Unlocking the Power of Existing Data
Baker Hughes has a strong culture of evaluating their B2B customer training programs, primarily utilizing the industry-standard Net Promoter Score (NPS). To help the team extract even deeper insights from their evaluations, I designed and validated a multi-metric dashboard. By triangulating the NPS with eight other quantitative metrics the company was already diligently collecting—such as instructor competence and training documentation—we empowered the team to easily visualize trends and identify specific areas for curriculum enhancement.
A task-based usability study revealed a 92% task completion rate and an "excellent" 87.75 System Usability Scale (SUS) score, proving that the tool successfully supported data-driven decision-making.
"How likely are you to recommend this course?"
That single question is the foundation of the Net Promoter Score (NPS). The metric is elegantly simple: respondents are categorized into Promoters (scores of 9-10), Passives (scores of 7-8), and Detractors (scores of 0-6). The final score is calculated by subtracting the percentage of Detractors from the percentage of Promoters. At Baker Hughes, this score served as a reliable, high-level indicator of course quality and customer loyalty.
The Challenge: Extracting Deeper Diagnostic Value
While the NPS is fantastic for a high-level overview, as a standalone metric it can sometimes oversimplify the multidimensional nature of a learner's experience. It indicates how participants feel overall, but it doesn't inherently explain why they feel that way. Was a high score driven by engaging instructor communication, or by the quality of the documentation?
To uncover the "why" behind the scores, the evaluation team often had to read through every single participant comment, which was incredibly time-consuming. But Baker Hughes is a highly data-mature organization. Over the years, they had already collected a wealth of data, consisting of eight additional quantitative evaluation metrics covering everything from logistics to course consistency.
However, these rich supplementary metrics were typically only analyzed in-depth when an NPS score fell below a specific threshold. This meant that a lot of valuable insights about average or high-performing courses remained unexplored.
My mission was clear: the solution wasn't to collect new data, but to design a tool that could effectively triangulate and visualize the rich data foundation Baker Hughes had already built. I wanted to create a bridge between their high-level satisfaction scores and deep, actionable insights.
Building with Purpose: Design Science Research
To ensure the dashboard would seamlessly integrate into the team's workflow and provide genuine value, I adopted the Design Science Research (DSR) paradigm. DSR is an iterative problem-solving approach that aims to enhance human knowledge by creating innovative, practical artifacts.
Before designing any visual components, I needed to deeply understand the stakeholders who evaluate these training programs. I conducted semi-structured interviews across the digital customer service team. Using an Influence-Interest matrix, I mapped out the key users to ensure the design focused on the primary decision-makers.
Collaborative Requirement Gathering
Through these conversations, we identified opportunities to enhance the current evaluation process. Using the MoSCoW prioritization method, we aligned on core requirements. A crucial "Must-Have" identified by the stakeholders was the ability to systematically analyze all courses, allowing the team to uncover trends in successful programs as well, rather than solely focusing on courses with lower NPS scores.
Structuring the Data: The Kirkpatrick Model
Baker Hughes had done an excellent job collecting diverse evaluation data, but to maximize its diagnostic potential, it needed a structured analytical lens. I introduced Kirkpatrick’s four-level training evaluation framework—a highly regarded industry standard.
I analyzed the existing dataset and mapped their eight supplementary metrics to the first two levels of the Kirkpatrick model to create a cohesive narrative:
Level 1: Reaction
"How did participants respond to the training?"
I mapped metrics like Course Duration, Training Logistics, and Instructor Communication to this level. This provides insights into the immediate, qualitative experience and engagement of the learner.
Level 2: Learning
"Have participants learned from the training?"
I mapped metrics like Instructor Competence, Course Consistency, and Training Documentation here. These metrics indicate whether the environment successfully facilitated knowledge transfer.
This theoretical grounding was a powerful addition. It enabled the team to connect a Level 1 issue (e.g., poor instructor communication) with a Level 2 metric (e.g., instructor competence) to instantly diagnose the context behind a specific NPS. Data triangulation was transformed into a structured, highly actionable diagnostic system.
Designing for Clarity: The Single-Page Strategy
My initial approach involved exploring multiple visualizations through low-fidelity sketches, which evolved into a prototype with multiple tabs to organize the wealth of information Baker Hughes had available. However, early collaborative testing with the team highlighted that splitting the data across multiple pages caused cognitive overload and hindered their ability to quickly connect the dots.
To better support their fast-paced decision-making, I pivoted the UX strategy. I embraced the principle of "focused simplicity" and consolidated the experience into a single-page Power BI dashboard. This eliminated complex navigation and allowed stakeholders to view high-level scores and detailed diagnostic metrics side-by-side.
Directing the Eye: The Gutenberg Diagram
A powerful dashboard is one where users instinctively know where to look. To structure the layout of the data, I applied the Gutenberg Diagram, a key Human-Computer Interaction principle that maps the natural western reading pattern from top-left to bottom-right.
1. The Primary Optical Area (Top-Left)
This is where the eyes land first. I strategically placed the most critical KPIs here: the Overall NPS and the Overall Satisfaction score. Stakeholders immediately get the high-level answers they are used to tracking the millisecond the page loads.
2. The Center Focus: The Radar Chart
To add the missing diagnostic layer, I introduced a central Radar Chart. This visualization simultaneously displays the eight supplementary training metrics. Instead of reading through hundreds of comments, evaluators can instantly spot the shape of the data and see exactly which metric (e.g., Course Interaction) is influencing the overall NPS.
3. The Terminal Area (Bottom): Exception Reporting
Following the natural visual flow, the bottom of the screen houses the "Training Sessions Requiring Attention" table. This implements the principle of management by exception: it automatically highlights courses that fall below Baker Hughes' internal benchmarks, making it effortless for the team to transition from analysis to targeted action.
Validating the Impact
To ensure the dashboard truly met the needs of the evaluation team, I conducted a rigorous task-based usability test with 10 Baker Hughes employees. They were given real-world scenarios to retrieve specific information, analyze trends over time, and identify course weaknesses.
Significantly exceeding the 80% success target.
Placing the dashboard's UX firmly in the "excellent" category.
The results were a fantastic validation of the team's data maturity combined with user-centric design. 100% of the participants successfully completed tasks related to extracting participant sentiments and identifying specific metric weaknesses. While building a full narrative across filters (Task 6) proved slightly challenging due to interactive limitations within Power BI, 80% of users still completed it successfully.
Final Takeaways
This project beautifully demonstrated the value of bridging raw data and human decision-making. By applying UX and Human-Computer Interaction principles to Business Intelligence tools, I helped Baker Hughes evolve their evaluation process. We successfully unlocked the potential of their existing data, providing the team with a multi-dimensional diagnostic tool that doesn't just ask "What is our score?" but provides the actionable insights to answer "How do we elevate our training even further?"