Analytics Platform at Voxxify
During my internship at Voxxify, I was the UI/UX designer working on a live B2B analytics platform.
I designed 5 product experiences and redesigned 6 core features, working from early requirements through to prototypes, developer handoff and implementation support.
Some details have been left out for confidentiality.
Duration
September 2024 - August 2025
Role
UI/UX Designer
Company
Voxxify
Design
UI, interaction design, data visualisation, Figma prototypes, design systems.
Product
Backlog management, sprint planning, agile sprint retrospectives, developer collaboration.
01 / AI INSIGHTS TRENDING
The original view showed recurring topics from each survey round, but everything looked equally important. It was hard to know where to start, and the AI summaries felt disconnected from the employee feedback behind them.
I designed a ranked view that gives the most recurring topics more prominence. Users can filter the results, open a topic, and read the comments that contributed to it.
The ranking idea came from leaderboards in games and apps like Duolingo. I liked that it gave the page an obvious starting point, while still letting users investigate the results for themselves.
02 / PERSONA HEATMAP
I needed to make sentiment easy to compare, but I also wanted to show how many responses were behind each result. A strong score from a small group should not carry the same weight as one supported by hundreds of responses.
I explored a few chart ideas that worked nicely with tidy sample data, then became much less convincing once I tried them with real data. I eventually moved towards a heatmap where colour represents sentiment and cell size represents response volume.
This made the main patterns easy to spot, while giving users a better sense of how much evidence supported each result.
03 / X-GRID
Early versions used red and green to show how results had moved between survey rounds. It was easy to read, but it also made every movement look like a clear win or loss.
The meaning of a change depends on where the score started, where it ended, and what was happening around it. I rebuilt the view using a fixed scale and made the movement between rounds visible without depending entirely on colour.
The fixed scale gives users a consistent reference point, and the extra visual cues make the view easier to read for people with colour-vision differences. More importantly, the interface shows what changed without deciding how the user should feel about it.
04 / MOJIMAP
There was no quick way to get a feel for employee sentiment across the platform. MojiMap started as a more approachable way into that information, but I did not want it to become a page of emojis without much meaning behind them.
I used scale to show how much feedback sat behind each emotion, then added the related employee comments to the same view.
This keeps the overview playful and easy to scan, while still letting users answer the more useful question: why are people feeling this way?
05 / ACTION PLAN
The platform could show teams where something needed attention, but the actual planning usually happened somewhere else. Users had to move between tools and keep returning to the survey results for context.
I designed an action-planning flow that keeps the original insight visible while a team decides what to do next. Users can add owners, priorities, statuses, and targets, with support for both numerical goals and more qualitative measures of progress.
I also explored how the product’s AI could suggest a starting point using the survey context. The suggestion is editable and only gets added if the user chooses it. I wanted it to help with the blank-page problem without taking control away from the team.
Keeping the insight and the action together makes the reason behind the work much easier to follow.
process
The evidence came in throughout the work: the existing product, customer conversations, support feedback, survey data, internal reviews and questions raised during development.
For some features, I used customer conversations to understand how teams currently recorded issues and followed up on them. For data visualisations, the most useful test was often putting real data into the design. Values that looked perfectly reasonable in a mock-up could behave very differently once long labels and uneven scores were introduced.
Formal usability testing and post-launch analytics were not part of my role at the time. Designs were reviewed throughout development and signed off by team leaders once they met the agreed completion criteria, but I don’t treat that as proof that every decision worked for users. Testing those assumptions would be the key next step.
Nearly every feature included an interactive prototype so the team could review the behaviour, not just a collection of static screens.
For development, I documented component measurements, states, interaction rules, conditional behaviour and edge cases. I then met directly with engineers to talk through constraints and reviewed the implemented work when questions came up.
The harsh reality was that the polished screen was only one part of the job. Redlining and documentation were gruelling, but working towards a more established design system streamlined the process massively. And, of course, special shoutout to the various Figma plugins that made the whole thing a little less painful.
REFLECTION
At the start of the placement, I was mostly thinking about whether a screen worked and looked finished. By the end, I was thinking about what was going into it, how it would behave with real data, and whether someone else could actually build it.
I also learned how much a good system can save you. Rebuilding a similar page in hours instead of weeks was deeply satisfying. Not quite as impressive now that vibe coding enables much quicker iterations, but very exciting at the time.
If I had continued the work, I'd have liked to test the main workflows with end users. I’d also have taken the accessibility work further with more structured testing.
More than anything, Voxxify showed me the kind of designer I want to be: someone who can dig into how people understand information, then turn that into a meaningful product. Also that ideally, if someone spends ten minutes looking at something I designed, I want it to be because there’s something worth noticing, not because they’re still trying to figure out what it means.