business performance.

‘Data product thinking’ means understanding where there are repeatable opportunities for data to add value. This is the mindset that informs the following three data product propositions.

Performance Diagnostic

A clear diagnosis on which actionable metrics will drive the greatest performance improvement.

From: Employees overwhelmed with the number of reports they receive and where to focus actions.

To: The value proposition of a performance diagnostic is to identify where the allocation of resources will have the biggest impact on financial metrics. Make it transparent and visible to everyone in the organisation what actions their role can take to improve financial performance.

Imagine the CEO as the user, they have a single view of performance across all the customer journeys and processes within a business. The consumption is designed so you can drill through metrics into the granular root causes. For every role in the organisation, you’ve identified what actions that specific role can take to improve performance and give them a single role-based view of this. The entire organisation has transparency of performance and can align on the priority actions. Where possible the metrics are monetised to show what the improvement will make in terms of direct impact on financial performance like reduced lost sales, reduced costs, and increased margin. The user interface design should enable the user to easily identify the priority actions.

Actionable Alerts

Enabling action to be taken at the earliest opportunity.

From: Not giving employees the prioritised action they need to take at the point in time it will be most effective.

To: The value proposition for actionable alerts is that employees of an organisation take action to address performance at the earliest opportunity. This may require them to be prompted as they undertake their tasks. Guardrails are applied to metrics so that when performance deteriorates outside of a defined tolerance, the alert is triggered for a defined action to be taken to address the performance issue. These alerts can either be based on a prediction of a deterioration in performance — in this way, they are preventative action alerts or can be based on actual performance which makes them a remedial action alert. The speed at which you can let a user know of an action they need to take helps prevent further impacts on performance.

Automated Decisioning

Using ML and AI can support automating the most valuable decisions that impact performance.

From: ML and AI not adding business value.

To: A strategic focus on value creation. Starting with understanding what decisions in the business model could have the largest benefit from applying ML and AI techniques. Secondly, understanding who makes those decisions now, how they make them, and how they currently perform. Mature organisations will be clear on the measurable business value they think they can achieve before they start building solutions. Central to this is understanding the organisational impact of solutions before they are built. Most of the complexity in achieving success is in the organisational and business change needed to facilitate the use of new solutions to make decisions in new ways. This needs to be designed for and sponsored as part of data product discovery.

Conclusion

Companies that are successful in delivering data products to drive business performance will be the ones that recognise that empowering strong Data Product Managers and adopting ‘data product thinking’ are essential elements of a modern data strategy.

References:

*CEO Excellence, by Carolyn Dewar/Scott Keller/Vikram Malhorta.

**Why Do Chief Data Officers Have Such Short Tenures? by Tom Davenport, Randy Bean and Josh King

This article was featured in the first Edition of our Driven by Data Magazine. You can download the magazine and read more articles like this by clicking here.