selected works

/

kitravia

Kitravia

Kitravia

Kitravia

Website Redesign

Website Redesign

A website redesign of an AI-powered travel agency that lets travellers plan and book trips through a conversational interface. Three user dashboard experiences created from scratch.

A website redesign of an AI-powered travel agency that lets travellers plan and book trips through a conversational interface. Three user dashboard experiences created from scratch.

Role

UX designer on a team of 3, dashboard design lead

Timeframe

5 Weeks

Type

Mobile and desktop website redesign

Tools

Figma, Figmake, Maze, Claude Design

Project overview

Project overview
Kitravia technology asked users to trust it. The website gave them no reason to.
Kitravia technology asked users to trust it. The website gave them no reason to.

Kitravia is an AI-powered travel agency that lets travellers plan, compare, and book entire trips through a conversational interface. The ambition goes past competitors: an assistant that understands natural language, remembers preferences, and handles everything from flights to visa documentation in one place. The technology delivered on that. The website didn't. Fragmented navigation, a broken mobile experience, and no visible trust signals meant new users had little reason to believe the product worked, let alone book with it.

I worked in a three-person team across a five-week accelerated sprint, leading the research synthesis and owning the design of all the user dashboards — traveller, partner, and admin. My teammates covered the homepage, AI search, booking flow, and onboarding. The redesign gave the interface a single navigation system, a mobile experience that put the AI chat where a phone user would actually find it, and a visual language consistent enough to make the underlying intelligence believable.

THE CHALLENGE

How might we make a conversational travel platform feel credible enough to book with, before the user has said a word to it?

How might we make a conversational travel platform feel credible enough to book with, before the user has said a word to it?

Research

Research

We wanted to find out if the UX problem lay within the product or the way it was presented.

We wanted to find out if the UX problem lay within the product or the way it was presented.

We wanted to find out if the UX problem lay within the product or the way it was presented.

We audited the current website before we designed anything. I ran a heuristic evaluation of the live site against Nielsen's ten heuristics, documenting every finding, then set the results against a competitive analysis of four major OTAs. Interviews with the target audience were conducted. Finally, an updated site map was created. The audit told us what was broken. The competitive analysis told us whether fixing it would be enough to matter. The interviews told us what users really needed from the product. The site map gave us direction.

38

issues found during the website audit

6

interviews, open-ended, 30 minutes each

4

competitor products analysed

Method 01

Website Audit

NIELSEN'S TEN HEURISTICS APPLIED

38 ISSUES IDENTIFIED

Method 01
Website Audit

NIELSEN'S TEN HEURISTICS APPLIED

38 ISSUES IDENTIFIED

Method 02

Competitor Analysis

6 COMPETITORS

SWOT ANALYSIS

Method 02
Competitor Analysis

6 COMPETITORS

SWOT ANALYSIS

Method 03

Interviews

6 INTERVIEWS

OPEN-ENDED

REACTION TEST

Method 03
Interviews

6 INTERVIEWS

OPEN-ENDED

REACTION TEST

Method 04

Site Map

REVISED INFORMATION ARCHITECTURE

Method 04
Site Map

REVISED INFORMATION ARCHITECTURE

Key insights

Key insights

Booking is multi-platform by default.

Booking is multi-platform by default.

Booking is multi-platform by default.

All participants use more than one platform or information source during a booking session. Even those who described themselves as loyal to one platform (e.g. Booking.com) admitted to cross-checking hotel websites or running parallel searches. The cognitive burden of managing this fragmentation is a shared pain point.

Filters are broken for complex needs.

Filters are broken for complex needs.

Filters are broken for complex needs.

Family travellers, work travellers with specific constraints, and experiential travellers all hit the same wall: standard platform filters don't go deep enough. This forces manual review-scanning, which is time-consuming and unreliable.

AI is welcome, but trust must be earned.

AI is welcome, but trust must be earned.

AI is welcome, but trust must be earned.

Every participant expressed openness to AI-assisted travel planning. However, the level of autonomy they were comfortable granting the AI varied significantly. The common thread was: users want to remain in control of decisions, especially financial ones. AI should augment the process, not remove human agency.

First impressions of Kitravia: Promising but unpolished.

First impressions of Kitravia: Promising but unpolished.

First impressions of Kitravia: Promising but unpolished.

Participants reacted positively to the concept and the AI-centric positioning. However, the current execution has credibility gaps that prevent them from trusting it with a real booking. Several critical UX issues were identified without prompting.

Personas

Personas

Nadia

Nadia

Age: 34

Occupation: Product manager

B2C traveler

Budget: Mid-range

Travels for work + family

5-6 trips a year

BIO

Nadia plans her own trips for work and her family's holidays. She compares three or four sites every single booking — deeply skeptical that any one platform shows her the best deal. She's used ChatGPT to shortlist destinations and was impressed by the speed, but she's not about to give over banking information to an AI. She wants help filtering and curating — not deciding for her.

BEHAVIOURS

Cross-checks prices, reads reviews before booking stays, uses AI to shortlist destinations, books in advance.

CHALLENGES

Generic filters miss unique needs, AI doesn’t fully understand her needs, cross-checking tabs across sites becomes frustrating.

NEEDS FROM KITRAVIA

Multi-itinerary AI output, decision gate before payment, family/group filters, unified post-booking hub for docs.

Marcus

Marcus

Age: 47

Occupation: Hotel owner

Property size: 32 rooms

B2B partner

Location: Cape Town

Platforms: Booking, Expedia

own website

BIO

Marcus and his wife run a boutique hotel in Cape Town. He operates across Booking.com, Expedia, and his own site — and resents how much commission he pays. He wants faster review notifications, clearer revenue trends, and a way to run promotions himself without depending on a marketing agency.

BEHAVIOURS

Checks bookings and inbox daily, replies personally to all reviews, runs promotions by himself.

CHALLENGES

High OTA commissions eat margin, promotion tools are complex on current platforms, no visibility into who his guests are.

NEEDS FROM KITRAVIA

Real-time analytics, optimised review management, promotion templates.

Caroline

Caroline

Age: 29

Occupation: Customer operations lead

Responsible for Kitravia customer support

Super admin

BIO

Caroline joined Kitravia from a fintech support team. She's responsible for service-level agreements compliance, refund processing, and partner escalations. Her day swings between managing a queue and digging into edge-case bookings to figure out what went wrong. She's the calmest person in the room when something breaks— but she's tired of stitching context together from three separate tools

BEHAVIOURS

Handles all partner and customer disputes, manages refunds, reviews purchase orders.

CHALLENGES

Doesn’t have one tool for all tasks, no single view of system health, refunds need multi-system approvals.

NEEDS FROM KITRAVIA

Single tool for users, bookings, and tickets, one-click actions, live system health and ticket queue, audit trail of admin actions.

UI Kit

UI Kit

UI Kit

My role in the design

My role in the design

My role in the design

I was tasked with creating user dashboard for B2C and B2B customers and for the Kitravia admin support team.
I was tasked with creating user dashboard for B2C and B2B customers and for the Kitravia admin support team.

The B2C dashboard for mobile and desktop were prioritised for this project and will mostly be what is discussed for the remainder of this case study. Due to time and testing sample limitations, it made sense to prioritise the B2C dashboards as we could test and validate these designs. For this reason, I only created a UI design for the Partner and Admin dashboard with minimal interaction between elements. This was noted to client for future work and iterations.

User flows

User flows

Three flows were created for testing purposes. Due to time and testing sample limitations, only the B2C User Dashboards were to be tested. All three flows below are for those tests.

Three flows were created for testing purposes. Due to time and testing sample limitations, only the B2C User Dashboards were to be tested. All three flows below are for those tests.

Three flows were created for testing purposes. Due to time and testing sample limitations, only the B2C User Dashboards were to be tested. All three flows below are for those tests.

Lo-fi Designs

Lo-fi Designs

Lo-fi Designs

Low fidelity designs were created for all three dashboards. This helped prioritise content and sections that would meet each specific persona's needs from Kitravia

Low fidelity designs were created for all three dashboards. This helped prioritise content and sections that would meet each specific persona's needs from Kitravia

TRAVELER DASHBOARD (B2C)

PARTNER DASHBOARD (B2B)

ADMIN DASHBOARD

High fidelity & prototype

High fidelity & prototype

High fidelity & prototype

The traveller dashboard was turned into a function prototype while the partner and admin dashboards were designed to the highest fidelity but without any interactions created for a prototype.

The traveller dashboard was turned into a function prototype while the partner and admin dashboards were designed to the highest fidelity but without any interactions created for a prototype.
Traveller Dashboard:
Traveller Dashboard:

Dashboard overview

Dashboard overview

Traveller trips

Traveller trips

Traveller guides

Traveller guides

Traveller profile

Traveller profile

Partner Dashboard:
Partner Dashboard:

Dashboard overview

Dashboard overview

Listing management

Listing management

Promotion center

Promotion center

Admin Dashboard:
Admin Dashboard:

Dashboard overview

Dashboard overview

Ticket view

Ticket view

Usability testing

Usability testing

Unmoderated remote sessions with three participants, covering all three task flows. Users were given a scenario and then asked to move through each flow. Follow up questions helped me understand their process and gather feedback.

Success was measured by time of completion, item discoverability, user confidence and trust, and desirability of the dashboard.

Unmoderated remote sessions with three participants, covering all three task flows. Users were given a scenario and then asked to move through each flow. Follow up questions helped me understand their process and gather feedback.

Success was measured by time of completion, item discoverability, user confidence and trust, and desirability of the dashboard.

Task 1: Find return flight number

Task 1: Find return flight number

Scenario

Scenario

You're packing for your Cape Town trip. Your partner wants to know your flight number for your return journey. Find it on the user dashboard.

Task

Task

Navigate to where you will find a flight number for your return journey from Cape Town to London.

Success rate

Success rate

100%

Avg time on task

Avg time on task

384 seconds

Ease of use

Ease of use

Mixed reviews. No users found it particularly challenging but one rated ease of use 2/5.


‘After clicking the "My trip" button, I easily found my return trip information on the page page with both departure and returns.’

Task 2: Find the first item in a guide

Task 2: Find the first item in a guide

Scenario

Scenario

Before your Tokyo trip, you saved a travel guide called Tokyo Explorer. You want to remind yourself what the first thing on the itinerary is.

Task

Task

Find your first itinerary item in your saved guide - Tokyo Explorer.

Success rate

Success rate

100%

Avg time on task

Avg time on task

31 seconds

Ease of use

Ease of use

All users found this task easy. 2 out of 3 participants rated it a 5/5 in ease of use.

Task 3: Navigate to subscription settings

Task 3: Navigate to subscription settings

Scenario

Scenario

You'd like to upgrade your Kitravia subscription plan. Navigate to where you would do that.

Task

Task

Find the subscription settings in the user dashboard.

Success rate

Success rate

100%

Avg time on task

Avg time on task

113 seconds

Ease of use

Ease of use

All users found this task easy. 2 out of 3 participants rated it a 5/5 in ease of use.

Overall experience

Overall experience

Users were asked a series of questions about universal dashboard components and their overall experience.

Was there a point where you weren't sure if you were in the right place?

Was there a point where you weren't sure if you were in the right place?

All users were sure they were in the right place at all times.


‘No, it was simple to follow.’

Was the dashboard easy to navigate?

Was the dashboard easy to navigate?

All users found the left navigation menu to be sufficient in navigating around the dashboard.


‘The main dashboard is clear and concise along the directions.’

Is there anything you expected to find on the main dashboard itself that wasn't there?

Is there anything you expected to find on the main dashboard itself that wasn't there?

Overall no users found the dashboard to be lacking anything.

On a scale of 1–5, how confident would you feel using this dashboard for a real trip?

On a scale of 1–5, how confident would you feel using this dashboard for a real trip?

67% rated it five and 33% rated it 4.

Iterations

Iterations

Iterations

Most changes were made by client request as the usability testing proved highly successful and required no major changes.

DASHBOARD OVERVIEW

An AI search bar that is used on the home page of the website was added by client request. This allows users to plan a new trip and gives a constant reminder of Kitravia’s main feature.






The ‘My trips’ card colours were updated.

BEFORE

AFTER

MY TRIPS

The AI search bar was also added here by client request. Trying to keep the main function of Kitravia front of mind as much as possible.

BEFORE

AFTER

MY GUIDES

The design was updated to match the design one of my team mates did on the home page flow.

BEFORE

AFTER

Reflections & take aways

Reflections & take aways
The hardest thing about the five week sprint wasn’t deciding what to design, it was deciding what to leave out.
A feature is only as good as the intent
behind it.

The volume of the design brief was daunting at first — three dashboard across different user groups in five weeks. This meant I had to prioritise and be realistic about what I can design, test, and prove in that time. Early on I decided to put the focus on the Traveller Dashboard, because it was the only one I could realistically test with the participants available to me. That meant the admin and partner dashboards would take a backseat and only get a UI design to match the overall design system. I raised this with the client early on got agreement rather than just assuming and heading off in my own direction.

One lesson that carried of a lot weight for me is how important a shared design system is in a team where everyone is working on a different feature. Working from fixed tokens — colour, type scale, radius, spacing — meant all three team mates designed cohesive features for the client. It also helped me build three dashboards, made for entirely different users, still read as one product.

FUTURE ITERATIONS

Usability testing for the Partner Dashboard with real hotel owners

The remaining Super Admin pages built from the existing component set.

Error state variants built across trip detail and booking management

A WCAG 2.1 AA audit before sign-off.

KEY INSIGHT

A five-week sprint doesn't reward fitting everything in. It rewards choosing what to cut, agreeing that cut with the client, and documenting it clearly enough that the next person loses no time picking it up.