Implementing Boolean Search
Overview
SEEK Asia is a combination of two job portal brands, JobsDB and JobStreet, under one roof. They help facilitate the matching and communication of job opportunities between candidates and hirers. Our main markets are Malaysia, Singapore, Thailand, Indonesia, Vietnam, the Philippines and Hong Kong.
Challenges
The challenge that I faced during this implementation was understanding Boolean Search and how it has been utilised by hirers to do candidate sourcing. It was my first time hearing about Boolean Search and interesting to understand how powerful it is.
​
But the main challenges that I face are how to incorporate the search engine that we have in the existing SiVA RC (new system) and how it will impact the hirer's experiences.
Role
Product Designer
Lead Designer, User Research, Competitor analysis, Wireframing, Interaction Design, Prototyping, Usability testing, enhancement, handover to engineers and continuous support.
Project Time
Phase 1 took 5 months
(Aug 2021 - Dec 2021)
​
Phase 2 until successfully launch took 2 months
(Jan 2022 - Feb 2022)
Goal
Allowing user to have flexibility to do search that suit with their sourcing preferences and create a seamless search experience
Background
SEEK Asia is an extension of SEEK, an employment portal in Australia and they have expanded globally throughout Asia and American Latin. SEEK Asia is a combination of two job portal brands, JobsDB and JobStreet, under one roof and operating across six countries.
​
I have been working with SEEK Asia for almost 1 year and 6 months, mainly focusing on the JobStreet platform as well as help supporting design changes in JobsDB. When I first joined in, I was involved in one of the major projects that SEEK Asia has which was retiring our legacy system on the JobStreet Employer site called SiVA 11 (SiVA stand for Staffing Intelligent Virtual Agent) and migrating all our power users in SiVA 11 to SiVA RC (our new employer platform and RC stand for Recruitment Centre).
​
The reason for migrating was legacy system no longer can be sustained (using old system code) and SEEK main vision is to unify its product (JobStreet, JobsDB, and SEEK) into one platform to maximise the benefit of a streamlined online employment platform
Active in collaborating with Products, Data Analysts, UX Researcher and Software Engineers
For this major scheme, there are multiple medium and large-size projects that I handle as a product designer. I work closely with Product Managers, UX researchers, Data analysts, QA testers, and Software engineers.
​
Boolean Search is one of the big-size projects that we did to support SiVA 11 migration. I was the lead designer for this project and got supported by my Design Lead and others designers. They help by sharing advice and feedback on my design. This was my first time doing the structure design process from discovery, ideation, usability testing, and enhancement and successfully launched to all JobStreet markets.
Based on this project, I highlight a few things achievements which are:
-
Increased % NPS score for this product and % of purchasing credit for Talent Search. Since we incorporated Boolean Search, we gain a growth NPS score and purchasing credit
-
Increased returned customers rate by 6.4% compare before we have Boolean Search. Boolean Search was one of the features highly demanded from our hirers, and successfully building it, increased the number of returned customers rate.
-
Gain experienced remote moderated usability testing with customers using Figma prototype with different markets in Singapore, Malaysia, Indonesia and the Philippines. In my previous experiences, I didn't have structure usability testing. But in SEEK Asia, I got research expertise to guide us to do proper usability testing.
What is Boolean?
"Boolean Search uses a combination of keywords and the three main Boolean operators (AND, OR and NOT) to organise and sift through your searches. It produces more accurate and relevant results, allowing you to navigate through appropriate candidates, while disregarding the unrelated ones." - Social Talents
​
Boolean consists 5 types of operators which are
-
AND
-
OR
-
NOT
-
Brackets ()
-
Quotation marks ""
​
In recruitment, Boolean search allows recruiter to narrow down candidate search by using specific operators
For reference on how Boolean works in recruitment:-
​
Understanding the problem
To our product, Talent search is a powerful instrument where it helps employers/recruiters to do active seeking talents rather than waiting for job seekers/talents to apply to their job ads. Users be able to search, filters and purchase candidate profile to view their details and information, download their resumes or approach the candidate via email.
​
End of Nov 2020, we migrated some of our users to the new platform call SiVA RC and SiVA RC is the MVP product with basic functionality that we took from legacy platform (SiVA 11). This include search functionality in Talent Search.
​
Since the migration, there is increasing number of complaint about 43% from new users and migrated user on the absence of Boolean Search in SiVA RC through our Customer service team (Date collected from Jan 2021 to Apr 2021)
In the existing platform, we only incorporated Bubble search where allowing hirer to type keywords and entering them will turn the keyword into a tag. This type of search does not support any Boolean operator such as OR, NOT, quotation mark or brackets.
​
To understand further around the reason why there is increasing numbers of complains, I did a discovery research with help from UX research team to uncover why Boolean, how they utilising Boolean in their workflow and the impact absence of Boolean.
Bubble Search in SiVA RC
In SiVA 11, users are able to choose whether they want to utilise Boolean Search (we call as advanced search) or Bubble search depend on their job requirement
Discovery
For discovery, these are the research that I did and was involved in to uncover the impact without having Boolean Search in the platform, ways of users utilising Boolean in their workflow and the pain point they have when using our existing search:-
​
-
Feature and Customer churn for SiVA 11 (Legacy system) migrations to SiVA RC (Existing system) - Quantitative and Qualitative
-
Analysis of comparison of company's product (JobsDB, JobStreet SiVA 11 and SiVA RC and SEEK)
-
Hotjar survey to investigate users' feedback on existing search
-
Data tracking that we plant into the legacy system to observe boolean search usage in SiVA 11
-
Competitor analysis (LinkedIn, Google, Indeed, and Jora)
​
Our target participants, we pick variety range of users, from new users to the power user that have utilising Talent Search daily or weekly within 6 months
1. Research on feature and customer churn for SiVA 11 migrations to SiVA RC (Quantitative and Qualitative)
Our users, hirers in Asian markets are accustomed to using features available on SiVA11. Some of these features will not be available or will fundamentally change once migration to SiVARC, and ultimately the SiVA ANZ platform, has occurred.
Loss and change of features will impact the Customer Experience (CX) and this may not be off set by new value introduced by SiVA RC or by GTM efforts. Some these users may not be able to be migrated from SiVA11 to RC because of bespoke and complex requirements.
​
We did both qualitative and quantitative surveys across key Asian markets in Singapore, Philippines, Malaysia and Indonesia to understand usage,attitudes and opinions of features impacted by the migration, as well as new value added by RC platform features.
For Qualitative Research:
We conducted semi-structured interviews with 18 users currently using SiVA11. Eight interviews lasted 90 minutes, ten interviews lasted 60 minutes.
​
All interviews covered the following key topics:
-
The hirer’s role, their main responsibilities, and their main needs of SiVA11
-
Interactive demonstration of hirer’s usage of SiVA11 to reveal recruitment workflows and feature usage;
-
Hirer’s response to a potential scenario of feature change in terms of perceived impact on workflows or consideration of different solutions.
​
For Quantitative Research:
We distributed a survey to 614 users that currently using SiVA 11. These survey distributed to the key major user segment in SiVA 11 that called as Corporate, Business Process Outsourcing (BPO), and Recruitment Firm (RF).
​
Our survey design covered the following key topics:
-
How many users use SiVA 11 for each feature?
-
How much friction will be caused by the loss of features?
-
Will the loss of features result in customer churn?
​
Overall it took about 6 months. 3 months for Qualitative and 3 months for Quantitative. This duration includes kick-off, planning, preparing research brief, conducting research, synthesis of data, and creating reports and presentations.
​
Based on this research particularly on Boolean Search, we found out that:
-
Boolean Search is a feature that is relied upon for complex skill set combinations to find candidates for a position.
-
We found during discovery research, that most of our power users, come from well-known Recruitment firms, Corporate companies or Business Process Outsourcing companies (BPO), their recruiters or HR recruitments have been trained to use Boolean Search when actively seeking candidates/talents
-
It is the fourth highest used feature in the survey, with 63%
-
Estimated about 2 Millions revenue lost if we don't incorporate Boolean search into our SiVA RC platform
-
Loss of this feature will cause hirers level of pain, with NPS -3.3, the 3rd highest drop in NPS
-
Highest overall churn rate (I would stop using JobStreet and use another job board), 30.4%
Workshop brainstorming research question, research targeted user segment, and research design
Summary of the quantitative survey we did target SiVA 11 users
2. Company's product analysis (JobsDB, JobStreet SiVA 11 and SiVA RC and SEEK)
Comparing SEEK Asia products and SEEK products on ways they provide Boolean Search and seek if we can have the same approach to get consistency in search for every platform.
Showing comparison searching in SEEK, JobStreet SiVA 11 and RC, JobsDB
3. Analyse Hotjar survey from SiVA RC on pain point relate with Boolean Search
In SiVA RC, we did a satisfaction survey every 3 months, to understand the user's level of satisfaction, pain points or things we could improve.
​
From that surveys, the highest feedback we received was the issue they faced using our Bubble search and their request on Boolean Search
​
The requests mentioned numbers of times in Malaysia, Singapore and the Philippines markets and it contribute toward declining rate of users satisfaction level toward our product
.jpeg)
.jpeg)
4. Analyse data tracking that we have on SiVA 11 comparing usage of Boolean Search and Bubble Search
In our platform, we plant tracking for each call-to-action component on our side. This is to track usage, conversion rate and bound rate.
From our analysis of data tracking we have in the legacy system, we found that:
-
Overall, Malaysia's users are the highest user for Boolean Search
-
We have a total of 280,092 usage on Boolean searches performed in SiVA 11 for 1 month (July 2021 - August 2022)
-
AND operator is the highest keyword used when they perform a Boolean search
5. Competitor analysis on Boolean Search use in LinkedIn, Indeed, Google, Workabroad (Philippines's Job portal) and JORA
Comparing the implementation of Boolean Search in another job portal to see the differences and how well they executed the Boolean Search. I won't be able to go into their employer's site, I took this from their candidate's side.
Summary of the Discovery
-
We found during discovery research, that most of our power users, come from well-known Recruitment firms, Corporate companies or Business Process Outsourcing companies (BPO), their recruiters or HR recruitments have been trained to use Boolean Search when actively seeking candidates/talents
-
The issue lies when users can only key in/input only 1 keyword at a time and are not able to do any Boolean combination which is something they have been trained to use whenever doing a talent search as that part of company training to help narrow down the search to the specific candidate that matches the job requirement as well as maximise their workflow efficiency (reduce on time spent looking for a candidate as recruiter usually need to handle more than 1 job portfolio).
-
Boolean Search is a must-have feature that we need to build before migrating SiVA 11 hirers to SiVA RC
-
Prioritise this feature to be the top list risk feature to be adapted in SiVA RC
-
SEEK, JobStreet SiVA 11 and JobsDB have established Boolean Search. In all 3 platform, we found significant number of users using Boolean operator
-
Found that competitors are using the same search approach but no Bubble Search
Ideation MVP
Based on the summary of the research, I prepare two series of ideation sessions to address the requirements. As these users are from SiVA 11, one of the ideas I have is to try keeping familiar terminology and functionality.
-
Flexibility to use the two types of Boolean Search that we have in SiVA 11 called Advanced Search and More option search
-
Ability to have granular and broader search options
​
I mocked up some basic wireframes to gather feedback from the internal team consist Designers, Products, and Engineering on the overall layout, flow, logic, technicality and brainstorming things we could improve.
​
For low-fidelity design, we mainly focus to get the user flow right and a bit of review on the visual perspective. We did 2 sets of design reviews to get the user flow, UX logic and visuals right before I proceed to high-fidelity design that mainly focus on the interface.
​
1st round of wireframing, my design ideas was:
-
Removing the Bubble Search from the platform and at default, free text field search that allows users to apply Boolean operators
-
Adding search focus where users able to choose which area of candidate profile that the search need to focus on. It either search overall candidate's profile, search only specific to job positions or search only specific to skills.
-
Adding another type of search allowing users to have similar broken down Boolean Search, this search mainly focus toward migrated users
​
I got few feedbacks on the 1st round design checkpoints with the teams and did some enhancement of the flow before proceed to high-fidelity design and test it to the users. The feedback that I received around:-
-
Should we add an information on the Advance Search? What is advance search and how it can help users to do talent searching?
-
Visual perspective, having the visual bit more balance and play around with UI arrangement of the search section
​
2nd round of wireframing, enhancing the design ideas to:
-
Enhancing the visual look and feel
-
Add (i) for the Advanced Search to educate users on what is Advanced Search and how it can help their recruitment process
​
I got few minor feedbacks on the 2nd round design checkpoints with the teams and we brainstorming usability testing questions and what we want to find out​.
1st round of wireframing
2nd round of wireframing
Prototype the MVP
After 2 rounds of feedbacks, I finalise the flow and create it into prototype where internal teams able to play around with the new search functionality and share feedbacks on the experience. In this stage, we discussing and brainstorming on usability testing details.
Validating the design - Moderated Usability Testing
I conducted usability testing sessions with our users from 4 different markets. I divided this participant into this criteria:-
-
Participants must be SiVA 11 users who use heavily use our search to source candidates (At least 1 search per day)
-
Heavily use Bubble search​
-
Heavily use Advanced search (Boolean search with a free text field)
-
Heavily use More options search (Boolean search but structure Boolean)
-
-
Participants must be SiVA RC users who heavily use Bubble search
​
This usability testing goal is to validate the assumption we have
-
Overall users rely more on Boolean Search compared to Bubble search
-
A small percentage of risk removing Bubble search in SiVA RC
-
Small impact if we remove auto-suggestion and Bubble search in SiVA RC
-
Users need to have both free text field Boolean and structured Boolean to do sourcing
​
I tested them with 3 concepts of search to validate all the research goals. I gather feedback on the current search work, likes and dislikes and problems they have when using this search type
-
Current SiVA RC Bubble search
-
Free text field Boolean Search
-
Structure Boolean Search
​
I used Figma tool for prototyping that allows users to type in into the search bar, observe their behaviours when searching and keywords that they used when talent searching (Investigate whether they utilising Boolean operator)
Prototype for usability testing
Search 1: We test the existing search to user
Search 2: Free text search field allowing to input Boolean operators
Search 2: Structure Boolean keywords. Divided into Boolean category and user can only put single keywords with comma
Findings:​
Enhancement based on Research
Wireframing the MVP enhancement
Based on the Usability testing outcomes, these are the recommendation for the MVP enhancement:-
-
Maintain Bubble search and work around having both Bubble and Boolean search
-
To educate users that not familiar with Boolean, include a help page hyperlink
-
Maintain auto-suggest for Bubble search and include suggestion of recent search for Boolean search
Final Design
In the final design after a few rounds of feedback and discussions from design, product and technical perspectives, we finalise the overall flow and UI. Below are the key highlight of the enhancement that I did:-
-
Capability for users to switch search preferences
-
By default, we maintain Bubble search and we did a track on the "Switch" button to see the number of Boolean Search usage
-
Having auto-suggestion for Bubble search and recent search dropdown for Boolean search
-
Include a help page hyperlink and design the help page to have a visual describing Boolean operator works
​
Below are the example of high-fidelity prototype:-
Development
I created my high-fidelity mockups in Figma and then share the file to allow the engineers to inspect the file and export the HTML and CSS code.
I worked very closely with the Front End team to spec out any missing interactions that were not covered in the high-fidelity mockups. I conducted a UX review of each front-end component that was implemented to ensure it was aligned with the designs before it went live.
​
We successfully launch to all markets last March 2022. Below are the live production video:-
Outcome & Learning
Since the implementation of the Boolean search, we see there is an increasing number of users purchasing our Talent Search credit to purchase candidates' profiles. Also, we launch the Hotjar survey and we received more than 60% positive feedback from users on the Boolean search implementation as it help them to distinguish and broaden their search results which help to find the right candidates.
Some key takeaways from this project are:​
-
User testing validates the design solution. Design is a constant iteration of improving the experience for the end-user. Always find ways to collect and listen to your user's feedback. It would be best to have usability testing before proceeding to development. It would save a lot of time to do enhancement
-
Involve engineering upfront. This helps to reduce any rework later on as an understanding of the technical limitations upfront will help to inform your design strategy.
-
User research helps to uncover opportunities, helps to prioritise and risks that we have not been aware of before.
Project Information
Credits
Design Tools
PRODUCT, DESIGN LEAD, RESEARCHERS & DATA ANALYST
Figma, Dovetail, and Miro
Elliot Leung, Faidzal, Siti Nurain, Mimi Turner, Cheng Yew,
ENGINEERING
Ed, Hans, Haen Seng, Yow Kee, Adrian
