Develop a map-based, resource platform that enables Life After Hate to more effectively provide service to people trying to leave hate groups.
Product Designer, Front-end Developer
September 2019 - May 2020 (9 months)
Design: Sketch, Adobe Photoshop
Coding: React.js, Redux, Node.js, MongoDB
"Each year, more than 250,000 people in the United States are victims of hate crimes. The vast majority are violent and more than half go unreported. Between 2008 and 2017, 71 percent of extremist-related fatalities in the U.S. were committed by members of the far right or white-supremacist movements. LAH helps people leave the violent far-right to connect with humanity and lead compassionate lives." - Life After Hate
Life After Hate (LAH) is a nonprofit focused on helping people leave hate groups to connect with humanity and lead compassionate lives. They accomplish this goal this by connecting former hate group members or members trying to leave hate groups (called formers) to resources and partners within the LAH network. Some examples of resources are tattoo removal, career preparation, and education.
Life After Hate needed a way to manage and quickly search their resources. Two project teams helped create a web application that would allow users to search for resources and visualize on a map. The platform also had a way to manage accounts so that LAH staff could manage access to the application. This platform would allow LAH and volunteers to quickly find the information they need to assist ex-hate group members.
This is a 2 semester project done with Hack4Impact aimed at extending the reach and efficiency for the nonprofit Life After Hate. Hack4Impact is an UIUC that develops products for nonprofits to help them further their mission and better engage their clients.
When helping people (called "formers") leave hate groups, LAH staff members need to sift through their partner resources located on many different platforms such as Excel, email, paper, etc. LAH has had hundreds of businesses and individuals volunteer for their cause, but it can be hard to find the right resource to fulfill a person's specific request.
This bottleneck in the support process can also take days to complete. By automating and organizing the search process, we can speed up the process of matching resources to formers. We wanted to create a highly-secure, easily searchable, centralized web application to host and manage these resources. This would enable LAH to focus more on outreach and accomplish their mission on a larger scale.
After a few calls with LAH, we found that a map view would be necessary since location and proximity of resources are important when helping ex-hate group members. The map view would be similar to websites such as Redfin by giving a visual representation of data to help illustrate distance and positioning. While Redfin has a heavy focus on landscape, our map view focuses more on proximity.
There is also the chance that the user may want to do a high-intent search where they search using details such as a particular contact name. In this case, we need to display search results in a way that allows users to quickly skim through resource details. We did this by adding a separate directory view. This directory view displays resource details in a spreadsheet format to make it easy to find the intended resource.
I served as the Product Designer on a team with a Product Manager, Technical Lead and 5 software developers. We had two different teams for the fall and spring semesters, only difference in composition being that the spring team had one less developer.
As a designer and front-end developer, I was in charge of the UI and UX of the application. I worked closely with the Product Manager to help create user-centric prototypes and mockups, while also working with front-end developers to implement these designs in React.
I gave a brief overview of the rationale behind the choice of having the map view along with the directory view. This gives a broad solution to LAH's problems, but there are many details that still need to be explored.
One feature we passed on was mobile optimization. Given our users, we found that the application would be used almost exclusively on desktop. The time we would spend making the map view on mobile would be better spent working on another feature, so we chose to pass on scoping a complex feature that would see little to no use.
Since Hack4Impact works in increments of 1-2 semesters, it is important that we weight the complexity and impact of features against each other when deciding to include them or not. We have to prioritize completion of more important features to ensure we create the optimal product in our defined time frame.
One constraint that was immediately obvious to us as an organization was the handoff from one team to another between the fall and spring semesters. We paid special attention to naming conventions, code formatting, and standardizing site structure to ensure the spring team could easily work off of our existing codebase.
The way the project would pan out is that the fall team would focus on the map and directory view while the spring team would focus on account management, increasing engineering reliability, and building features based on user feedback. Focusing on the main map and directory views first would ensure we had time to process user testing and feedback to improve principal features.
Another constraint we came across concerned the security of our application. The data that we are displaying on our application has personally identifying information for individuals and businesses volunteering for LAH. If a hate group were to get their hands on this information, it could be catastrophic for those businesses and individuals. To ensure the security of this information, we chose to use Google OAuth to handle sign-ins. Google is a widely used and highly secure service, so it was the optimal choice to ensure security and inclusivity for those who access the platform.
Before we start creating designs for the application, we need to fully understand our audience and explore existing alternatives.
Starting with some alternatives to our application, the most obvious one is Excel/Google sheets. Both products allow for easy filtering, storage, and display information effectively. These products are similar to our directory view, but our map view remains unique. We can use this as an opportunity to differentiate our product by addressing LAH's pain points with spreadsheet products.
LAH's main pain points with Excel:
The first thing we need to do is define the user. In terms of our immediate users, we have 2 LAH staff. We also need to keep the potential volunteers in mind. Our product has a well defined use case, where all our users will be using this platform to search for resources and occasionally add/edit them.
We couldn't rely on only 1-2 users for all user feedback to build this application. We had to extrapolate some of the feedback to see if our platform would work for a generalized audience with little to no friction. To do this, we had Hack4Impact members test the function of our application prototypes and extensively researched the audiences and design decisions of similar products like Yelp, Google Maps, WebMD, and Redfin.
By doing some research on our users, we are able to come up with a list of user needs and other characteristics that will influence our design. By asking questions to LAH staff and volunteers, we were able to define the following user needs:
After identifying the user needs, we can go straight into building out the user flow diagram. This will help us lay out the architecture by creating flows of actions/screens that revolve around use cases.
Some of the use cases we can identify are:
This flow diagram combines all 6 actions into one graphic:
In the user flow diagram, the dark orange shapes represent completion of user tasks.
We have two account types, admins and volunteers. There are a few key differences between these two account types. Volunteers do not have access to the Account management page and are not able to add or view resources while admins do not have these restrictions.
After perfecting the user flow diagram, I got to work on the wireframes for the app. When creating these wireframes, I worked off of our user research and based the layout off of similar products (Yelp, Google Maps, WebMD, and Redfin).
All the mockups and designs displayed throughout this case study are from an admin point of view, meaning that there will be the option to edit and the account management page will be accessible. For volunteer accounts everything is the same except the management tab, edit button, and add resource button will be hidden.
There were quite a few problems we really dug deep to solve. I've outlined some of our more impactful design decisions we made to provide a better user experience.
When searching, we had 3 different types of data that factored into the search ranking: location, tag, and the keywords. Having 3 different text inputs would make searching take longer. How do we keep the search form user friendly while still collecting necessary data to make a search?
We had to go back to the differences between high-intent search and general search. We had already decided that high-intent search would be on the directory view page, while the general search would be on the map view page. So we broke down exactly what we needed to make both searches.
General search: Of the 3 different possible inputs, we found that for general searches LAH staff would use location and either keywords or tags. Location was a staple for this type of search, so we gave it a dedicated input field. Following principles of Postel's Law, we made the second input highly versatile by accepting both tag and keyword queries. Postel's Law states that you should be be liberal in what you accept, and conservative in what you send. We keep the search open ended and kept results relevant to the given search query.
To make this work, we created a weighted search algorithm that weighs keywords found within resources. Based on research and user needs, we created a search ranking system with the following components contributing to the rankings:
Each components has a weight based on how important we found them when displaying search results.
High-intent search: Because high-intent search is highly specific, we needed to use all 3 data components: location, tags, and keywords. These 3 independent fields give the user a high level of control over searches. The user still has the choice of leaving any field blank, but all options are provided.
We used a slightly modified version of the keyword scoring above, with the only difference being that the tag can be explicitly defined. This allows the user to really narrow down search results.
We revised the location input field to be extremely forgiving by accepting many location formats. Queries as simple as just "Chicago" or as complex as "600 E Grand Ave, Chicago, IL 60611" would both be accepted. With some extra engineering work and some research into basic search algorithms, we were able to optimize the user experience in both search cases.
LAH's resources are very complex and contain a lot of data. When displaying many resource cards on the map view, we needed to find a way to display the data in a digestible manner so users can easily skim and request more details.
LAH's resources are very complex and contain a lot of data. When displaying many resource cards on the map view, we needed to find a way to display the data in a digestible manner so users can easily skim and request more details.
Information is progressively displayed through 3 states:
condensed card → expanded card → modal window
Each progressive level will present more data, eventually displaying all details when you reach the modal window. This progression is representative of user behavior on the map view. When browsing for an item the distance, name, and tags are generally the most important, with a description to add some context. Then if you want more, expanding the card reveals relevant information to give to the former. This card would include details such as full description, email, and address. More specific results are listed on the modal window if necessary. But we kept all the most relevant information within one click.
Currently our map view is focused on slowly guiding you to the resource information you need through progressive disclosure. This method works great for users who are not familiar with the platform, but for some users (LAH staff) this application will be used extensively on a daily basis. The staff will be extremely familiar with the resources and the use of progressive disclosure will unnecessary in many cases. So how do we make our card design more user friendly for power users?
We need to implement a feature that the power users can utilize without interrupting the user experience for other users. The LAH staff (power users) are highly familiar with the resources because they are the ones who most likely are in contact with them and added them to the platform. Often times they know some specifics of a resource, they are only looking it up because they need a specific detail. Because of this, the intermediate step of expanding the card is unnecessary. Power users want to get directly to the modal window.
To cut the expanded view out of the process, we add a non-intrusive "shortcut" button in the top right of the card. This button takes you directly to the modal window. This button disappears and is replaced with a "close" button when the card is expanded. By doing this, we give the option to reduce the number of clicks it takes to go from condensed card to modal window from 2 to 1 click.
Since our project spanned 2 semesters, we were able to invest time and resources into user testing. We had most of the main components built out in the fall semester (map view, directory view) which allowed us to spend spring semester pinpointing usability issues with ample time to revise.
We primarily did user testing with the LAH staff and people who did not have a deep understanding of the application. We asked the users to try and complete certain tasks and documented the processes that each user followed to reach the end goal.
Through testing, we found that a lot of users were clicking on the "tags" of a resource. After further research, we found that the tags resembled buttons. They had an outline and used the same bright orange accent color as the buttons and map pins, both clickable objects.
Our initial solution was to revise the design to have no border and use a more neutral color, but instead felt there was a better way to address this. The pilot users were trying to filter by tag. They needed it to complete their task and were left frustrated when it did nothing and were forced to plan B. We left a lot of time to work on usability issues, and decided interactive tag filtering could be added to the scope without any adverse effect to our current timeline.
I already went through the reasons for leaving a dedicated tag filtering input on the map view search in the "Key Issues" section above. Despite that, we decided that tag filtering somehow needed to be incorporated but without compromising the simplicity of the current search experience.
We took a look at the user behavior and noticed that users were only clicking on the category tags after they couldn't find what they were looking for in the top results. For most users, filtering becomes important after they have started scrolling through the resources and want to look at related resources. Going back up to the search bar and typing the tag in would be more work, when the user could simply click on the tag.
This map view is being used for general searches, meaning that they often will not have a tag in mind unless they come across it in the search results. If they are looking for a specific tag, they would search in the directory view.
Having a clickable tag acts as a "recommended" section to find related resources. So the implementation we designed was to have users click on the tag they want to filter by, and this would be added under the search bar and re-trigger the search.
By adding this feature we will make it easier to browse for resources. This may be a little bit less useful for power users who know exactly what they are looking for, but when this product is expanded to volunteers we expect to see this feature utilized more.
After all of the user research, prototyping, and user testing we came to our final designs. The web app was created over 2 semesters by 13 developers using React, Redux, Node.js, and MongoDB.
Life After Hate already had a pretty established brand, with logos and a general color scheme. For colors, we used the main brand orange along with the gray as seen in the logo. We found it hard to design the application with one accent color and black, white, and the one gray, so we added a few different shades of orange and neutral colors while staying true to their original brand.
There wasn't an established typeface for LAH, so we went with a common sans-serif font, Roboto. The font is well suited for the card design that was used throughout the application.
The login system is relatively simple from a design perspective. There are only three login states: not logged in, logged in but pending approval, and logged in with approval (full access to platform). The UI is very simple, with no text input since authentication is done with Google.
When you first arrive on the platform, you are greeted with the sign in page. This page will bring you to a Google hosted page and prompt you to select a Google account. After this step, you are either directed to the map view (if you are a returning user with access to the application) or to the pending approval page. A sign-out button was included in case you select the wrong account when signing in.
The map view is page we expect to be used the most. Based on feedback, the map view fits the largest percentage of use cases. As a result, the map view has more features and is the default page after login. We tried to make the UI feel similar to other map based platforms (Google Maps and Redfin) to help the user feel some familiarity and replicate some of the efficiencies in these widely used products.
A lot of the features have been previously covered, but the design hasn't been covered at a macro level. Starting with the layout, we took inspiration from Google maps and created a left-aligned search result box on top of the map. We chose the left side for the search box because it follows the Z-pattern (blue arrows). Our hope is for users to quickly find the search box, and we used general user viewing patterns to position it correctly.
Beyond the search box, we used the F-pattern (eye tracking pattern coined by Nielsen Norman Group) to help us model how we would display search results. Users first read across. In this case they read the first card and the name of the resource along with maybe a line of the description. If it isn't what they are looking for, they move down the page, reading less and less as they go. This creates an F-pattern (red arrows). The amount they read across decreases as the user gets further down the list. Because of this, we needed to make sure that the search results were displayed based on relevance rather than something like distance.
We also needed the resource cards to be quick to read. The less time spent going horizontally, the faster the user will go down vertically. This is part of the justification for progressive disclosure. We want to present a small amount of extremely relevant information so that the user can easily make a decision to continue with the current resource or move on to the next.
The directory view is meant for high-intent searches and quickly adding resources. You can edit/view resources just as on the map view. We have the 3 field search bar followed by the grid view of resources.
I covered most of the UX decisions with the high-intent search in the "Key Problems" section, but I haven't gone through the UI components. For displaying the resources, we opted for a row and column design. Line separated rows would have been more space efficient, but we chose to keep the drop-shadow card design for resources on this page. The reason being that we want to keep our design consistent. The map view has cards holding resources, and we wanted to mirror this same design style on the map view to help users make the connection that these are the same resources as on the map view.
While the directory page will usually be used for high-intent search, we also wanted to allow quick search. We made sure that the 3 fields could all be used, only one be used, or a combination of them be used when searching for resources. This flexibility makes the directory highly versatile, and able to fulfill more than just specific queries. You lose the visualization that the map view provides, but still have a lot of flexibility when using this view. Skimming and quick filtering resources were some pain points in the map view, and we used the directory view to address them.
This page was added to help allow LAH to expand very efficiently. By having a dedicated page for admins to review and grant/deny access to the platform, it is easy to manage hundreds of users and their permissions.
This page is very similar to the directory view, except this page has tabs to separate the different account states. There are 4 different account states: admin, volunteer, pending, and rejected.
For this page, we group admins and volunteers together as "active users" but we keep the groups separate by displaying admins on top. We have separate categories for pending and rejected users. Separating pending and rejected states was important because it solves a few issues. If a user is denied access, we want it to stay that way. Without a rejected state they could keep requesting access. Instead, we can move them to the rejected pile and their requests for access will not show up. The rejected state also allows LAH to keep track of who has previously requested access and give or revoke access to the platform easily.
We also have a "Title" field which serves as a way for admins to mark down notes for users. This would make it easier for them to remember who someone is and why they have a certain permission/role.
This being my first project as a product designer for Hack4Impact, I got a ton of experience working with a team. Most of my previous design projects were done in smaller groups and on a more undefined timeline. With Hack4Impact, there was a lot going on at once, and I had to make sure my designs were easy to implement and ready for development. I also found myself reviewing PR's, assigning and creating tickets, and other various development related tasks. My role went beyond just designing, and I got a lot of experience developing parts of the front-end and learning more about how to build a product.
One of the biggest takeaways for me was learning the importance of prototyping. Early phase prototyping allowed me to deliver low-fidelity designs to devs so they could get the structure of the application created while I focused on perfecting the prototypes through user research and testing. I spent less time on intricate details and more time on the high-level experience of my designs.
I also found out how important it is to keep the development in mind. Typically the designer just creates a mockup with no regard for implementation and instead focuses purely on UI and UX. While these are the most important aspects for a designer, I learned it is important to also factor in the engineering side. There may be stylings or components that work better, but might be extremely hard for developers to implement or would not work well with the tech stack. Knowing the limitations of your development stack and staying up to date with the code-base can keep you aware of the product at a higher level. Which allows you to design features that could use the API more efficiently or gracefully build off of existing components. Keeping things realistic for developers helps the product move along and leaves time for more features to be built.
Overall, the project was a great learning experience and I learned how effective teams function. I want to thank my Product Manager, Alan Fang, and my Technical Lead, Josh Byster for their hard work leading this project, teaching our team, and being great mentors. I would also like to thank the fall and spring LAH developers for their hard work which helped make this product a success. Life After Hate also played a huge role in the creation of this product by spending time meeting and emailing us, making sure we had everything we needed.
We put a lot of thought and work into this product, and we hope that it helps Life After Hate grow and make a larger impact.