TASK

Katana was a forward thinking company using machine learning way before the AI boom. Using machine learning Katana analyses up to 200 million bond pairs to show reverting dislocations with a 91% accuracy in order to reveal specific trade ideas and see broader trends earlier, faster and more precisely. Primarily aimed at the largest financial trading institutions, portfolio managers, financial analysts, execution traders, tier-1-3 sales. Meant to be a trade discovery tool alongside industry leader tool like the Bloomberg Terminal.

My main task was to make sure the user experience is as smooth as can be which was one of the main selling points and quite uncommon in the bond trading industry. We were building the next generation bond analytics platform.

I had to introduce a new tool to users that brought a new way of thinking with itself. Presenting big data through our cloud computing platform and user trust in a machine learning algorithm brought further challenges.

Execution

Workflow efficiency was at the core of my user research. I had to analyse a typical trade discovery workflow which is many times time consuming and inaccurate. Based on spreadsheets and often no direct links between tools used for information gathering analysis and modelling.

I identified the issues needed to be solved to achieve optimum user flow, a quick and easy way to get to a viable trade idea.

1. First, we had to improve the way our data can be manipulated. Filtered, sorted, searched for and the unique way it's presented in pairs (buy side - sell side) all keeping in mind the various personas. I user tested different layouts to nail the two sides problem intuitively.

2. We then improved monitoring and user-managed monitor/watchlists. This turned out to be the heart of the platform, a new way to keep track of and analyse bond portfolios. There was a seamless integration between Monitoring and Idea discovery.

3. I further improved presenting a list of bonds ideas with a new tag system which served two purposes: matching sides and visualizing filter settings.

4. I then focused on the first-user experience, onboarding, help and notification.

5. I identified persona-based task flows and made sure there is a smooth end-to-end journey for each we often reduced 6-10 clicks to just 3-7. After these improvements, we fed back these to users, presenting these task flows visually actually improved their individual performance.

6. Solving all the basics we turned towards integration for live pricing and liquidity.

7. The next step was a new way to visualize search results with aggregated metrics to be able to make quicker decisions on the quality of the presented ideas.

TOOLS USED

Figma, Adobe XD, Illustrator, Photoshop, Airtable

Webflow: I mantained the website alone with the help of Webflow which turned out to be an extremely helpful strategy to not divide focus for our front end developers.

MY ROLE

Lead User Experience Designer, strategy

Results

I am extremely pleased with the improvements and new features we added to the platform. Our data-driven approach we defined at the start resulted in a fast-paced and continuous releases. At every major stage, we evaluated our solutions with the relevant users and we followed up with them a few weeks later for further confirmation and reassurance. Along the way, we uncovered future business opportunities and aligned our strategy accordingly.