case study – Nevy’s language for lower levels
In 2018, I worked with Nevy’s Language for Lower Levels to lead UX research and strategy. Nevy’s Language is a language training non-profit bringing innovations in second language learning, adult education and program delivery. Their mission is to improve the English levels of underserved newcomers with limited language skills through a range of innovative language products and services. My goal was to better understand how Service Providing Organizations (SPO) efficiently advertise and engage newcomers and how newcomers easily learn about and access language services.
a lapsed system
There are tens of settlement agencies in Toronto alone and tens and hundreds of thousands of newcomers to Canada every year. This means that the ratio of newcomers to agencies and service providers is very lopsided. In addition, most agencies are using the same methods and approaches to serve clients. A gap was identified in the current industry; marginalized individuals are not able to reach their potential and gain the skills needed to be able to gain employment in their desired occupation.
Some of the problems identified with the current system were:
- Service providing agencies are never fully resourced.
- Settlement workers are inundated in calls and requests for information.
- Agencies struggle to recruit participants because they use old approaches to advertise and recruit.
- Lack of innovation in sector prevents clients from accessing information they need using integrated online solutions.
the potential for a digital solution
We set out to design a platform that utilizes a chatbot through several apps, allowing agencies to deliver information to clients about their services. It would allow newcomers to engage in conversation with this chatbot in order to ask quick questions that they need help with, and inquire about upcoming classes, and to ultimately enrol in them.
Our aim was to develop a digital approach to outreach and recruit newcomers by:
- Facilitating direct and efficient communication and dialogue with newcomers
- Building better relationships and engagement with newcomers
- Delivering the experience via a familiar and conversational digital method
a two part research process
My research and strategy was done in two phases, and involved both attitudinal and behavioural methods, as well as quantitative and qualitative ones. During the first phase, I planned some exercises and activities with the internal team to get a better understanding of the problem space, the sector, and our assumptions about our users and their behaviours.
I used three methodologies: Lightning Talks, How Might We (HMW), and Affinity Diagraming. They are described in more detail below. I then created deliverables such as a UX blueprint to bring all the information together.
In the second phase, I conducted user research; an online questionnaire with those who had worked or are currently working in service providing agencies, and one-on-one interviews with newcomers who would potentially use our solution. The research goal was to identify key challenges in the current outreach and recruitment process, define a future vision for what an ideal experience should be, and determine how a technology solution could help alleviate existing challenges.
One-on-one interviews with newcomers included those who were diverse in gender, age, years in Canada and languages spoken. The goal was to better understand newcomer’s needs, goals and challenges when accessing services and to define a roadmap for what a desired tech-savvy system looks like.
key findings and opportunities
My research resulted in a series of findings and recommendations that shed light on what this potential digital system should focus on. Language barriers, meaning the inability to speak English was a major limitation for newcomers, as it prohibits them from knowing about existing services, accessing these services, as well as navigating the overall system. In addition, delivery methods are outdated and not in line with current newcomer behaviours; the sector does not change fast enough to accommodate for new and improved methods.
Based on the research conducted, a series of ideal outcomes for a digital system were identified. Below are the requirements we had to keep in mind as we moved into the design and build phases of the project.
1. Culturally appropriate
- System should address the diversity of users.
- It should consider cultural context, sensitivities and limitations.
- The tool should be available in many languages that reflect current newcomers’ cultural background.
- Plain language should be used at all times, with minimal use of lingo.
- It should consider technological literacy as well as behaviour.
2. Better client experience
- Service providers will spend less time on administration and focus instead on improving client experience.
- Less focus on outreach will allow for more focus on programs that are valuable, unique and diverse.
- Provide an experience that is conversational, friendly and personable, in order to bridge the gap between virtual and in-person support.
- An engaging and positive interface that can perhaps be a replacement for a one-on-one experience.
- It should resonate emotionally.
- There should be follow-ups to ensure that newcomers get recruited in the program. They should be kept up to date with the status of their request.
- The system should provide access to available and updated information and resources in the sector.
- Information should be up to date and relevant.
- Users should be able to trust the people providing the service as well as the service being provided.
- Provide an experience that is trusting; tool should allow users to feel comfortable with sharing information.
5. No barriers to access
- Short and digestible content.
- Useful features and functionalities: list of available services, directions and location.
- Provide an experience that is seamless and reduces learnability load: content should not be overwhelming to users.
- Digital system should be simple to use.
- Users should be able to find services they need easily.
- Tool should provide clear information about programs and make it accessible for users to access.
- The digital system needs to be less time consuming in its recruitment and enrolment process.
- Enrolment and application to programs should be online.
- Include a centralized data base where all clients can be registered and reached.
- Provide functionality and connectivity with other social platforms and communication technologies that they currently use (i.e. Facebook, WhatsApp).
- System should be measurable in order to better understand what works and what doesn’t.
- Quantitative data will help determine what is and isn’t successful for later iterations and an anticipated phase 2.
chatbot and conversational design
Although the main goal of this tool is to recruit newcomers in language classes, our aim is that it also helps with case management, helping perform internal evaluation tasks. We see a huge potential for this tool to become a meaningful conversational tool and a means to efficiently engage and share information with non-profit beneficiaries.
The design of the chatbot was based on best practices in conversational design, taken from Google’s Design Guide. We based our principles on the main premise that the chatbot has to be as humanistic, natural, conversational and trustworthy as possible. Some of the considerations that were important when designing the conversation flows were:
- Cultural tendencies must influence the phrases and words communicated, i.e. there are different dialects in the Arabic language.
- Conversation can’t be too structured and restricting, otherwise it will feel like a chatbot: automated and impersonal.
- Conversation can’t be too fluid and loose, otherwise it is hard to guide conversations, making it harder to build.
- Outputs can’t just spit out paragraphs of text. They have to be short responses, emulating how a real conversation would go.
- The bot has to follow the mannerisms of a conversation to appear as human as possible.
We created examples of flow diagrams to illustrate the client recruitment system, and where the chatbot would fit inside the entirety of the journey. We also created a table to display conversation scenarios with different audiences, the languages they speak, their first touchpoint, the medium used, and finally a list of reasons they they choose to reach out. Lastly, we brainstormed and developed a translation library; English phrases and its Arabic translation were recorded to inform conversations that would be implemented on the chatbot.
In order to build the chatbot, we conducted a tool assessment, evaluating each tool and software based on the requirements that we discovered during our research phase, as well as language support, user friendliness. and subscription costs. Some of the technical considerations we had to keep in mind were:
- How will the solution be built?
- Is the solution likely to be web-based? Mobile?
- Where will the data and information come from?
- Will user data be used for personalization?
- How will privacy be addressed?
- How will accessibility be addressed?
The chatbot is currently running with Facebook messenger and Twilio (sms messages). Our goal for next steps is to reach all social media apps and in all native languages. Our conversation design will continue to follow Google’s best practices, be culturally tailored to establish trust and reliability with the client, and be tested to ensure that designs are always being iterated upon based on user feedback. We will continue to work with developers to build a system that will be able to automatically translate the many languages clients will use to respond to the chatbot into English, based on a library of phrases.
This project was a huge learning opportunity. Researching conversational design and applying it to a context and users who speak a language other than English proved challenging. In addition, assessing existing technologies and ideating on innovative methods to solve a problem with real human and social impact was extremely rewarding.