/
/
Qlik
Qlik · Lead Product Designer · Sole Designer, Pre & Post Acquisition
increase in learner engagement
(target +10%)
Reduced bounce on the logged-out homepage and increased entry into certifications and learning paths.
program bookings
Exceeded $500K target, driven by increased certification enrollment.
certifications completed
Surpassed 3K target through improved progression clarity.
academic program growth
Exceeded 95% target as structured paths increased program visibility.
hands-on cloud training
More than doubled the 8-hour target through labs integration.
CONTEXT
I had been part of Talend's learning team since 2021 and continued as the sole designer through the acquisition, carrying product context across both systems. The question was whether I'd use that knowledge to patch the old system or redesign around what learners actually needed.
THE REAL PROBLEM
The two platforms merged without a shared architecture. Navigation was inconsistent, discovery was driven by system structure rather than learner goals, and there was no clear entry point for first-time users.
HOW I DIAGNOSED THIS
I mapped every surface across both platforms before touching any designs. The problems weren't hidden — they were structural. Every decision I made afterward had to function within the system as it existed, not as I wanted it to be.
PLATFORM CONSTRAINTS
Template-locked layout. No reliable progress states on custom content. System-controlled hierarchy. Limited onboarding affordances. Catalog structure driven by LMS, not learner intent.
WHAT I COULD HAVE DONE
Pushed engineering to build workarounds for progress tracking on custom pages. Advocated for a platform migration. Designed the ideal experience and handed the constraints problem to someone else.
WHAT I CHOSE AND WHY
I designed toward where the LMS was reliable, not where I wished it would be. That constraint became a structural principle: every design decision had to function within the system as it existed.
TEMPLATE-LOCKED LAYOUT
No ability to customize section order or hierarchy.
NO RELIABLE PROGRESS STATES
Custom content couldn't surface completion data.
LIMITED ONBOARDING
No native flows to guide first-time learners.
SYSTEM-DEFINED CATEGORIES
Organized around LMS structure, not learner intent.
NO FILTERING LOGIC
Learners couldn't narrow by role, product, or format.
CATALOG DRIVEN BY PLATFORM
Content buried with no prioritization or guidance.
Every design decision had to function within the system as it existed, not as I wanted it to be.
Before: every design decision fought the platform.
After: every design decision worked within it.
THE DECISION
Stakeholders pushed for product-forward navigation because it mapped to how the platform was built. I pushed for learner-led because testing showed learners browsed by topic, not by product. I won that argument with behavior, not opinion.
WHAT I GAVE UP
Product-forward navigation meant content teams could update it without design involvement. Choosing learner-led meant we took on more IA responsibility long term. That was the right tradeoff.
Navbar Variant A: Product-forward navigation structured around LMS categories. Stakeholder testing revealed learners felt disoriented without a clear starting point.
Navbar Variant B: Learner-led navigation built around the Explore dropdown and catalog. Testing showed learners preferred browsing by topic over searching, only using search when looking for something specific.
ABOVE-THE-FOLD HIERARCHY
Designed around learner outcomes, not product structure.
CORE LEARNING ENTRY POINTS
Grouped by intent to reduce cognitive load.
MODULAR SECTIONS
Designed to support future initiatives (e.g. AI Academy).
Before: navigation organized around how the platform was built.
After: navigation organized around what learners were trying to do.
Onboarding (0–10 days) — Reduce anxiety, show first steps, and introduce guided starter content.
Adopt (momentum building) — Reinforce habit and relevance while surfacing personalized next actions.
Discover (active learners) — Support deeper exploration and clarify paths and progression.
Mastery — Align certifications to readiness and signal proof of expertise.
THE INSIGHT THAT CHANGED EVERYTHING
Solving entry clarity wasn't enough. The logged-in experience needed to prioritize guidance over available content, surface the right next action at the right time, and support learner maturity through structured onboarding.
THE CONSTRAINT THAT SHAPED IT
Course completion couldn't be surfaced in custom sections. Progress was only visible in Intellum's native Continue Learning section. I designed flows to guide learners there rather than fighting the platform.
WHAT I DEPRIORITIZED
A simpler homepage would have been faster to ship and easier to maintain. I deprioritized that because the data was clear: generic experiences were failing returning learners. Simple wasn't the right answer.
HOW I TESTED THE DECISION
I ran 10+ moderated sessions across first-time and returning learners. Findings directly informed the 4-stage model and the onboarding grouping strategy.
Before: one logged-in homepage showing whatever content was available, regardless of where the learner was in their journey.
After: a system that adapted to the learner's stage — different entry points, different content, different guidance at every phase.
NO FILTERING OR SORTING
Content surfaced without prioritization or guidance.
NO FREE VS SUBSCRIPTION DISTINCTION
Learners couldn't tell what required a subscription.
VOLUME WITHOUT HIERARCHY
Increasing content made discovery progressively harder.
STRUCTURED FILTERING
Filter by price, product, and role.
FREE VS SUBSCRIPTION CLEARLY LABELLED
Learners know what they can access before clicking.
PRODUCT-BASED GROUPING
Content organized around learner intent, not LMS category.
THE PROBLEM
No filtering or sorting controls. No distinction between free and subscription content. No product-level organization. Increasing content volume made discovery progressively harder.
WHAT I COULD HAVE DONE
Accepted the native catalog. Pushed engineering to rebuild the LMS catalog from scratch. Advocated for a platform migration and handed the problem to someone else.
WHAT I BUILT INSTEAD
A custom discovery layer in HTML and CSS only — no backend access, no platform rebuild. Structured filtering, product-based grouping, and free vs subscription segmentation, all inside LMS constraints. Every other Intellum customer had accepted the native experience. We didn't.
WHAT CHANGED BECAUSE I WAS IN THE ROOM
Navigation shifted from product-forward to learner-led after I presented behavioral data. Intellum escalated to a new CSM and added a feature request to their roadmap. A design limitation became a product roadmap item.
Before: a content dump with no hierarchy, no filtering, no way to know where to start.
After: a custom discovery layer built within LMS constraints, organized around learner intent.
WHAT I ESCALATED
Progress tracking was only available in Intellum's native sections, not on custom pages. Rather than building around it, I flagged it as a structural blocker — which led to a new CSM relationship and a feature request on their roadmap.
WHAT I PUSHED BACK ON
Our director suggested benchmarking against other Intellum customers. I found a fundamental difference: no one else was building an adaptive learner lifecycle. Their logged-out and logged-in homepages were identical. There was no benchmark to follow. We were building something that hadn't been done on this platform before.
WHAT CHANGED BECAUSE I WAS IN THE ROOM
Navigation direction shifted from product-forward to learner-led after I presented behavioral testing data. Intellum escalation resulted in a new CSM and a feature request now on their roadmap. The 4-stage learner model was adopted as the framework for all future logged-in experience decisions.
This project was designed before AI generation was a meaningful part of my design workflow. After key decisions shipped, I ran experiments with two tools to see where AI could accelerate the work — and where it couldn't.
FIGMA MAKE — ITERATING FASTER ON LAYOUT
The AI Academy needed to feel distinct from the rest of Qlik Learning. I prompted Figma Make with my existing design, color system, and learning path logic.
What it got right: content format, hero structure, course row with progress states.
What I changed: stripped unsupported progress logic, restructured the step sequence, and pushed the aesthetic further from standard Qlik Learning pages.
Initial design, Figma only
After Figma Make iterations
Final version after my edits
LOVABLE — ADVOCATING FOR THE GAP
Mobile was deprioritized at launch because learner data showed no meaningful mobile usage. I pushed back on one exception: the logged-out page. That surface is a marketing page — the people seeing it on mobile are prospects, not existing users. I made the case, then built it myself using Lovable, brought it into Figma via Codia, and fixed the hero treatment. The whole process took a fraction of the time it would have taken from scratch.
THE HONEST TAKEAWAY
Both tools accelerated layout work significantly. Neither replaced judgment — knowing what to strip out of the Figma Make output, knowing which surface was worth fighting for on mobile, knowing how to adapt a generated layout to match a real design system. The gap between what AI generates and what ships is exactly where design judgment lives.
These outcomes weren't the result of more resources or a platform rebuild. They came from designing the right system within constraints that already existed.
increase in learner engagement
(target +10%)
Reduced bounce on the logged-out homepage and increased entry into certifications and learning paths.
program bookings
Exceeded $500K target, driven by increased certification enrollment.
certifications completed
Surpassed 3K target through improved progression clarity.
academic program growth
Exceeded 95% target as structured paths increased program visibility.
hands-on cloud training
More than doubled the 8-hour target through labs integration.
WHAT I WOULD HAVE MEASURED
What I would have measured: activation rate among new learners completing their first module within 10 days, stage progression rate from Discover to Onboard to Adopt, catalog search-to-enrollment conversion by filter type, return visit rate within 30 days per learner stage, and certification attempt rate among learners who reach Mastery.
REFLECTION
This work transformed Qlik Learning from a collection of LMS pages into a scalable learning system. Designing within platform constraints rather than around them created a structure that improved learner clarity while enabling long-term content growth without increasing UX or engineering overhead.
Longitudinal research. I designed the 4-stage model from behavioral testing data and synthesis, but never got to observe learners evolving through the stages over months. I'd want to validate stage transitions with real usage data before committing the framework to future product decisions.
Personalization. The logged-in homepage adapts by stage, but within each stage every learner sees the same content. The next layer would use role, product, and prior activity to surface genuinely relevant content rather than broadly relevant content.
Cross-functional handoff. I was the sole designer, which meant I held context that never got documented. Investing in decision documentation earlier would have made the system easier to hand off and extend.
