IDC forecasts that spending on data and analytics technology and services will grow annually at a rate of 11.7 percent over the coming five years[i]. Why are companies prioritizing their investments in data and analytics over other technologies? Because new data about customers, as well as, data streamed from connected products is promising insights that will transform customer relationships, products and business models. However, a recent Gartner survey cites a tapering of investments[ii]. I believe this is because companies are realizing that pursuing analytics proof-of-concepts and exciting isolated stories is neither sustainable nor scalable.
Here are seven prerequisites required to turn investments in analytics into sustainable and scalable business outcomes:
1. Analytics are prescriptive and deployed to make operational decisions with a closed measurement loop. A large durable goods manufacturer wanted to quantify the value the IT portfolio to its product development function. It developed an advanced analytics model that forecast the contribution of an IT project to product development efficiency. The model rank ordered the projects, and prescribed an optimal portfolio[iii]. It also enabled verification of a project’s benefits post deployment by comparing them to the benefits claimed at the time the project was approved. Traditionally, the senior directors and VPs decided the portfolio. This was now replaced by the recommendation from the model (with minor tweaks from leadership). For analytics to deliver on their potential the company needs to embrace analytical methods for decision making where feasible, with a closed loop to monitor performance.
2. The company creates and continuously improves Analytical Data Assets such as Customer 360, Product 360 etc. A bank wanted to increase its mortgage sales by improving the conversion rate of customers looking for mortgage information on its web site. Because mortgage was traditionally handled as a silo product, that area did not have information to tailor personalized offers for the customer. To address this, the bank created a data set that combines information about the customer across all products, a Customer 360 Analytical Data Asset. The new data asset enabled powerful analytical models that generated personalized offers with superior take-rates. Analytically mature companies develop integrated data assets in critical areas such as “customer,” “supplier,” “plant,” etc.
3. Data is a corporate asset supported by all stakeholders. Data circulates throughout the enterprise, it is created in all areas and flows into others. Analytics is a grand collector of data. If the data is polluted, hard to get to, or does not represent the physicals of the company it impedes and distorts insights. Analytically mature companies have created processes, governance, and a culture to create timely, accessible, accurate, and high quality data. They continuously work on improving the value of their data and its availability, and route out duplication and bureaucracy.
4. The company scans the external environment for data sources that can increase the predictive power of its analytics. For example, weather data, cell phone usage data, or various membership lists can add predictive power to a data set because they are not correlated with internal data, yet may influence customer or product behavior. Once an opportunity for analytics is identified, the analytics teams may not have the time to source and evaluate new external data sets. One durable goods manufacturer that has made a sizable commitment to analytics set up a dedicated team whose sole mission is to identify opportunities for purchasing potentially relevant external data.
5. The company proactively manages analytics knowledge and drives analytics innovation. Producing high quality prescriptive analytics that make better decision than human experts requires a “secret sauces.” These “sauces” are highly evolved modeling procedures that consist of multiple steps each requiring different statistical, machine learning, and data science tools and techniques. Credit score development and credit card fraud detection are two examples of such procedures. In contrast to an individual data scientist working on a model to solve a problem, evolved analytics procedures require large teams to develop and carry out. Documenting these procedures, purposefully evolving them, and developing new procedures to solve new problems is the key to sustainable high performance analytics.
6. The company has a capable data and analytics technology architecture in place. Delivering exciting analytics insights in a timely way to the relevant points of decision requires a specific technology stack. Many of its elements (data replication, metadata management, SQL accelerators) are familiar to most IT shops, and some may be new (Hadoop, SPARK, niche tools, model server.) However, the integration pattern to reduce the time to insights is new. Figuring out an architecture that works in a given IT landscape, yet supports the agility and integration required for analytics is a challenge to overcome.
7. The company has as an effective operating model across operations, data, analytics, and IT. Should analytics be a service provided by IT, or should it be embedded in the business? Is IT responsible for data quality or is it a business responsibility? Who should own the analytical data assets, assure their utilization and that fractions of them are not recreated? Many questions need to be identified and addressed to operationalize analytics at scale. Companies need to work through these questions in a purposeful planning process. Interestingly, there is no “one size fits all” answer, as I’ve previously discussed[iv].
Delivering on the promise of analytics is a multidimensional puzzle of organizational change management, business transformation, development of new practices and deep technical skills, as well as technology. Where is your company on this journey? Have you encountered another prerequisite not mentioned above? Share your thoughts!