Case studyUser Research3 days
DANA: AI-powered User Insights Platform
An AI-powered way to run user research, turning raw user feedback into insights the product team can act on directly.

How much time do you have?
1. Problem Space Analysis
Despite efforts to gather user feedback through structured surveys—like the recent DIANA Awareness Survey—results often fall short of guiding real product decisions. Stakeholders across CS/CX and Product find the insights too shallow, lacking the behavioral depth and contextual nuance needed to uncover root causes or unlock opportunities.
But, why does this happen? What seems to be the problem? Let’s break this down with the 5 Whys analysis:
- The current user research method fails to deliver sharp and actionable insights. Why?
- Only measures surface-level metrics, not user behavior or decision-making context. Why?
- The questions only focus on awareness, not real journeys, usage pain points, or emotional triggers. Why?
- The questions are not tailored to specific journeys or product cases. Why?
- The survey was not co-created with product or CS/CX teams, causing a mismatch between insights and needs. Why?
- The user research lacks collaboration and is often rushed, with no early discovery phase to align goals and hypotheses.
So, the root cause is that the current way of doing surveys is not collaborative and often rushed, prioritizing speed and output volume over strategic alignment and behavioral depth. This results in data that is statistically rich, but with poor and unactionable insights.
2. Proposed Solution Analysis
To address the lack of collaboration and depth in the current user research process, we need to shift from traditional and static user research method to an AI-powered in-app user insights platform embedded directly within the DANA app. This solution enables continuous, contextual, and actionable insights for all teams, including Product, CS/CX, and UX. This platform aims to transform user research at DANA from something occasional and reactive into a real-time, dynamic, and behavior-triggered engine for understanding users.
How does this AI-powered in-app user insights platform work?
1. Event-Based Triggers
Rather than relying on general campaigns or scheduled surveys, the platform triggers micro-survey moments in real time, based on user behavior within the app. For example:
- After completing a top-up or payment
- After encountering a failed transaction
- After engaging with Help Center or CS chatbot
- When users explore a new feature for the first time
2. Smart Micro-Surveys
Each interaction starts with a short, friendly question tailored to the event. AI will dynamically select follow-up questions based on users’ responses. For example:
- If a user expresses confusion after trying a new feature, follow-ups explore which part was unclear.
- If satisfaction is low, the AI digs into root causes: “Was it the speed? the clarity? or the final outcome?”
3. Real-Time Insight Dashboard
All feedback is aggregated into a centralized insight dashboard, accessible to PMs, CS/CX, and UX teams. It provides:
- Feedback breakdown by persona and behavior pattern for tailoring solutions to different user segments.
- Sentiment tracking per feature or journey for spotting emotional patterns.
- Pain point clustering (e.g. Top 5 UX blockers in payment flow)
3. Value Proposition and Business Impact
Value Proposition: DANA Users
| Canvas | Details |
|---|---|
| Jobs | Share feedback and get support in-app without disruption, and feel heard when it matters most |
| Pains | Feedback requests are either too intrusive or too disconnected · Surveys often feel irrelevant, repetitive, or too general · Lack of follow-up or resolution clarity when reporting issues · It’s unclear whether feedback ever leads to changes |
| Gains | Contextual feedback requests that feel natural and personalized · Smart, seamless experience that knows when and how to ask questions · Confidence that their voice is valued and contributes to improvements · Easy reporting flows that feel like a part of the app experience |
| Product | Behavior-triggered, micro-surveys embedded seamlessly in the DANA app experience |
| Pain Relievers | Replace generic, one-size-fits-all surveys with contextual questions · Remove random or disconnected survey requests · Avoid over-surveying through frequency and persona-based logic |
| Gain Creators | Ask questions at high-impact moments (e.g. post-failure, during feature use) · Tailor survey tone and length based on user profile or past actions · Provide visual or lightweight formats (like quick polls, emojis, swipes) · Show how feedback contributed to improvements |
Value Proposition: Stakeholder Users
| Canvas | Details |
|---|---|
| Jobs | Quickly gather actionable insights that align with real behavior, without relying on manual research. |
| Pains | Current research is too siloed, slow, or surface-level · Surveys don’t adapt to behavior or target the right users · Data lacks depth and actionability · Research tools don’t integrate smoothly into workflows |
| Gains | High-signal insights delivered right when teams need them · Easier to advocate for changes with clear user-driven evidence · Continuous discovery framework that adapts to user journeys · Sharper personas and pain points uncovered over time |
| Product | Modular research engine with smart triggers and integration into existing tools and workflows |
| Pain Relievers | Replace generic surveys with adaptive, behavior-aware logic · Reduce manual research workload through automation and smart targeting · Prevent over-collection by surfacing only what’s relevant to each team · Enable cross-functional visibility without data segregation |
| Gain Creators | Reveal user problems product teams didn’t know existed · Turn every feature release into an insight opportunity · Drive roadmap alignment with sharp, persona-based patterns · Build a scalable insights machine that improves over time |
Business Impact
The AI-powered in-app user insights platform delivers mutual value for both DANA users and stakeholders by seamlessly integrating lightweight, behavior-triggered feedback mechanisms into the app experience. This innovation enables context-aware interactions that respect the user’s journey while unlocking powerful, real-time insights for continuous product improvement.
From the users’ perspective, this platform…
- Enables contextual feedback, not spam: Micro-surveys are triggered by real actions (e.g., failed transaction, new feature use), avoiding irrelevant interruptions.
- Feels like part of the journey: Feedback flows are seamlessly embedded in the app experience, keeping the tone friendly, short, and intuitive.
- Ensures their voice feels heard: Users can see how their feedback contributes to visible changes, increasing trust and engagement.
- Eliminates feedback fatigue: Smart logic avoids over-surveying, and uses formats like emojis, quick polls, or swipes to keep it frictionless.
From the stakeholders’ perspective, this platform…
- Enables faster and smarter decisions: High-signal insights delivered at the right time, tied to actual user behavior—not just static metrics.
- Delivers effortless scalability: Reduces dependency on third parties or time-intensive manual surveys, freeing up research and product workload.
- Sharpens team alignment: Brings visibility across teams with auto-generated dashboards, prioritized themes, and persona-level pain points.
- Embeds insight into every product aspect: Every interaction becomes a learning moment, which transforms releases, failures, and feature explorations into continuous discovery loops.
4. KPI and OKR Analysis
| KPI | What it Measures | SMART Target |
|---|---|---|
| Survey Response Rate | % of users completing in-app surveys | Increase monthly in-app survey completion rate consistently over the next 3 months from current baseline |
| Survey Completion Time | Avg. time users spend completing a feedback interaction | Keep user feedback completion time under 30 seconds for 90% of users in every release cycle |
| Time-to-Insight | Avg. time from feedback collection to usable insights | Reduce average insight generation time progressively within the first 3 months of implementation. |
| Insight Utilization Rate | % of insights that directly inform team decisions | Achieve a steady increase in insight utilization rate per sprint, starting from baseline |
| Stakeholder Adoption | # of teams regularly using platform-generated insights | Ensure regular usage of the platform by key teams (e.g., at least monthly active engagement from 3+ teams during the first 3 months) |
Objective: Smarter and scalable user insights that improve products without disrupting DANA users or stakeholders.
Key Results:
- Expand feedback coverage across high-impact user journeys
- Accelerate time-to-insight from user feedback
- Reduce manual effort in collecting and analyzing user insight
- Improve stakeholder satisfaction with research relevance and actionability
5. Solution Design: DANA Users Journey & Flow
| Stage | Action | Touch Points | Pain Points | Solution |
|---|---|---|---|---|
| Trigger Moment 😐 | Completes a meaningful action in the app (e.g. top-up, transaction, bill payment) | Action confirmation / failed / success screen | Occasionally unsure if the action went as expected, minor friction during process | A prompt asks for feedback triggered contextually and phrased in a friendly, human tone |
| Feedback Invitation 😐 | A personalized micro-survey is triggered based on context and user segment | In-app notification / pop-up after relevant user flow | Past feedback experiences felt long or didn’t seem to lead to improvements | Contextual, non-intrusive micro-survey with clear framing and optional skip |
| Interaction 🙂 | Responds to 1–2 quick questions | Embedded chat-style prompt or survey card | Doesn’t want to type or think too hard. May ignore if it feels irrelevant | Mobile-native UI (emojis, swipes, tap-to-select), fun & familiar tone |
| Follow Up 🙂 | If feedback suggests an issue or confusion, AI follows up with relevant clarifying question | Conversational follow-up, or card-style additional question | Branching logic with clear signal that user can exit any time | |
| Resolution 😀 | Later, user notices a changelog or message highlighting updates based on user feedback | Inbox message, pop-up on feature revisit, or “What’s New” banner | In traditional surveys, feedback feels like having no real impact to the app | Feedback loop closed with “Thanks to your input…” style message |
DANA users flow
- User completes a key action (e.g. transaction, feature use, Help Center visit).
- System checks if feedback should be triggered (based on frequency, segment, event type). Is user eligible? If not, the flow ends.
- System: micro-survey is shown (personalized prompt type & tone).
- User sees and interacts with micro-survey (emoji, 1-question poll, etc.), or closes the survey.
- Is follow-up needed? If yes, the system triggers adaptive follow-up and the user gives further context to their feedback.
- System logs the user feedback.
- User revisits the feature after update. Is matching feedback linked to improvement? If yes, the system shows a closure message.
6. Solution Design: Stakeholder Users Journey & Flow
| Stage | Action | Touch Points | Pain Points | Solution |
|---|---|---|---|---|
| Need for User Signal 😐 | A team needs user feedback to explore a problem, support a feature decision, or investigate an issue | User insights dashboard | Traditional research takes too long and lacks real-time relevance | Real-time dashboard with behavior-linked user feedback—no waiting, no manual digging |
| Explore Insights 😐 | Browse user quotes, sentiment trends, pain points, and relevant to their journey or feature | Feedback breakdown, sentiment trends, pain point clusters interface | Past insights lacked context or depth, often overwhelming | Smart filters, clustering, and sentiment tagging to simplify spotting patterns and user frustrations |
| Trigger Research 🙂 | Identifies a feedback gap or hypothesis and initiates a new micro-survey campaign | Create survey campaign, trigger campaign, and audience selector interface | Generic surveys couldn’t drill into specific product experiences | Stakeholders can self-launch contextual feedback flows in minutes with smart templates |
| Implement & Monitor 🙂 | Stakeholder watches how user sentiment evolves after a new feature or fix goes live | Time-based insight tracker, before vs after release view, or alerts | Previously had to wait for reports or user complaints | Platform auto-monitors user journeys in sentiment or behavior tied to product updates |
| Reflect & Plan Forward 😀 | Uses accumulated insights for quarterly planning, discovery, and roadmap shaping | Historical insight, journey score tracker, user segmentation filters | Long-term planning used to rely on guesswork or anecdotal feedback | Historical insight heatmaps, journey scores, and segmented data help teams better strategies |
Stakeholders flow
- Stakeholder opens platform to explore insights.
- System displays recent feedback breakdowns, sentiment trends, and pain points by journey/feature.
- Stakeholder applies filters (user segment, journey, feature, date).
- Should this insight be explored deeper? If yes, the stakeholder configures a new micro-survey campaign.
- Launch micro-survey campaign? If not yet, the campaign is saved as a draft for future edits. If yes, the campaign is launched and the system monitors response.
- System surfaces and saves ongoing response data & top insights, and the stakeholder monitors impact in real time.
- Has feedback and sentiment improved? If yes, flag insight as resolved and share outcome with team; if not, flag issue for further review and discussions with team.
- Stakeholder uses historical insight trends in planning.
7. Solution Design: High Level System Framework & Flow
This high-level system architecture illustrates how DANA’s AI-powered in-app user insights platform captures, processes, and delivers real-time user insights seamlessly:
- User Feedback is Triggered Contextually: Firebase events (e.g., failed transactions, onboarding) activate the Trigger Engine, which dynamically generates smart surveys via the GPT-powered Survey Generator.
- Lightweight Surveys are Delivered In-App: The Embedded Feedback SDK pushes micro-surveys to users directly within the DANA app interface, ensuring high contextual relevance and low friction.
- Responses & Events Flow into the Feedback Pipeline: Raw feedback and event logs are streamed via Pub/Sub into the User Feedback Database (PostgreSQL) and consumed by the Processing Engine (Python).
- Insights are Enriched and Analyzed: The engine powers its NLP services using spaCy for natural language parsing and structuring, MonkeyLearn for sentiment analysis, and Pinecone for semantic similarity matching and clustering of user feedback.
- Real-Time Alerts and Dashboard Updates: Processed insights are visualized in the React + Tailwind Dashboard, and critical patterns trigger notifications via Slack (API) and SendGrid (Email Alerts).
8. Product Life-Cycle Plan
To build an impactful and scalable AI-powered in-app user insights platform over the first 12 months, we need to develop a robust product lifecycle plan. This plan is structured into four key phases, each building on the progress of the previous—starting from a lean MVP and gradually evolving into a company-wide insights engine. It outlines a clear growth path from initial launch to embedding insights into team rituals, giving stakeholders visibility into what will happen, when it will happen, and why it matters at each step.
| Phase | Timeline | Core Focus | Goal |
|---|---|---|---|
| Minimum Viable Product | Months 0–3 | Trigger-based in-app micro-surveys and a basic dashboard | Launch the first version and gather early directional feedback |
| Minimum Lovable Product | Months 3–6 | Smart follow-ups, fatigue control, and response clustering | Demonstrate that insights can scale and adapt without additional burden |
| Beta/Expansion | Months 6–9 | Personalization, tone tuning, and insight scoring | Improve feedback quality and relevance for product decision-making |
| Scaling Up | Months 9–12 | Cross-functional onboarding and automated reports | Embed user insights into day-to-day planning and team rituals |
| Phase | Key Activities | How to Measure | Why it Matters |
|---|---|---|---|
| Minimum Viable Product | Design lightweight, event-based triggers · Build the first version of the dashboard · Enable basic insights segmentation · Onboard a small pilot group | Increase in survey completion rate · The team actively explores segmented insights at least 3 journeys a week | Validating both user and internal value by enabling contextual feedback |
| Minimum Lovable Product | Implement smart follow-up logic · Introduce fatigue control to prevent over-surveying the same users · Introduce automatic clustering for open-ended responses | Improved response depth and number of completion on smart follow-ups · Lower drop-off rates from over-surveyed segments | Shifting from passive data collection to active insight generation and decision support |
| Beta/Expansion | Test tone and messaging logic · Fine-tune personalization logic · Add scoring mechanisms to rank responses · Initiate structured onboarding sessions for core teams (Product, CS/CX, UX) | Increase in feedback volume per user, without drop in satisfaction or tone · Insights appear in at least 3 roadmap or sprint planning discussions | Insights become clearer and more digestible for teams’ decisions |
| Scaling Up | Introduce automated insight distribution through Slack or email · Role-based dashboards for each function · Broaden onboarding to include non-core teams (Ops, Marketing, Legal) | Reduction in ad-hoc research requests to UX research teams · Adoption across at least 3 more teams, including non-product functions | Platform becomes internal infrastructure to streamline decision-making |