Thrive! How To Create A Data-driven Organization

Building a Data-driven Organization and Culture: the Best Practices

According to Gartner estimates, 80% of businesses will start internal data-focused initiatives this year. Mainly, companies will focus on data literacy and the promotion of data-driven organizational changes. And it’s obvious because the importance of data is undeniable. The ability to collect information, extract valuable insights, and make business decisions based on this data is essential for long and sustainable development.

This guide provides core info and suggestions that can help your team turn into a truly data-driven organization. Before we proceed to the actual steps, metrics, and tools, let’s agree on the basics. Put simply, a data-driven business is one that makes all the decisions based on data rather than on gut feeling or unproven intuition. These decisions include all the levels, from company-wide actions like entering a new market to minor HR stimulations.

Overall, we think that such businesses should focus on three areas to succeed: a data-driven culture that promotes the importance of proper information gathering and usage, a data-driven structure that organizes this culture, and data-driven technologies that act as implementation tools for culture and structure. Further, we’ll cover all these points.

How to Start Implementing a Data-driven Culture and Structure

First and foremost, let’s understand how a team can praise data on all levels. Today, a lot of small and large enterprises rely on the traditional top-down structure. It provides for all the decisions approved and pushed by the leadership. This approach can’t survive in the data-driven world. While top management still should make global decisions and define development directions, all employees should be able to make their own decisions based on data, surely. It’s a cornerstone idea that forms data-driven cultures.

Talking about the next element you should think about in the beginning, there’s a so-called hub-and-spoke structure. It provides for building the main data team and integrating data analysts or scientists into all other departments. Having access to all main data flows in all business areas, analysts can help other employees with data-driven decisions and, most importantly, affect the global company decisions by providing the needed insights.

Source: O’Reilly

 

Well, it should be clearer now how data-driven cultures and structures look like. Keeping these ideas in mind, check the three main processes that form the foundation of any successful data-driven organization:

  • Data extraction. Stands for creating, collecting, and obtaining information from all sources, structuring this data, and integrating it into the company.
  • Data analytics. Stands for getting valuable knowledge and business insights from the collected data based on the best data analysis practices.
  • Data management. Stands for making data-driven decisions that improve the business operations, customer experience, the company’s metrics.

In other words, you should start building a data-driven organization by establishing the needed cultural and structural basics, as well as the processes. It’s crucial to understand that just hiring professional analysts and getting the needed tools isn’t enough. Data-driven transformation is a complex process that starts with general organizational changes and strategic planning. 

Overall, we can divide the process of becoming a data-driven organization in two parts. The first one is the roadmap itself that includes planning and vision forming processes. During it, you should analyze and evaluate your company to understand the potential transformation risks, benefits, and goals. The key result here is the ready data-focused company-wide strategy.

The second part is known as the actual change program. It provides for preparing the global internal change plan, as well as specific transition plans for departments and even employees. Implementation of these plans is possible through awareness and education. Learning and training ensure the final transition while analytics and monitoring let you track the results.

Hence, using the knowledge about data-driven cultures, structures, and roadmaps, you can start making the necessary changes. Eventually, they’ll lead your team to turn into a data-driven organization.

What to Measure: Core Business and Employee Metrics

Let’s move to more exact tips and suggestions now. This part focuses on business and employee metrics that help you with data-driven transformation. The next section uncovers useful technologies and tools that help to evolve smoothly.

Talking about metrics, we should identify the main stakeholders. Of course, on the business level, there’s one core entity – your company. You care about its success, want the team to thrive and rise, attract new customers, improve their experience, and so on. Still, looking at the business as a team of employees helps us identify specific personas:

  • Data analysts are people integrated into the departments. They know the business goals and pain points but also can work with data sets. The main goal of analysts is to connect business and data, translate questions to queries, and vice versa.
  • Data engineers are programmers and tech folks who know how to work with data. They ensure that information is accurately gathered, structured, stored, and that all the sets are available to analysts and other members.
  • Data scientists are adepts of AI, ML, and stats. Scientists focus on deep learning and mining to optimize data flows and extract valuable insights. These experts can be quants – people who design and tune mathematical models.
  • Security and compliance representatives ensure that all these processes and the general data-driven strategy are both safe to all parties and compliant to regulative requirements. Particularly, they care about data privacy.
  • System administrators are people behind the infrastructure. They manage and maintain databases, data warehouses, infrastructure hardware, and access rights. So, they focus both on security and data efficiency.
  • Ultimate decision-makers are the end-users within your data-driven organization who use the information to make various decisions. These stakeholders vary from project managers and UX writers to CTOs and CEOs.

When it comes to metrics, the choice is up to you. There are dozens of different parameters to measure, and the exact set depends on your goals and niches. For example, if a company wants to retain customers, improving UX, it should focus on metrics that reflect client interaction with the product/service. And teams that focus on rapid growth may be interested in marketing metrics, lead acquisition, etc.

In a nutshell, metrics help measure success (and failures), show how various related decisions affect the process. Further, we want to show some examples of metrics for a data-driven organization in three categories.

Revenue Metrics

This group unites parameters related to sales and marketing. Here are the examples:

  • Click rate.
  • Product views.
  • The most searchable items.
  • Cost of custom acquisition.
  • Customer lifetime value.
  • Net custom worth.
  • Conversion rate.
  • Leads.

Profitability Metrics

The metrics in this category refer to the efficiency of your business:

  • Days inventory.
  • Cost of product/service sold.
  • Gross profitability.
  • Earnings before taxes.
  • Return on investment.
  • Cash flow.

Risk Metrics

The third group is all about business sustainability:

  • Time to a product recall.
  • Volatility.
  • Probability of failure.
  • Value at risk.
  • Cost of risk management.
  • Number of risks identified/occured/closed/etc.

Essential Tools for Data-driven Organizations

By now, the aspects of data-driven cultures and structures should look clearer. We described how a company can start evolving in these spheres and which metrics it can leverage. Let’s also look at more specific cases – tools and technologies that your team may need. They differ by the maturity level of a data-driven organization and the employee role.

Source: Cloud Technology Partners

For example, check the maturity levels. Most likely, your team will be somewhere near the first or second stage if you’re just beginning the data-led transformation. Looking at the table above, we’re interested in the last column. Let’s compare:

  • Data Access level (your company stores info but don’t use it for decision-making often) – only spreadsheets like Excel or Google Spreadsheet are enough here. You also may need a dedicated database solution.
  • Consolidation level (your company starts integrating various sets of data) – relational database management systems or RDBMS are core tools here. The examples are popular databases like MySQL, Oracle Database, PostgreSQL, etc.
  • Reporting level (your company initiates data-based reporting and decision-making) – you may need more sophisticated enterprise-level warehouses. The examples may be Amazon Redshift, SQL Data Warehouse, and BigQuery.
  • Alerting level (your company notifies decision-makers about data/metric changes) – Big Data enters the game at this moment. To process, store, deliver, and track sets, you may need specific tools such as Spark, Datapric, EMR, etc.
  • Engaging level (your company handles deep analysis in all aspects) – moving from post-change alerting to prediction, you’ll need even more expert tools and custom models. They can be created with R, Python, and cloud ML software.

These are the basic technologies your team may need to gather and use data properly. However, tools also differ by use cases. It means that data analysts and data engineers often use different software. That’s why you should remember the needs of all stakeholders. Take data engineers. To handle their tasks, they need the mentioned RDBMS and Spark. But they also require other solutions: Hadoop, NoSQL databases, message queueing systems, etc.

Wherein, data scientists focus on different approaches. They create and tune models, use algorithms. That’s why their core technologies belong to the highest level of maturity of your data-driven organization. Scientists use R and Python, SAS, Mahout, and so on. Finally, data analysts focus on BI and research. That’s why their software includes Tableau, Cognos, and OLAP tools.

Summing it up, we can say that there are some core technologies and there are more specific tools for departments. If you’re starting your data-driven journey, it’s a good idea to look at the tools used by all stakeholders: relational databases and data warehouses.

Data-driven Testing + KPIs

In conclusion, we want to cover one specific aspect that can be interesting for any data-driven organization. We talk about testing here. According to the ideas described above, advanced data-focused teams bring information-based decision-making to all processes. In this case, software testing isn’t an exclusion. Why are we focusing on this process now? It’s simple. No data-driven team can succeed without proper software testing that ensures customer satisfaction.

Generally, data-driven testing or DDT is an approach to testing based on table conditions or values. To perform it, you need to design a testing algorithm and feed it with two types of data: inputs that are tested and outputs that act as the expected results. Then, the system compares two values and identifies which of them passed the test and which didn’t.

It’s a great tool when you need to run many repetitive test runs. Thus, instead of designing unique tests for each purpose, you tune the algorithm and just use different input and output data. What’s unique about this type is that you don’t include data in the test. Instead, you use data sources connected to the algorithm, e.g., CSV or XML files, or even MySQL databases.

DDT approaches are great as they facilitate app testing that requires different data inputs. After setting the test, you can use it for many sets and at many stages of development. Moreover, if everything’s correct, you’ll need just one file and one data repository for proper work.

On the other hand, data-driven testing may be inefficient if you don’t repeat tests regularly or if it’s impossible to create clear expected outcomes. Overall, it depends on your goals but DDT can boost automated testing significantly if implemented correctly.

At the end of the day, let’s look at some metrics and KPIs that you may be interested in during data-driven testing. Here they are:

  • Code coverage.
  • Defects closure rate.
  • Executed tests.
  • Passed tests.
  • Requirements passed.
  • Requirements reviewed. 

These are just a few examples. Each testing case is unique and it’s especially true for a data-driven organization. If you’re only beginning the transformation, don’t hesitate to ask us about the best software testing practices. We’re always happy to help!

Get started today
Grzegorz Kłos
Co-Founder
office@apphawks.com
Grzegorz Kłos - Apphawks Co-founder
© 2020 Apphawks. All rights Reserved.