Insights

The Three Cornerstones of a Modern Data Organization

An enterprise data strategy is a blueprint to help an organization modernize from data poor to data informed. Like many metaphorical journeys, it begins with honest self-reflection. An overestimation of existing data capabilities can be a significant barrier to change. Organizations often focus on cutting-edge AI prototypes as a sign of modernity and progress, but a more prosaic approach often reveals fundamental issues.

Having sponsored a series of AI conferences, we were struck by how many executives we engaged with revealed their own businesses’ struggles with data quality and access. Experiments in AI were not the problem; building the right team with the right technical capabilities to organize data and provide them to data scientists was the real barrier.

To change, organizations need a plan. At bare minimum, that plan should include the following:

- A clear and undisputed owner of the data within an organization

- Executive sponsorship for that person to mandate change within an organization

- Acceptable trade-offs between short-term impact and long-term value

- A business plan that clearly articulates where data value resides within the organization

Data are an asset and require the right level of investment to deliver growth. Capability must be developed over time so the business has the right skills to match newfound demand. Process and governance will lock in repeatable value that will outlast any person, program or initiative.

Barriers to smart data use within a business usually come down to data quality and access. Data quality underpins any data strategy, and without it, momentum is hard to establish.

Slow or limited access to data is a barrier to rapid decision-making. Sometimes the data are literally inaccessible in any meaningful way. Getting to 20:20 hindsight, while seemingly unambitious, is a necessary condition to creating a path to the future.

Sometimes there is a human barrier. Siloed departments or officially sanctioned gatekeepers may create a roadblock to data sharing. If data are not available at the speed in which decisions are made, decisions will be made without it.

Data-forward companies tend to focus on empowering colleagues through a culture of self-service and accountability. Collapsing the distance between data and its internal consumers is a great method for generating actionable feedback and improving systems to increase engagement and ensure development is for the benefit of the entire organization.

An emotive topic within data enablement is the single customer view. The theory is easy: create a single version of the truth that consolidates everything an organization knows about its customers. The reality is that a single customer view is a moving target. The customer data available are constantly changing, and each new data source makes the pool larger than before. This makes storage, computation and access increasingly difficult. A single customer view is out of date the moment it is identified unless it is understood as an iterative platform where new data can be plugged in. The high cost and low performance of single customer view projects usually revolve around the world moving faster than the project can deliver.

There is no intrinsic value in building data platforms and pipelines if the fundamentals are not in place. Value is only unlocked from data when put to use.

Insight often comes from context, and it requires multiple people with different skills and experiences to come together around a business problem. The future of data analysis is in its democratization. Providing people access to data more often is key to democratization.

Insights most often come from bringing two data sets together. A simple example of this is supply and demand. Retailers waste money promoting products that are out of stock. Inventory data are not shared with the marketing department in real time, which leads to ad buys that do not reflect accurate inventory levels. Data-forward companies transform this situation into an arbitrage opportunity, maximizing their revenue by shaping their demand to fit their inventory.

With the right experiment design and testing framework, incremental decisions can be made in parallel to minimize exposure to risk and create a balanced portfolio approach to investment opportunities. The process of making data-informed decisions creates its own data, which in turn generates more insight and more opportunity to improve performance.

Being data driven is a virtuous circle once a company gets inside the circle of decision-making. Insight to action closes the feedback loop, isolates the causes of success and failure and creates an environment to learn and grow.

Without it, companies just have expensive data technology.