Challenge

Research

What issue this project was trying to solve

Commercial real estate traditionally made investments of millions based on subjective opinions. One broker may value a building 10% off from the next broker. This project aimed to equip them with objective reasons with machine learning APIs and make better investment decisions.

Finding the common grounds

The main goal is to know what to offer and how to offer the data. We have a list of 200+ datasets, 5 APIs, but not enough knowledge to know how and whom to deliver to. I mapped out the existing workflow, pain & gain points of users working in different fields and sectors to find the best product-market fit.

Key activities

• Analyze traditional information, touchpoints and their workflow along with pain/gain points.

• Gather user insights from stakeholders’ conversations with top tier clients and align expectations and priorities.

• Understand the data, source and model behavior of our datasets

• Get insights from reports and analytical companies to understand what data type and delivery methods is useful.

Key Insights

Most users have an assumption of what the value should be and an intuition on their impact on value.

The biggest pain is the long discussions in value drivers due to the subjective nature of the industry.

On first glance all a user want to know is if there are many serious crimes in the scene, then more detailed information on demand.

Most users consider AI an “objective source for value drivers”, however the AI must prove its accuracy first.

Value drivers are much more valuable when enriched and showing the holistic market trend.

Information architecture

Understanding the market and the models

Being a complete novice to the market and to the models, a large part was simply to absorb and talk to lots of people, it’s also accepting that information is never complete and prepare for that in the design.

Design user flow around use case scenarios

The architecture focuses on lowering adoption hurdle and display information by imitating an appraiser’s journey on valuating a property while adding additional macro values.

Give users what they are used to

With many variations and testing, I realised what works best is to give users what they are used to. The IA is rather flat as our users are used to printing the reports for documentation. Instead, the architecture shows layers vertically to help users navigate through the data pile. The data are shown in categories with benchmark data directly visible.

Prototyping & Testing

Test, Fail, Repeat

In the beginning, I teamed up with a rapid prototyper and we prepared the early versions of product value drivers to get feedback from customers. Feedback we heard most was that the UI looked good but the data did not make sense. After many conversations we found out that the data simply lacked a conclusion and was too complicated to digest. Internalising the insights, I restructured the IA, benchmarked the property drivers against the metro area and national area and added other features to help users understand the property and the market influence. This helped move the conversation forward for better insights and opportunities.

Interface Design

Continuous efforts

The product is the core section of our Saas Product offering. As with all agile efforts, I am still working on improving the usability and help our users make better decisions. On the side, I have also developed and maintained a design system management to facilitate our designers and developers from other projects to collaborate better.

It’s a challenge

It was in the beginning and we had little access to users. I had little knowledge of the market and it was a long process of trial and error. When we showed the users the first prototype they were not interested. "AI' is a fascinating, yet incomprehenisble and illogical. Step-by-step we improved the product by making it more relatable and the product started to gain momentum.