* By JM Benedetto
Any entrepreneurial activity must begin with the end in mind. This includes those related to data. Data is a means, not an end. They are tools to achieve business goals. This basic principle is often forgotten or overtaken by the hectic day-to-day activities of companies. Our teams end up so focused on what they do that they lose sight of why they took action.
Many of the conceptual models we use to operationalize our ideas in the evidentiary world also tend to look at activities and flows from start to finish. The data value chain is an example, with pipelines starting at raw data and ending at value created for the business. However, when assembling a data product, we must follow the opposite path, discussing what effect we want to cause, what human behavior we need to change. Everything else is a consequence of this first analysis.
A data strategy must combine processes, knowledge and assets into a data-enabling capability capable of creating value in the company's various business areas. At least three components are essential to a successful strategy:
- Applications: the business problems we are trying to solve with data (eg product recommendation);
- Tech & Data: the technological base to support the applications. Here enter the data, its governance, the software and hardware used (eg data lake, data viz, data science, integration, …);
- People: data influenced culture, data literacy, organization, profile (e.g. training, mentoring, job requirements, performance appraisal, leadership ceremonies, incentives).
The road to building a data capability is tortuous, with many pitfalls. As a former CDO (Chief Data Officer) of a major home improvement retailer, I had the opportunity to learn firsthand some lessons about introducing data in a century-old organization. I share some of those learnings below.
data democratization
The democratization of data calls for more governance, not less. It wasn't the car that transformed urban mobility; the ecosystem we build around the car, yes (Detran, driver's license, driving school, roads, traffic legislation, traffic guards). Giving access to databases to the business expecting teams to create value is just as effective as doing away with the police and arming the entire population, expecting people to defend themselves.
If we want to give more freedom to the business, we must structure, prepare, support, enable and control the data platform. One of the main missions of the data teams is to build Montessori rooms for our business teams designed to support autonomy and experimentation. This is nothing more than an application of UX precepts to data.
At the same time, we must not make the mistake of trying to govern everything. Governing is a risk management activity and has a cost, either in money or in speed. Govern what is essential. We don't have a traffic cop in every car, yet we still have acceptable levels of road hazards.
Code to solve the complicated, no-code/low-code for everything else
We are still young on the journey of data in companies, and much is under construction. When the journey began, coding was the only option. Our technical teams worked like craftsmen, building each pipeline like an artist makes a sculpture. Over time, the accumulated knowledge and cost share associated with data opened the door to a myriad of technical solutions, with a focus on simplifying, robustness and automation of the work of data teams.
It is high time, therefore, to leave the artisan mindset and embrace automation as a new development paradigm. Of course, we will still have parts of the pipeline in Python, but only those that cannot be resolved by the no-code/low-code path (tools for building applications without having to write code). This shift will do for data products what industrialization did for physical products: a drastic drop in cost, leading to an explosion in consumption.
We are all different (thank goodness!)
A big part of the challenge of using data in business processes is not in the technology. Many times, a simple linear regression could drastically change the performance of a process, if used. The barrier is in the organization's culture, steeped in what is considered right and fair, which is reinforced by the incentives we give our teams.
There is no magic pill for cultural transformation. However, we know that it involves empowering teams with knowledge, support and direction, as well as, of course, encouraging desired behaviors. Increasing the data literacy level of teams falls under the empowerment category, and is critical in the data-influenced transformation.
As data touches the entire enterprise, we sometimes think about corporate and standardized initiatives. However, different people need different levels of data literacy and sometimes adapted learning approaches. The needs of each person are linked to the role they play in the organization, and also to their personality. Thus, any data literacy program must start by identifying the teams responsible for the priority use cases, grouping people into populations with similar profiles and needs. Only then can we think of actions to increase the data literacy level of each population.
DataViz tools (data visualization) can be a great ally in introducing data to people's daily lives. These tools are teams' first contact with applying data to business problems (after Excel!) and they are ubiquitous in companies. A good DataViz tool can be the poster child for data.
DataViz, by definition, seeks to simplify insight and make interaction with data intuitive. The evolution of DataViz tools are the deCe platforms, in particular those that support Augmented Analytics. These tools offer a combination of virtual assistants, natural language and AutoML, among other enablers. When combined, these functionalities drastically reduce the level of knowledge necessary for the good use of the tool by business teams and allow laymen to perform complex analyses.
Conclusion
We data professionals have a mission to democratize access to data in our organizations. As with any organizational transformation, this process will require a lot of effort, and not everyone will be able to finish the journey. To increase the chance of success, we can leverage the collective knowledge we are accumulating as a community. Some important points:
- Data is a means, not an end. The end is the purpose of the business;
- Without governance, there is no democratization of data;
- Automation is essential to reduce the cost of data solutions;
- Simplicity first, bespoke data literacy later.
Qlik Cloud touches all the points mentioned above. We have an open platform that covers everything from integration to analytics. On the data management side, we are able to automate the construction of data lakes and data warehouses, as well as govern your data estate, regardless of your storage, compute and analytics providers. On the analytics side, we take your teams from insight to action using automations and AutoML.
*JM Benedetto is now part of the Qlik group as Manager of Solutions and Value Acceleration
Notice: The opinion presented in this article is the responsibility of its author and not of ABES - Brazilian Association of Software Companies
EN 
PT 
