Revolutionising voter engagement: Insights from Sprint 8 of MyDemokrasi
A blog by Gopi Ponnudurai, Implementing partner
Pilot: Pnyx, the digital public assembly; redefining representation
In today’s digital democracy, empowering voters to make informed decisions is vital for strengthening civic participation. Read more about how MyDemokrasi, an innovative platform designed to cater to Malaysian voters, aims to meet the needs of voters and transform political engagement in Malaysia in our previous blog posts. Sprint 8 was crucial for exploring user needs through real-world user testing. This sprint also aimed to further validate the accuracy of the AI sentiment analysis tool.
Understanding Our Users: The Power of Personas
One of the primary objectives of Sprint 8 was to delve deeper into voter behaviour by identifying and analysing user personas. Understanding user personas was critical as it allowed the team to hypothesise what different groups of voters would find valuable (in other words, understand who our customers are, what they are buying and why) and whether they would actively engage with politicians through the platform. We focused on youth aged 19 to 25, primarily from Malaysian public universities. This demographic was chosen as they are a group the research team had immediate access to and represent a potential key user base for MyDemokrasi. While this group is expected to engage with the platform, not all participants were tech-savvy or politically engaged, as revealed through the segmentation survey results.
To achieve this, segmentation surveys were conducted to categorise students into personas and explore what each persona valued. The 102 individuals responded to the segmentation survey (75 females and 27 males), with 97% reporting prior voting experience. Five distinct personas emerged: Busy/Time-Constrained Voters, First-Time Voters, Social Media Influenced Voters, Personally Invested Voters, and Politically Interested Voters. This detailed segmentation provided insights into tailoring the platform to meet diverse user needs and preferences.
Photo 1 & 2: Students from 3 different universities gathered for MyDemokrasi workshop
Engaging Voters Through Four Activities
To comprehensively evaluate how these voter personas interact with the platform, Sprint 8 adopted a multi-faceted approach through four activities. These activities were crucial to validating key assumptions and understanding user behaviour. The primary goal was to explore whether voters would find the platform valuable and if it could encourage active engagement with politicians.
Follow-Up Surveys: These surveys aimed to capture what voters from different personas valued most in a political platform, their preferences, and the challenges they faced in political engagement. Participants were asked about their expectations, preferred features, and perceived gaps in political content access.
Task 1 & Task 2: These tasks explored the ease of finding political information. Task 1 required users to search for political information via the internet, while Task 2 involved searching for similar information using the MyDemokrasi platform. The aim was to determine if the platform made it faster and easier for users to find relevant information compared to conventional methods.
Behavioural Tests: Using Clarity, we tracked user interactions on the platform to assess engagement patterns and usability. This included analysing click-through rates, time spent on different pages, and navigation paths to understand how intuitive the platform was for different voter personas.
User Acceptance Test (UAT) Survey: This activity evaluated user satisfaction, feedback on platform features, and recommendations for improvement. Participants rated various aspects of the platform, such as usability, information sufficiency, and design, providing actionable insights for future enhancements.
Table 1: Sprint 8 design and samples involved in each activity
While we did not explicitly evaluate whether users would revisit the platform, the data collected indirectly provided insights into features that could encourage retention, such as ease of access and trust in the platform's reliability. 96 participants completed some or all of the planned activities. The findings have provided insights into how users (voters) might interact with the platform which will inform how we forward.
Persona-Based Insights and Key Takeaways
Busy/Time-Constrained Voters
The follow-up survey revealed that this group seeks quick and concise information due to their packed schedules. Survey findings revealed that 50% identified insufficient time for candidate research as a significant barrier to meaningful political engagement, while 66.7% expressed a general lack of interest in political issues. Despite this, 83.3% of respondents favoured social media posts as the easiest way to access political updates, showcasing their preference for quick, digestible formats like short news summaries or engaging social media content. In Task 1 and Task 2 comparisons, Busy Voters appreciated the platform's efficiency, with MyDemokrasi reducing information retrieval time by over 50% compared to generic internet searches. This directly addressed their need for quick and concise access to political updates.
Behavioural tests further supported this finding, showing that 80% of these users scrolled through the most relevant sections, such as Profile and Key Achievements, while largely avoiding non-essential content like blogs and FAQs. Their minimal engagement with the politicians' communication features in Experiment 2 indicated a preference for information consumption over direct interaction, likely due to their time constraints.
The UAT survey reinforced these observations, with 60% rating the platform’s time-saving aspect as "very efficient." However, usability challenges, such as slow navigation and unclear menu structures, emerged as barriers to sustained engagement.
To effectively cater to Busy/Time-Constrained Voters, streamlining content presentation and improving navigation speed are paramount. Introducing features like quick summaries, offline access, and enhanced search functionality could significantly enhance their experience, ensuring continued platform usage and deeper engagement.
Table 3: Presents Task 1 & 2 Insights; Table 4: Presents UAT insights
First-Time Voters
First-Time Voters highlighted the need for simplicity and trust-building features during the follow-up survey. This group was heavily influenced by family and social networks, with 66.7% relying on these channels for voting guidance. A key concern for these voters was the lack of clarity regarding voting procedures, which created uncertainty in navigating the electoral process.
During Task 1 and Task 2, MyDemokrasi’s visual aids and clear instructions outperformed internet searches, saving an average of 5-10 minutes per search. This significantly boosted their confidence in understanding the voting process. Behaviour tests further validated these findings, with 90% of users scrolling to the bottom of candidate profiles, showing a strong focus on foundational voting knowledge. However, supplementary tools like FAQs saw limited interaction, suggesting that this persona prioritises basic, actionable content over exploratory features.
In Experiment 2, First-Time Voters showed moderate engagement with politicians’ communication features, primarily reading responses rather than initiating interactions. This behaviour reflects their reliance on guidance and foundational information rather than active participation in discussions.
The UAT survey confirmed these insights, with 71.4% of First-Time Voters agreeing that MyDemokrasi provided valuable guidance. However, they emphasised the need for more explicit details on voting procedures, highlighting their demand for transparency and clarity.
To effectively engage First-Time Voters, MyDemokrasi should enhance its educational features, incorporating step-by-step voting guides, interactive tutorials, and comprehensive FAQs. Additionally, strengthening trust-building elements such as verified candidate profiles and endorsements can encourage their participation and foster confidence in the electoral process.
Social Media Influenced Voters
The follow-up survey revealed that 50% of Social Media Influenced Voters consumed political content several times a week, predominantly on TikTok and Instagram. This group preferred dynamic, shareable content and engaging debates over straightforward news, indicating a demand for visually appealing and thought-provoking material. The survey also underscored a need for higher-quality, in-depth political discourse to align with their expectations on these platforms.
In Task 1 and Task 2 comparisons, MyDemokrasi demonstrated clear advantages over traditional internet searches, particularly in providing credible and organised information. Social Media Influenced Voters appreciated the platform’s ability to serve as a reliable counterpoint to less trustworthy online sources, which they frequently encountered on social platforms.
Behaviour tests revealed high engagement with interactive features, with 95% clicking on Profile pages and 80% scrolling through the content. However, static sections like blogs received minimal attention, highlighting their preference for dynamic, visually stimulating material. In Experiment 2, their engagement with politicians was moderate; they preferred consuming politicians' responses rather than initiating conversations, showing interest in learning rather than direct interaction.
The UAT survey further validated these insights, with 100% of Social Media Influenced Voters recommending the platform, citing its credibility and user-friendly design. However, they emphasised the need for more dynamic content tailored for social media sharing to align with their consumption habits.
To better engage Social Media Influenced Voters, MyDemokrasi should integrate shareable content, such as infographics, short-form videos, and interactive quizzes. Collaborating with influencers and creating a TikTok-specific strategy could significantly enhance the platform’s reach and engagement with this audience.
Personally Invested Voters
Values-driven content focusing on social justice, economic stability, and equality resonates deeply with Personally Invested Voters. Their political interest is often triggered by media coverage and significant current events. A majority (63.6%) preferred structured news and commentary, while 22.7% engaged in in-depth reading, indicating a focus on well-researched, issue-specific content. This persona reflects a balanced commitment to both community and personal well-being, valuing comprehensive and targeted information.
In Task 1 and Task 2 evaluations, Personally Invested Voters demonstrated a clear preference for MyDemokrasi’s policy-specific tools. Compared to internet searches, the platform saved an average of 7–10 minutes per search, offering more relevant information aligned with their core values. This efficiency strengthened their perception of the platform as a reliable and focused resource.
Behaviour tests revealed high interaction rates with localised tools and MP profiles, with 75% actively engaging. These users showed limited interest in general features like FAQs, instead gravitating towards issue-specific data. During Experiment 2, this group displayed a moderate level of engagement with politicians, focusing on exploring candidates' stances on policies rather than initiating direct interactions.
The UAT survey highlighted strong satisfaction among Personally Invested Voters, with 85% rating MyDemokrasi as "very helpful." This persona valued the platform's alignment with their priorities but suggested adding more comprehensive issue-based data and advocacy tools to deepen their involvement and make their engagement more impactful.
To better serve Personally Invested Voters, MyDemokrasi should expand its issue-focused resources and create more opportunities for advocacy. Introducing interactive tools that clearly outline candidates’ positions on specific policies, alongside mechanisms for voter-driven campaigns, could significantly enhance this persona’s engagement and participation.
Politically Interested Voters
The follow-up survey identified Politically Interested Voters as highly engaged, with 60% highlighting major political events as their primary motivator for staying informed. This group values meaningful conversations, often gravitating towards structured debates to enhance their understanding of issues. While social media serves as an important source of information, personal discussions and networks also play a significant role in shaping their political perspectives.
During Task 1 and Task 2, these voters praised MyDemokrasi’s detailed and organised design. Compared to internet searches, the platform significantly reduced navigation challenges, saving an average of 10–12 minutes per task. Their feedback confirmed that the platform’s layout and structure meet their demand for clarity and comprehensiveness. Although improved trust was not explicitly mentioned in the feedback, the positive responses regarding the platform's detailed and organised design suggest it may have fostered a sense of trust among users. The feedback indicates that these qualities likely contributed to building confidence in the platform, even if not directly stated.
Behaviour tests demonstrated strong engagement, with 90% of Politically Interested Voters interacting with dynamic features such as live Q&A sessions and exploring detailed candidate profiles. However, they expressed a desire for more real-time updates and enhanced communication channels with representatives, signalling a demand for immediacy and interactivity.
The UAT survey revealed that 70% of this persona rated MyDemokrasi as "highly effective," reflecting its capacity to address their information needs. Suggestions for improvements included adding live Q&A sessions and discussion forums to facilitate active participation and real-time updates to keep the platform relevant to their evolving interests.
Summary
The insights gathered from all five personas emphasise the critical role of tailored features in sustaining user engagement with MyDemokrasi. Busy/Time-Constrained Voters and First-Time Voters prioritise efficiency, simplicity, and clear guidance, while Social Media Influenced Voters and Politically Interested Voters thrive on dynamic, interactive content and robust communication channels. Personally Invested Voters value depth, alignment with specific issues, and opportunities for advocacy.
The findings also highlighted a shared interest across personas in having meaningful engagement with politicians. This underscores the platform’s potential to bridge the gap between voters and their representatives by fostering trust and direct communication. To achieve its goal of creating a more informed and engaged electorate, MyDemokrasi must continue refining its features to address these diverse preferences, ensuring the information provided is not only relevant but also encourages active political participation.
AI vs. Human Sentiment Analysis: Validating AI’s Role in MyDemokrasi
A pivotal aspect of Sprint 8 was the evaluation of the AI-driven sentiment analysis tool to determine its alignment with human evaluations. This experiment was designed without associating specific personas, instead involving participants from Experiment 1 and 2 who voluntarily joined. Two groups, each consisting of 10 voters, participated. Each group was assigned a unique set of 10 news articles to analyse, ensuring comprehensive testing across diverse political content.
The AI sentiment analysis tool was tasked with interpreting the sentiment of these articles, with results compared to human assessments. Findings revealed a strong positive correlation between AI and human scores (Set 1: r=0.93, Set 2: r=0.89), underscoring the AI's capability in accurately reflecting public sentiment. For instance, in Set 1, the AI and human scores showed close alignment for Article 8 (AI: 1, Human: 1.5, Margin: 0.5), demonstrating consistency in detecting positive sentiments. Similarly, in Set 2, Article 9 showcased comparable results (AI: 1, Human: 1.5, Margin: 0.5), reinforcing AI's reliability.
Despite these promising outcomes, challenges were noted, particularly with mixed-tone articles where nuanced sentiments were less effectively captured. For example, in Set 1, Article 4 (AI: -2, Human: -0.5, Margin: 1.5) highlighted the AI's tendency to overestimate negativity. Such observations point to the need for further refinement of the tool to enhance its interpretation of complex content.
This experiment highlighted AI's potential to empower voters by offering quick, reliable sentiment insights, thus enhancing their decision-making process. To unlock its full potential, the AI must be trained using a broader and more diverse dataset reflective of Malaysia's political landscape. Additionally, incorporating user feedback mechanisms could further refine the AI model, ensuring greater accuracy and user trust.
With these advancements, the AI-driven sentiment analysis tool can position MyDemokrasi as a cutting-edge platform for political engagement, delivering both transparency and efficiency to its users.
What did we learn and what do we still need to test?
Throughout Sprint 8, our experiments offered vital insights into voter behaviour, engagement, and the functionality of the MyDemokrasi platform. A key discovery was the platform’s ability to deliver concise and relevant information tailored to the distinct needs of identified personas. This validated the importance of segmentation, as user interactions varied significantly across voter categories. Tools like Clarity provided detailed navigation insights, suggesting a preference for MP profiles over other sections. However, challenges such as difficulties in searching for MPs—stemming from spelling or location inaccuracies—highlighted areas requiring immediate attention. Moving forward, we aim to test whether presenting data more interactively, refining search functionalities, and collaborating with politicians to ensure data accuracy can enhance user satisfaction.
Additionally, the platform showed significant promise in facilitating voter-politician engagement. User feedback indicated a strong desire for direct communication with representatives, with all participants expressing their willingness to recommend the platform. However, managing expectations around response frequency from politicians emerged as a critical challenge. To address this, future testing should explore implementing tools such as response templates for politicians and moderation strategies to ensure effective and respectful communication channels.
Another highlight of Sprint 8 was the validation of the AI-driven sentiment analysis tool. The AI exhibited a strong alignment with human evaluations, with statistically significant results indicating high accuracy in capturing overall sentiment. However, its limitations in interpreting nuanced, mixed-tone articles underscored the need for refinement. Expanding the training dataset to include diverse, politically relevant content and real-time feedback from users and politicians will be essential. Additionally, addressing biases introduced by media representation is crucial, as politicians with significant negative media coverage may disproportionately influence sentiment scores. Future testing should explore how to mitigate such biases, potentially through normalisation techniques or context-based adjustments. Educating users on how to interpret AI scores effectively will also be a key priority in making the tool more accessible and reliable.
Finally, we acknowledge the limitations of our current testing, which primarily involved a young, urban demographic. While this group represents a tech-savvy and politically aware segment, the ambition for MyDemokrasi is to serve the entire electorate, including rural and older populations. Ensuring accessibility and inclusivity will be vital to achieving this vision. Moving forward, efforts will focus on gathering user feedback from a more diverse audience, refining the platform’s features, and addressing the unique needs of all voter groups to ensure a truly equitable and impactful solution.
A Vision for the Future
Sprint 8 represents a significant milestone in our journey to foster a more informed and engaged electorate. While challenges such as usability issues, voter engagement with politicians, and AI refinement remain, the immense potential of MyDemokrasi to transform Malaysia’s voting experience is clear. By addressing these challenges and building on our successes, we can empower every voter with the tools and knowledge they need to participate actively in the democratic process.
Our path forward is guided by the understanding that enablers such as providing high-quality, unique data and clear, engaging presentations are pivotal in fostering meaningful voter engagement. However, we must also navigate blockers, such as addressing low engagement among specific demographics, refining moderation approaches, and bridging connectivity gaps to ensure inclusivity. From a technological standpoint, enhancing the scalability of AI tools, improving search functionalities, and integrating offline capabilities will strengthen the platform’s utility for diverse users. On the business front, exploring sustainable funding models and collaborations with local stakeholders will help ensure the platform’s longevity and scalability.
Equally important is the focus on equity and accessibility, especially for rural and underserved voters. Incorporating visually rich content in multiple languages and accommodating varied levels of digital literacy will ensure the platform resonates with all Malaysians. Lastly, policy alignment, including compliance with forthcoming AI regulations and election-related digitalisation standards, will solidify MyDemokrasi’s role as a trusted, future-ready platform.
As we move forward, MyDemokrasi aims to bridge gaps between citizens and their representatives while evolving to meet the needs of a broader, more diverse audience. By leveraging enablers, addressing blockers, and prioritising inclusivity, we can transform MyDemokrasi into a cornerstone of Malaysia’s democratic landscape. Stay tuned as we refine, innovate, and continue this exciting journey to empower Malaysia’s electorate!
If you’d like to dig in further into Pnyx’s journey…
🚀 Explore this pilot’s profile page
📚 Start from the beginning and read the learnings from Pnyx’s first sprint
