Finding a Single Source of Truth and Empowering 400 Self-Service BI Developers at Beeline


One of the most important steps toward success is understanding the reason behind actions. Defining that purpose through goals and objectives helps to clarify strategy and eliminate wasted efforts, and is usually one of the first steps of creating a strategic plan.

The purpose of moving business intelligence (BI) into a self-service model is to simultaneously empower end users and relieve some of the service burden. Granting customers of all types the power to guide their own data discovery is a critical step toward true data maturity. At the same time, reducing the resources spent on matters of upkeep and everyday assistance provides more opportunities for high-level productivity. The challenge, however, is in translating this idea into a workable model. The difficulty is especially acute when it involves coordinating the activities of hundreds of product developers.

Building a Single Source of Truth

Created in 1993, Beeline is one of the largest telecommunications company in Russia. With the slogan, “On your side” we provide mobile and broadband technologies to customers in Russia, the Commonwealth of Independent States (CIS), and Eurasia. Rather than just focus on technology, we focus on people and seek to use our products to connect hearts and minds.

Granting customers the power to guide their own data discovery is a critical step toward true data maturity.

My background is in computer science and applied math, but I have spent a lot of my professional career in BI. For the past year, I have been Head of Business Intelligence at Beeline. My mandate was to enable the company to better integrate data into our decision-making processes. Specifically, I was charged with building a single source of truth and helping the company to identify opportunities for automation and efficiency.

Automation is often viewed as low-hanging fruit that can quickly reduce costs and make operations more efficient. With automation, for example, companies can avoid the kind of errors that tired humans can make. However, there are still some things that humans do better than software. We wanted to automate wherever possible so we could put our human resources to their highest and best use.

Our Journey Toward Data Maturity

The word “beeline” refers to the shortest distance between two points. Despite our company name, our trip to a more effective, self-service BI operation was full of zigs and zags.

One of my first steps was to create a self-assessment survey to get a better sense of the environment. I asked our developers questions about their process, including the frequency and duration of the projects. Based on this feedback, we began to create an entirely new BI model that addressed some of their concerns. We learned from our assessment that the product team needed to improve their use of BI throughout the development process. The question was, should we achieve this by hiring outside analysts, or should we work to develop expertise in current employees?

In a prior position with a different organization, I was part of a team that took the latter route. We were able to take 25–30 people and build a complex, real-time, high load, end-to-end system. Based on that experience, I knew I could achieve the same transformation at Beeline.

Revitalizing Qlik Adoption and Architecture

When I arrived at Beeline, Qlik was already in use, but adoption was not consistent throughout the organization. The BI team had dashboards that tracked the number of overall users as well as actual developers, and where they were distributed in the organization. There was also a more detailed dashboard that was created for in-depth analysis of individual visits. This option could help staff to better understand what dashboards were useful to top management.

Since the eventual goal was an organization-wide transformation, we developed a heatmap to track users across all departments and units. This process was necessary to get a sense of who used the apps apart from top management. It also helped us develop a plan for approaching the remaining units. For example, if almost every member of a technical unit had already adopted Qlik into their processes, we could potentially highlight that group's successes and share with other groups where adoption lagged.

If we discovered that no one in an individual department used Qlik, we took that as an indicator that we needed to engage in more personal conversations. In some cases, we found opportunities to work together to develop some apps that would provide real value to those individual departments, and in turn, encourage adoption.

In addition to our efforts to transform the approach of users, we also had to completely change our architecture. Under the old process, a user might request the creation of a new dashboard that they believed might help them make a decision. The development team's process was complicated and cumbersome, involving a lot of steps and testing on our development platform. Only once this process was finished was the dashboard moved back to the production side and made available to users.

People generally request information when they need it, so a long lag time between a request and data delivery makes it impossible to make data-driven decisions.

The process worked, but it had a few problems. The first and most important was that it took a long time to receive the finished dashboard. People generally request information when they need it, so the lag time between a request and a usable document made it nearly impossible for any real business value to develop from our reports. By the time the information was available, decisions were already made.

The other problem was that once a request was made for a new dashboard, the requestor had no influence in the creation. The process itself was completely separate from the product, with the customer simply waiting while the developers managed the data. We knew that getting our staff involved in BI required giving them some ownership of the development process.

A New Process

After months of work, we began to build a new architecture in the fall of 2020. Under the new development model, we no longer have two disconnected platforms. Instead, we only have one. 

Dashboards are now created through a stream process. When someone requests a new dashboard, they receive three components of a new stream: development, staging, and production. As part of this process, we also created several roles attached to each stream. The Manager has complete control over the process and can manipulate the fields, schedule tasks, and move the dashboards. The Viewer can only view the production stream, and although they can see the incomplete dashboard, they cannot observe any of the activity behind the scenes. The Developers can see the development stage and any testing results. Testers are invited to try the dashboard before it goes live, so they have access to everything the final users can see. They can do everything from verify the numbers to evaluate the color scheme.

Through this new process, anyone who requests new dashboards are personally involved in creating the tool. That gives them ownership and encourages them to actually build something that has business value. The development team also finds it easier to manage the creation of the new dashboards via these streams. They can receive ongoing feedback from the eventual owner, allowing them to serve the needs of multiple clients without making them wait outside the process.

Six Tips for Developing Self Serve BI

While our process is far from finished, we have learned a few things that may be useful to others. 

  1. Take a two-pronged strategy to encourage the adoption of Qlik. Taking a top-down or bottom-up approach only addresses half of the company culture. You don't want business units constantly creating dashboards that no one will ever use. You also don't want your leaders searching for dashboards that were never built. Both sides have to grow toward the middle so that only useful dashboards are being created and that data has value to the organization. 
  2. Start with eager users. Working with people who are excited to be involved allows you to get moving quickly. The more people know about metrics and products, the more dashboards developed and distributed through the organization.
  3. Automate from the start. We have a dashboard catalog, but eventually, I would like to automate this process. Whenever the team completes a new dashboard, they must enter a description of the dashboard, the data sources, and the business purpose. That allows users to search through finished products to find things that may be useful for their purposes. Automating this will make the process seamless.
  4. Help people through the transition. We used a variety of methods to answer questions and help people to feel more comfortable. Meetups and open mic sessions encouraged dialog across users and units. Troubleshooting documents also helped new users to avoid common problems. We used features like Qlik Continuous Classroom to help people to develop data literacy.
  5. Use the available tools from Qlik to increase access. Give people the information however they want to receive it. For example, some of our units prefer to receive emailed reports. Rather than attempting to force them to actively use dashboards, we can set up Qlik NPrinting to provide them with emailed versions of the dashboard in a PDF format. Some of our call centers use this process to receive Monday morning reports about the metrics of the previous week.
  6. Counter resistance with a solution. Initially, we received some pushback from staff relating to the need to remember a password and sign-in process for yet another application. We dramatically changed the customer journey here by using Mobile ID. With Mobile ID, instead of remembering another password, people only have to enter their own mobile number. The system sends a prompt to the user's smartphone, and with a single click, they are logged in. This is especially useful for presentations and meetings when users are not sitting at a permanent desk.

Success Breeds Growth

We have seen some powerful results of our transformation. With each success story, the desire to learn more spread like wildfire throughout Beeline. Everyone is intrigued by the ability to create better data-based insights from a single point of truth, and that enthusiasm will continue to yield great results in terms of adoption and harnessing data for individual uses.

In our meetings, most of us has no need to prepare presentation decks or discuss numbers entered into Excel or PowerPoint, now, everyone can access the data directly and know they are examining the most up-to-date figures. This easy access to readily consumable data makes for better insights and smarter decisions for everyone.