Quantexa Search
Quantexa Search
Quantexa Search
TIMELINE
September 2022 - December 2022
MY ROLE
Lead Designer
THE TEAM
A Lead Product Manager and Data Science Engineering Team
OUTCOME
Approved new design for the Entity Search, called Search 2. A global Query Builder feature and Panel component being created for the platform, and the design of a global Help section for the platform.
OVERVIEW
Quanetexa’s Entity Search incorporates propriety AI Entity Resolution and is used for Customer Intelligence, Risk Management, Fraud Prevention and Anti-Money Laundering. The Entity Search is the primary entry point for most users when performing their key tasks on the platform and had consistently received high amounts of negative feedback during user focus groups.
OVERVIEW
Quanetexa’s Entity Search incorporates propriety AI Entity Resolution and is used for Customer Intelligence, Risk Management, Fraud Prevention and Anti-Money Laundering. The Entity Search is the primary entry point for most users when performing their key tasks on the platform and had consistently received high amounts of negative feedback during user focus groups.
THE OBJECTIVE
The search is the first entry point for most users to the software and sets the impression for the product as a whole. While daily users recognised Quantexa’s powerful capabilities, they described the interface and overall experience as clunky and unintuitive.
THE OBJECTIVE
The search is the first entry point for most users to the software and sets the impression for the product as a whole. While daily users recognised Quantexa’s powerful capabilities, they described the interface and overall experience as clunky and unintuitive.
GOALS
Improve perception of the platform
Create an experience that truly reflected the sophistication of the platform.
Boost retention
Mitigate customers leaving the platform in favour of competitors due to poor usability.
The platform before we started:
A prototype of our phase 1 release:
FOCUS GROUPS
The Product Owner and I held group interviews with Financial Crime Investigators from organisations such as: The Cabinet Office (HMRC), HSBC, Standard Chartered, Royal Bank of Canada and Barclays.
We also spoke to experts in the different verticals within Quantexa and members of the Customer Success team.
With these sessions we wanted to discover:
How and why they use the search and their common query types.
Their frustrations.
The things that were most helpful on the platform.
Other software they have used to search that they prefer.
Features they might want added.
FOCUS GROUPS
The Product Owner and I held group interviews with Financial Crime Investigators from organisations such as: The Cabinet Office (HMRC), HSBC, Standard Chartered, Royal Bank of Canada and Barclays.
We also spoke to experts in the different verticals within Quantexa and members of the Customer Success team.
With these sessions we wanted to discover:
How and why they use the search and their common query types.
Their frustrations.
The things that were most helpful on the platform.
Other software they have used to search that they prefer.
Features they might want added.
I created a persona to help us understand a day in their life
COLLATING INSIGHTS
I created an experience map using answers from the survey to help identify which step in the journey had the highest amount of pain points.
Key themes
Untrustworthy results
Users didn't trust our algorithm to return the most relevant results in the best order.
Inability to save work
The search includes a "Pinned List" feature where users can pin results to a list they can refer back to. If the page is refreshed the list is lost. Understandably, this was a huge source of frustration.
Cognitive overload
Users reported feeling like they couldn't easily scan the page or quickly find the information they needed.
Poor aesthetic-usability
Users often made comments about the look and feel of the product and said they felt it didn't match up to other apps they used in their everyday life.
COLLATING INSIGHTS
I created an experience map using answers from the survey to help identify which step in the journey had the highest amount of pain points.
Key themes
Untrustworthy results
Users didn't trust our algorithm to return the most relevant results in the best order.
Inability to save work
The search includes a "Pinned List" feature where users can pin results to a list they can refer back to. If the page is refreshed the list is lost. Understandably, this was a huge source of frustration.
Cognitive overload
Users reported feeling like they couldn't easily scan the page or quickly find the information they needed.
Poor aesthetic-usability
Users often made comments about the look and feel of the product and said they felt it didn't match up to other apps they used in their everyday life.
BRAINSTORMING WITH THE SQUAD
There is a dedicated Search team at Quantexa made up of four data science and engineers and one full stack designer who would handle the UI.
The team had never worked with a designer but were keen to get involved.
I organised a workshop where we ran through the research findings, brainstormed ideas and collaborated on a task flow to understand the technical touch points.
BRAINSTORMING WITH THE SQUAD
There is a dedicated Search team at Quantexa made up of four data science and engineers and one full stack designer who would handle the UI.
The team had never worked with a designer but were keen to get involved.
I organised a workshop where we ran through the research findings, brainstormed ideas and collaborated on a task flow to understand the technical touch points.
EXPLORING QUERY CREATION
I created some low fidelity wireframes and prototypes based on the two different methods of creating a query:
Most common: free-text search
Simply typing out their search query and sometimes using the available operators, such as wild cards and boolean logic. I investigated what operators were available and if we could add a cheat sheet of the operators to the UI.
Least common: constraint based search
Using constraints or fiters to search specific fields within the dataset. A form builder, advanced search type of experience.
I played around with different layout options and isolated some of the interaction flows. I discussed these options with the Product Owner and team. Based on the feedback, I created some more complete wireframes to discuss again with the team. Although Quantexa's search is unique in how it works I wanted to apply common interaction patterns where possible.
EXPLORING QUERY CREATION
I created some low fidelity wireframes and prototypes based on the two different methods of creating a query:
Most common: free-text search
Simply typing out their search query and sometimes using the available operators, such as wild cards and boolean logic. I investigated what operators were available and if we could add a cheat sheet of the operators to the UI.
Least common: constraint based search
Using constraints or fiters to search specific fields within the dataset. A form builder, advanced search type of experience.
I played around with different layout options and isolated some of the interaction flows. I discussed these options with the Product Owner and team. Based on the feedback, I created some more complete wireframes to discuss again with the team. Although Quantexa's search is unique in how it works I wanted to apply common interaction patterns where possible.
CO-CREATING WITH USERS
The PM and I held regular sessions with certain users who wanted to be involved in the design process.
There were two investigators from the Cabinet Office who were often on-sight at Quantexa. We had regular feedback and iteration rounds with them and let them talk us through what a day in their life as an investigator is like and the things they would and wouldn’t find useful in the new designs.
CO-CREATING WITH USERS
The PM and I held regular sessions with certain users who wanted to be involved in the design process.
There were two investigators from the Cabinet Office who were often on-sight at Quantexa. We had regular feedback and iteration rounds with them and let them talk us through what a day in their life as an investigator is like and the things they would and wouldn’t find useful in the new designs.
FIELD CONSTRAINTS & DATA FIELDS
I worked closely with the Lead Data Science Engineer on the team to map the data fields and contraints that make up a search query.
The data science engineers had been working on solving the biggest pain point: improving the quality of the results returned.
They had advised that if users applied correct constraints within their searches it would improve the quality of the results.
I had been exploring using a chip approach to the search constraints and terms, similar to typing in an Email recipient into most Email providers. This is to help the user identify the parameters applied to the search.
The PM and I had ran this past our users and they were in favour of the approach.
FIELD CONSTRAINTS & DATA FIELDS
I worked closely with the Lead Data Science Engineer on the team to map the data fields and contraints that make up a search query.
The data science engineers had been working on solving the biggest pain point: improving the quality of the results returned.
They had advised that if users applied correct constraints within their searches it would improve the quality of the results.
I had been exploring using a chip approach to the search constraints and terms, similar to typing in an Email recipient into most Email providers. This is to help the user identify the parameters applied to the search.
The PM and I had ran this past our users and they were in favour of the approach.
STAKEHOLDER PLAYBACK
The PM and I presented the strategy, research findings and prototypes to stakeholders including the Head of R&D and CTO who is a co-founder of Quantexa.
We put in regular two week design sprints to iterate and present prototypes. The CTO had asked us to explore an Entity first approach to the search constraints.
STAKEHOLDER PLAYBACK
The PM and I presented the strategy, research findings and prototypes to stakeholders including the Head of R&D and CTO who is a co-founder of Quantexa.
We put in regular two week design sprints to iterate and present prototypes. The CTO had asked us to explore an Entity first approach to the search constraints.
INCREMENTAL RELEASES
I worked with the PM and Full-Stack Engineer on the team to refine the Front-End tasks and to do list that we would update as designs were approved.
We decided to start with styling enhancements and designing a global panel component, this would be a great way for us to introduce scalable new styling updates to the entire system.
INCREMENTAL RELEASES
I worked with the PM and Full-Stack Engineer on the team to refine the Front-End tasks and to do list that we would update as designs were approved.
We decided to start with styling enhancements and designing a global panel component, this would be a great way for us to introduce scalable new styling updates to the entire system.
DESIGN SYSTEM
Given the search wasn’t the only screen on the platform using a query builder, we aligned with existing query builders and created a Query Builder feature level component for the global design system.
We created global components from the child components of the Query Builder that could be used for other filtering use cases.
DESIGN SYSTEM
Given the search wasn’t the only screen on the platform using a query builder, we aligned with existing query builders and created a Query Builder feature level component for the global design system.
We created global components from the child components of the Query Builder that could be used for other filtering use cases.
HELPING USERS LEARN
During our group sessions we had discovered that users had no help section to refer to after their initial training session of how to use the software.
I had designed a Search Tips page and spoke with the UI team about setting this page up in a way that it could be used as a global template for Help pages on the platform.
HELPING USERS LEARN
During our group sessions we had discovered that users had no help section to refer to after their initial training session of how to use the software.
I had designed a Search Tips page and spoke with the UI team about setting this page up in a way that it could be used as a global template for Help pages on the platform.
CONCLUSION
I worked with the PM to put together an evaluation plan using the SUS score and qualitative user testing insights to measure the improved usability of Search 2 against Search 1 before leaving the company in 2022.
The UI updates to the search results and the global Panels had been published at the time of my leaving the company in 2022. The general direction of the Query Builder had been approved and development work had started.
CONCLUSION
I worked with the PM to put together an evaluation plan using the SUS score and qualitative user testing insights to measure the improved usability of Search 2 against Search 1 before leaving the company in 2022.
The UI updates to the search results and the global Panels had been published at the time of my leaving the company in 2022. The general direction of the Query Builder had been approved and development work had started.