Memorable
AI
Role
Sr Product Designer
Type
SaaS · AI · Ad Tech
Scope
UX Research, UI Design, Prototyping, User Testing
Platform
Web (Desktop & Mobile)
Tools
Figma, FigJam, Balsamiq
Overview
Predicting ad performance before launch.

Memorable AI is a platform that uses artificial intelligence to predict the cognitive and emotional performance of ad creatives before a single impression is served. Users upload their creatives, run AI-powered pretests, and get actionable insights on memorability, aesthetics, engagement, and distinctiveness.

I redesigned the Creative Pretest Dashboard — the core product surface where marketers evaluate, iterate, and optimize their ads. The challenge: make complex AI data feel simple, build a faster edit-test loop, and surface recommendations users can act on immediately.

4
Cognitive dimensions scored by AI
2
Design directions explored (V1 & V2)
2
Users tested · insights validated
1
In-app editor shipped
The opportunity
Transform the pretest dashboard from a score readout into a creative optimization engine.
Understand your scores instantly · Act on AI recommendations · Edit without leaving · Re-test in seconds · Track improvements over time.
Memorable AI — Creative Pretest Dashboard on mobile
The challenge
Three friction points
Between AI output and creative action
Problem 01
Comprehension gap
Users didn't understand what cognitive scores meant or how to improve them. Memorability, Aesthetic, Engagement and Distinctiveness felt abstract — disconnected from their day-to-day decisions.
Problem 02
No clear path to action
Even when scores were understood, users had no way to act on them inside the platform. Every edit required switching to Photoshop or Canva, re-exporting, re-uploading, and re-testing.
Problem 03
Slow iteration loop
The test → edit → re-upload → re-test cycle was fragmented and slow. Users needed to iterate without context-switching between tools — and see the impact of changes instantly.
Action · 02 Design exploration
Two approaches
How users consume and act on cognitive data
V1 · Comparator
Track performance over time.

Focused on comparative data — sparkline trends, percentage changes, and an Engagement Rate bar. Designed for users who want to see how their current creative performs against previous versions and benchmarks.

The comparator layout makes score differences immediately visible, giving context to each number.

V1 Comparator view
V2 · Modular
Act on scores immediately.

Modular layout with large score numbers, color-coded metrics, expandable recommendation sections with swap options, and an AI prompt input for generating new creatives.

Optimized for users who need to understand their score at a glance and know exactly what to change to improve it.

V2 Modular cards view
Action · 03 Wireframing
Low-fidelity exploration
Mapping the full pretest flow from upload to results

I started with low-fidelity wireframes to map the complete user journey: uploading creatives, managing the ads library, the loading state while the AI runs, and viewing results with scores, heatmaps, and recommendations. This phase was about getting the information architecture right before committing to any visual direction.

Wireframe overview
Low-fidelity wireframes mapping the ads library, loading state, and results page — brand presence, cognitive scores, heatmap, and recommendations.
Action · 04 Key feature
The in-app editor
Iterate without leaving the platform
Crop, recolor and re-run the pretest in one place.

Users had to leave Memorable to edit their creatives in external tools, then re-upload and re-test. I designed an integrated editor that lets users make adjustments directly — crop, rotate, resize, and adjust colors — and immediately re-run the pretest.

The editor opens as a modal that preserves context: users see their creative alongside the pretest results, so every edit is informed by the AI feedback.

Editor capabilities
Crop with zoom, straighten, and resize sliders
Color adjustments — brightness, contrast, saturation
Rotate and reposition elements
Apply changes & re-run pretest instantly
Result · High-fidelity design
Production-ready
Scores, heatmap and copilot in one view
Result · Responsive
The same pretest, on mobile.

The full pretest experience — scores, heatmaps, recommendations, and the AI copilot — adapts to mobile so users can review creative performance on the go.

Mobile responsive view across cognitive score, results, and copilot recommendation screens
Mobile — cognitive scores, brand lift, and copilot recommendations in a scrollable, touch-friendly layout.
Validation · User testing
2 users tested
5 key insights that shaped the final direction
Insight 01
The performance comparator caught attention
Both users felt comfortable with V1's side-by-side layout — it immediately oriented them on how the creative was performing.
Insight 02
Big numbers + color coding = faster comprehension
V2's modular cards with large score numbers and color-coded metrics were praised as simpler and quicker to understand.
Insight 03
Comparative graphs are essential for small differences
When score differences are minimal, visual comparison graphs help users decode the real impact without extra effort.
Insight 04
Color-coded recommendations aid recognition
Recommendations color-matched to each metric were a hit — users instantly associated which improvements mapped to which scores.
Insight 05
The in-app editor was "fabulous"
Both users loved editing directly from the tool. Not having to switch to an external app saved significant time, and the editing quality was praised by both participants.

"Cognitive scores in V2 are simple, the big numbers help quickly to understand and having a color associated with each characteristic is very useful."

— User test participant, comparing V1 vs V2
Decision & next steps
A hybrid version
Best of V1 + best of V2 into a single direction
V1's comparator with V2's score cards.

After reviewing both approaches and the test insights, I created a hybrid final version combining V1's comparator layout and brand presence timeline with V2's modular cognitive score cards, color coding, expandable recommendations, and the AI copilot.

For the editor, my recommendation was to ship a lean V1.0 with core features, then use heatmap data to understand actual usage patterns before iterating.

Editor strategy
Ship lean, measure real behavior, then iterate.
Start with a V1.0 with core features. Deploy heatmaps to see how users actually use the editor — do they only adjust contrast? Do they resize logos but never touch the artwork? Let data guide the next version.
Heatmap
→
Iterate
→
A/B Test
→
Final Decision
Closing remarks
Learning 01
Size is not everything
Location matters, and AI can help you find the right one. Where elements sit on the canvas affects attention as much as how big they are.
Learning 02
Contrasts are your friend
Which version is better? Ask the AI. Data-driven comparisons remove subjective bias and let the creative performance speak for itself.
Learning 03
Keep it simple
Easy on the brain, delivers more gain. Reducing cognitive load in the dashboard mirrors what the platform itself recommends for ad creatives.
The core principle
The best dashboard is the one that makes complex AI feel like common sense.
Every design decision — from color-coded scores to the in-app editor — exists to close the gap between machine intelligence and human action.
↑
Score comprehension
↑
Iteration speed
↓
Tool switching
↑
Creative optimization
Memorable AI · Creative Pretest · Dashboard redesign · 2024
Sr Product Designer · Paula Elffman