01 EYECUE / CAMPAIGN CREATIONPRODUCT DESIGN
PROJECT TYPE: PRODUCT DESIGN
ROLE: PRODUCT + UX/UI DESIGNER
TIMELINE: 12 WEEKS
TOOLS: FIGMA, MACHINE LEARNING
Monitor and compare specific marketing strategies to see where your global campaign assets are being used worldwide.
THE CHALLENGE
Global marketing teams lacked an efficient way to track how their campaign assets were being used and adapted across local markets.
Existing tools made it difficult to compare global versus locally created content, often requiring manual searches and fragmented reporting, generating visibility gaps, slowed down decision-making, and made it harder for teams to understand which strategies performed best across regions.
There was a clear need for a centralized solution that could automatically detect asset usage, measure performance, and provide actionable insights to optimize global campaigns at scale.
“We see that our local content at times out performs our global content assets in product campaigns and we need and want to know why”
Head of Marketing of beauty company
GOALS + OBJECTIVES
The primary goal of the Campaign Creation feature was to give brands key insights and performance metrics of global and local content. Our key objectives included:
- Enhance cross-market visibility: Enable users to upload global assets and select multiple owned accounts for cross-comparison, providing deep insights into how global assets perform across both global and local markets. This strengthens the ability to track and optimize content strategies in diverse regions.
-
Streamline workflows: Provide an intuitive, AI-powered tool that reduces the need for manual tracking, making the process faster and more efficient.
-
Optimize internal reporting: Use AI to automatically generate insights in the form of a PDF, which can be easily shared across teams. This helps shorten the time spent creating presentations and reduces delays in transferring successful marketing strategies.
-
Automate asset identification: When users upload their assets, the AI tool automatically distinguishes between global and local content, ensuring an accurate comparison and eliminating the need for manual categorization.
CONSTRAINS
- Time: We were working under a rapid turnaround timeline. As a startup with a small client roster, we had to prioritize immediate client needs and deliver within tight deadlines. This meant there was limited time for formal discovery, in-depth research, or structured usability testing before launch. The trade-off was speed over depth. While this allowed us to validate value quickly and maintain strong client relationships, it also introduced risk — particularly around uncovering edge cases, testing assumptions at scale, and ensuring long-term usability. To mitigate this, we adopted a lean, iterative approach: shipping early versions, gathering real-time feedback from users, and continuously refining the feature post-launch. Rather than a traditional research phase, validation became embedded in the rollout process.
-
Small team: Our design team consisted of only two designers: the Lead Product Designer and myself. With limited design bandwidth, we had to be highly intentional about prioritization, scope definition, and decision-making. The trade-off was breadth over specialization. We were responsible for the full design lifecycle — from discovery and UX strategy to interaction design and stakeholder alignment — which required us to move quickly and make decisions with incomplete information. To manage this, we maintained close collaboration, divided ownership strategically, and focused on delivering high-impact features rather than pursuing exploratory or lower-priority enhancements.
-
Dashboard Information Architecture: Our dashboard lacked congruent information architecture and globally used terminology. Our dashboard had multiple pages, each with nested sub-pages, making it difficult for users to orient themselves and find what they needed quickly.
RESEARCH + DISCOVERY
Our approach to feature development had to be flexible. With a fast-paced roadmap and limited resources, we adapted the typical research-to-design pipeline into a more iterative, conversation-driven process. Instead of formal studies, real-time feedback from our clients guided our decisions (with the expection of heuristic evaluation!), allowing us to refine the feature quickly while staying grounded in actual user needs.
HEURISTIC EVALUATION
Given the tight timeline and the urgency of the client’s request, Sofia and I conducted a heuristic evaluation to quickly assess the usability of our initial concept. Using Jakob Nielsen’s 10 usability heuristics as our framework, we were able to identify early usability gaps and clarify which areas required immediate improvement. This approach gave us a structured, efficient way to validate our first iteration before moving deeper into design.
We had 4 participants do the Heuristic Evaluation the takeaways were:
Takeaways:
Visibility of System Status
Users had no clear feedback when assets were being processed or when the AI was actively analyzing content. There were no loading indicators, progress states, or confirmation messages to signal that the system had registered an action, such as uploading their images. This left users uncertain about whether their uploads had succeeded or if the tool was working at all.
What we did to solve it: We introduced progress indicators during asset upload and AI processing, added confirmation states once actions were completed, and included clear empty states when no data was available yet — giving users continuous, honest feedback about where they were in the process.
Consistency and Standards
Terminology and interaction patterns varied across the dashboard. Labels for the same actions differed between pages, and certain UI components behaved differently depending on where users encountered them. This inconsistency created confusion and made the learning curve steeper than it needed to be.
What we did to solve it: We audited the language used across all dashboard pages and standardized terminology to align with widely recognized conventions. We also aligned interaction patterns so that similar components behaved consistently throughout the product, reducing cognitive load and making the experience feel more cohesive.
Match Between the System and the Real World
Some labels and system messages used internal or technical language that didn't reflect how marketing teams actually talked about their work. This created a disconnect between the tool and the users' mental models.
What we did to solve it: We rewrote labels, headings, and system messages using language that mirrored the vocabulary of marketing professionals — replacing technical jargon with familiar, intuitive terms.
User Control and Freedom
Users had no way to remove an uploaded asset without discarding the entire campaign and starting over. This created unnecessary friction, especially when users made upload mistakes or needed to adjust their selections mid-flow. It also made users feel discouraged.
What we did to solve it: We added the ability to remove individual uploaded assets directly within the campaign creation flow, giving users a clear and accessible exit without forcing them to restart.
Error Prevention
The interface offered little guidance to prevent users from making mistakes during setup — such as selecting incompatible account combinations or uploading unsupported file types — before errors actually occurred.
What we did to solve it: We introduced inline validation and contextual guidance at key decision points in the flow to help users avoid errors before they happened, rather than surfacing them after the fact.
Recognition Rather than Recall
Users had to remember which accounts they had selected across steps, with no persistent visual indicator to reference. This forced users to rely on memory rather than recognition, adding unnecessary mental load.
What we did to solve it: We added a persistent accounts-selected indicator that remained visible throughout the campaign creation flow, so users could always see their selections at a glance without needing to backtrack.
Flexibility and Efficiency of Use
The flow didn't accommodate users who were returning to an existing campaign or wanted to make quick edits without going through the full creation process again.
What we did to solve it: We refined the flow to allow users to revisit and adjust campaign settings more efficiently, reducing the steps required for common repeat actions.
Help Users Recognize, Diagnose, and Recover from Errors
Error messages used internal, unclear language that didn't tell users what had gone wrong or how to fix it. Users were left to interpret vague alerts on their own.
What we did to solve it: We rewrote all error messages to use plain, actionable language — clearly describing what went wrong and providing a direct next step so users could recover quickly and with confidence.
Help and Documentation
There were no tooltips, onboarding cues, or first-time user guidance anywhere in the feature. New users had no scaffolding to help them understand the tool's capabilities or how to complete key tasks.
What we did to solve it: We designed a set of contextual tooltips for key interactions and introduced a lightweight first-time user experience that oriented new users to the feature without overwhelming them. This was prioritized as a high-impact, low-friction addition that significantly reduced confusion during onboarding.
OBSERVATION
Through recorded user sessions, we observed that users spent considerable time in Eyecue’s dashboard primarily analyzing performance tables for individual posts and manually extracting data to compare how posts performed in local accounts vs global accounts. Often times going back and forth from screen to screen, slowing insight finding.
INTERVIEWS
Our learning came directly from the people using the product. Throughout the design process, we met regularly with our clients’ marketing teams, the same teams who would eventually rely on this feature in their daily work.
These conversations quickly became our most valuable form of research. Each meeting revealed a little more about how global teams operated, where their workflows broke down, and how much time they spent manually tracking asset usage across regions. They often described the frustration of piecing together performance data from multiple sources, never having a single place to compare global and local campaign results.
Their feedback shaped every iteration of the design. We used their insights to refine the flow, simplify interactions, and focus on what mattered most: visibility and efficiency. Even without a traditional research setup, this ongoing dialogue allowed us to design with real users in mind, grounding every decision in their challenges and goals.
Issues surfaced through interviews and sessions included:
- Lack of hover states on uploaded asset thumbnails
- No indicator showing which accounts had been selected
- No tooltips or first-time user tutorial
- No way to remove an individual uploaded asset — users had to create an entirely new campaign to make changes
KEY INSIGHTS
User pain points:
- Users spent excessive time reviewing performance tables for individual posts, instead of having a streamlined way to access the information they needed quickly.
- Data extraction for each post was done manually, which added unnecessary complexity and slowed down the decision-making process.
- Users manually compared the performance of individual posts, requiring extra effort to generate actionable insights on the success of campaigns.
- Users struggled to visualize performance metrics across multiple posts simultaneously and had to go back and forth between screens to review the metrics for each post separately, making cross-comparing posts difficult and time-consuming.
OUTCOMES
PDF EXPORT
After launching the first iteration, we gathered client feedback from the marketing teams using the feature. While the insights were valuable and gave global teams strong visibility into asset performance across regions, one friction point remained: the insights weren't easily shareable.
Teams still had to manually translate dashboard data into presentations for internal stakeholders, adding time and effort to the reporting process.
In collaboration with engineering, we introduced a one-click PDF export that automatically generated a presentation-ready report. The feature synthesized key metrics, global vs. local comparisons, and performance highlights into a structured, branded document.
This reduced manual reporting time and enabled faster alignment across teams, extending the value of Campaign Creation beyond analysis into automated reporting.
PROTOTYPE / DEMO
MY THOUGHTS
Through this process Sofia and I spent a lot of time brainstorming how we could help our clients quickly get insights on their campaigns. The more we heard their feeback and saw how they interated with the feature we realized we had yet to shorten the time to insight. Along with our Machine Learning Team we