The Problem

A regional cruise line's internal knowledge base was essentially broken for its ~850 crew members spread across multiple ships. Exact-match search returned nothing useful. Navigation required 5+ clicks to reach a single document. Crew abandoned the system entirely and called the home office instead - a significant time sink for everyone.

For a head chef needing allergen procedures during a guest emergency at 6 AM, a knowledge base that fails isn't just frustrating - it's a liability.

Research & Discovery

Since crew were spread across ships, I got creative with my research approach:

• Shadowed crew on two ships during active shifts - galley, deck, guest services
• Remote interviews via video with crew from both vessels
• Analyzed 3 months of support ticket data to identify the most-searched (and most-failed) queries
• Card-sorting exercises to understand how crew mentally organized information

The Solution

I rebuilt the portal around four core upgrades: smarter search, role-based dashboards, mobile-first design, and a dramatically simplified information architecture.

Search That Actually Works

Replaced exact-match search with synonym mapping and a controlled vocabulary of tags. Searching "turndown" now finds "Daily Cabin Turndown Protocol" even if those words don't appear in the title.

Role-Based Dashboards

Each role sees what's actually relevant to their job. Galley staff see food safety and allergen protocols. Deck officers see navigation and safety drills. The homepage became genuinely useful.

Mobile-First Design

Built the entire interface mobile-first since crew access information on phones and tablets during active shifts. Added offline mode, compressed assets for satellite-speed internet. Load time: 7 seconds ? under 2.

Content Strategy

The interface was only half the problem. I rewrote ~150 critical documents with plain language principles, cutting reading level from graduate-school to clear and scannable. Added numbered steps, visual warning indicators, and quick-reference summaries.

Challenges

Results

The Problem

The Derma Loft is a high-end medical spa with a loyal in-person clientele - and a digital booking experience that was driving people away. Their mobile app had a 38% checkout abandonment rate. Most common review complaint: "confusing," "takes too long," "I just call instead."

For a luxury brand whose in-person experience is meticulously crafted, the app felt like an entirely different business. The UX mismatch was hurting both revenue and brand perception.

Research

• Session recording review: identified exact screens where 62% of abandonments happened (service selection + add-on screens)
• 5 user interviews with existing clients about expectations for a luxury spa booking experience
• Competitive audit of 8 luxury wellness brands (Heyday, Milk + Honey, Hand & Stone, etc.)
• Heuristic evaluation against Nielsen's 10 usability principles

What I Redesigned

Service Descriptions That Actually Help

Progressive Add-On Flow

Moved from a checkbox wall to one add-on per screen. Framing shifted from "Would you like to add?" to "Pair it with..." - add-on uptake increased 23%.

Persistent Booking Summary

Added a persistent summary drawer at the bottom of every booking step. No more surprise at final checkout. This single change drove measurable reduction in late-stage abandonment.

Results

Key Takeaway: This project is a reminder that content design isn't cosmetic - it's structural. The visual design was already clean. The copy was broken. Every service description rewrite was a UX intervention. The 47% increase came from words, not pixels.

What This Is

EverAlice AI is a creator workflow product built for Etsy sellers, digital product creators, and print-on-demand entrepreneurs who are tired of jumping between too many tools. The product story is straightforward: upload or generate product art once, then move through listing copy, mockups, exports, books, and delivery packaging in a workflow that actually stays connected.

The Core Insight

The real problem is not that creators lack ideas. It is that the workflow after the idea is fragmented. Listing copy lives in one place, mockups in another, keyword research somewhere else, books in a separate lane, and delivery packaging gets treated like an afterthought. EverAlice AI makes the strongest case when it closes those gaps instead of asking creators to stitch the whole business together manually.

What I Designed

1. A Workflow-First Product Narrative

The clearest positioning move is to frame EverAlice AI around outcomes, not isolated features. Publicly, that means showing how Magic Lister, Book Builder, and AI Copywriter fit into one creator workflow rather than reading like three unrelated AI utilities.

2. Multi-Platform Support Without the Sprawl

EverAlice AI currently supports marketplace-specific listing output for Etsy, Shopify, Gumroad, Creative Market, Amazon KDP, and more. That matters because a creator selling across platforms does not want to rebuild the same listing logic ten times. One of the product's strongest promises is that the workflow stays coherent while the output adapts.

3. Delivery and Packaging Are Part of the Story

A lot of competitors stop at SEO or mockups. EverAlice AI pushes further by treating delivery packaging and ZIP prep as part of the product workflow. That is a smart differentiator because it respects the handoff the customer actually experiences after purchase.

Product Snapshot

If you want to see the full product positioning, the best entry points are the main EverAlice AI site, the press page, and the core feature pages for Magic Lister, Book Builder, and AI Copywriter.

The Problem

A SaaS tool with strong feature depth was losing users in the first 7 days - not because the product was bad, but because users couldn't figure out the value fast enough. The onboarding flow was a 12-step checklist that nobody finished. In-app guidance was tooltip text that explained what buttons did, not why users would care.

Secondary problem: with a two-week release cycle, content couldn't keep up. Every new feature shipped with placeholder microcopy. I needed to build a system, not just write better copy.

Research & Discovery

• Analyzed drop-off data - 68% of users abandoned at step 4 of 12
• Reviewed 6 months of support tickets to identify most-asked "how do I" questions
• Interviewed 8 users who completed onboarding and 6 who churned in the first week
• Audited competitor onboarding flows to identify patterns and gaps

The Content System I Built

1. Goal-Based Onboarding Architecture

Redesigned the onboarding flow around user goals, not feature checklists. Instead of "Complete your profile ? Connect integrations ? Invite team," the new structure was: "What do you want to achieve first?" - three paths based on primary goal, each reaching core value in 3 steps or fewer.

2. Persona-Responsive Microcopy Framework

Built a content framework with four persona variants for key onboarding touchpoints. The system used 3 signals (role, team size, primary goal) to serve contextually appropriate guidance. Same interface, four different content experiences.

3. AI-Assisted Drafting Workflow

Built a prompt library and content templates that let me generate first drafts for new feature guidance in ~40% of the normal time. The system included: a voice and tone prompt, persona context blocks, and a QC checklist I ran every draft through before publishing. AI handled volume; I handled quality.

4. Content Governance for a Fast-Moving Roadmap

Created a content review protocol integrated into the engineering sprint process. Every new feature shipped with: reviewed microcopy, in-app guidance, and a help article stub. No more placeholder copy going to production.

Results

This project demonstrates what AI-assisted content design actually looks like in practice - not AI replacing UX writers, but UX writers building systems that use AI responsibly. The prompt library, QC checklist, and governance protocol outlasted my engagement and continued producing quality content after I left.