Make a Data-Driven Organization, CIO News, AND CIO

0
By Roja Boina

Large amounts of data can create a new era of evidence-based innovation in business, enabling the formulation of innovative ideas backed by solid evidence. Companies have accumulated data, invested in technology and paid dearly for analytical skills over the past decade in hopes of effectively delighting customers, streamlining processes and clarifying strategy.

However, a strong data-driven culture is elusive for many companies, and data is rarely used as the sole basis for decision-making. Transforming a business into a “data-driven” one isn’t always easy, and there will almost certainly be obstacles along the way. Indeed, data and technology alone will not make a business more successful. It requires a change of mindset and concerted effort from both management and people.

To ensure that the mission, goals and needs of the entire company are met – in terms of process and technology – executive advocacy, agility, data competence, and require a broad community to orchestrate change and do it effectively. With that, here are 10 steps to help create and sustain a data-driven culture.

1. The foundation for a data-driven culture is established at the top. Senior executives at companies with strong data-driven cultures assume that choices should be data-driven — that it’s the norm, not the exception. Employees who want to be taken seriously need to interact with senior leaders on their terms and in their language, and this habit is trickling down. Leadership from a few at the top can trigger significant changes in organizational norms.

2. Select metrics carefully. Leaders can have a significant impact on behavior by carefully selecting what to monitor and what metrics they expect staff to use. To do this, the company simply needed a much closer understanding of where its data is coming from and consuming than usual – which is precisely the goal.

3. Don’t put your data scientists in a box. Data scientists are often isolated, resulting in a knowledge gap between them and business leaders. If the analysis is performed independently of the rest of the business, it will not exist and will not create value. Those who have successfully met this challenge have done so in two ways. The first strategy is to make the divisions between business and data scientists as porous as possible. High-tech companies use another method. Not only do they bring data science closer to business, but they also attract business to data science, primarily by requiring people to master code and be conceptually proficient in quantitative matters.

4. Resolve simple data access difficulties quickly. Obtaining even the most basic data is the most common problem faced by people in various industries. Surprisingly, this condition persists despite a wave of initiatives to democratize access to data within organizations. To break the deadlock, the best companies use a simple method. They give everyone access to just a few critical metrics at a time rather than undertaking large-scale, time-consuming data reorganization projects.

5. Quantify the degree of uncertainty. Everyone recognizes the impossibility of ultimate certainty. Despite this, most managers continue to demand answers from their teams without the same degree of confidence, ultimately missing an opportunity. The fact that teams express and quantify their levels of uncertainty has three significant effects.

● First, it forces policy makers to confront potential sources of uncertainty head-on.

● Second, analysts get a better understanding of their models when they need to rigorously assess uncertainty.

● Finally, the emphasis on understanding uncertainty encourages companies to experiment.

6. Make your proofs of concept simple and durable. In analytics, the quantity of potential ideas far exceeds the number of practical ideas. Often, the distinction isn’t apparent until companies convert proofs of concept into production. In this regard, proof-of-concept engineering, where production feasibility is a key part of the idea, is a good approach. An effective method is to start with something industrial grade but trivial simplicity and then gradually increase in sophistication.

7. Specialized training must be available when needed. While you should include basic skills like coding in basic training, it’s more effective to train employees in specific analytics concepts and tools just before they’re needed.

8. Employees, not just customers, should also benefit from analytics. It’s easy to overlook the importance of data fluency in increasing employee happiness. However, empowering employees to manage data themselves can help. Few employees will be motivated to persevere and revisit their work if the idea of ​​learning new skills to better manage data is presented in the abstract. However, a task becomes a choice if the immediate results benefit them directly, such as saving time, avoiding rework, or retrieving frequently needed information.

9. Be willing to give up flexibility in exchange for consistency, at least for now. Many data-driven businesses have distinct “data tribes,” each with their preferred data sources, their own metrics, and preferred programming languages. This problem can be disastrous for an entire company. Trying to harmonize somewhat different versions of a metric that should be universal can be time consuming. Inconsistencies in the work of modellers also have an impact. If a company’s standards and coding languages ​​differ, every move of an analytics talent requires retraining, making it difficult to move around. Sharing ideas internally can also take too long if they have to be translated continuously. Instead, companies should use canonical measures and computer languages.

10. Make a habit of explaining analytical choices. There is rarely a single correct solution to most analytical problems. Instead, data scientists must make decisions based on a variety of trade-offs. Asking teams how they overcame a challenge, what alternatives they considered, what trade-offs were understood, and why they chose one technique over another is a smart idea. By doing this regularly, teams are more familiar with the approaches and are more likely to evaluate a wider range of alternatives or rethink key assumptions.

Conclusion

Any change, even becoming a data-driven organization, can be daunting. However, embracing change is an essential aspect of keeping up with industry trends and gaining competitive advantage. Ultimately, data-driven businesses are well-equipped, resilient and innovative, navigating the challenges of an ever-changing business landscape and satisfying customers who expect nothing less than the best.

The author is Principal Software Engineering Advisor at Evernorth

Share.

About Author

Comments are closed.