THE LOUNGE HUB - UX DESIGN/RESEARCH & DEVELOPMENT OF MOBILE APP AND CHATBOT
This is the Final Major Project. It is self-guided and chosen by the student.
The Lounge Hub is a reservations mobile application exclusive to shisha lounges. The app will allow users to book tables, join a virtual wait-list for reservations, discover new lounges as well as check any information related to them (overview, menu, reviews).
The aim of the project is to provide shisha lounges with a platform suitable to carry on doing business according to the new guidelines post-covid-19.
Timeframe: May – August 2020
- UX Research – market, user, target audience & personas, empathy maps, user flows & journeys, information architecture, sitemaps.
- Design Theory & Frameworks and Methodologies
- Marketing – AR Filters for FB & Instagram
- Conversational Design & Chatbot Development
- Branding, Wireframing, Prototyping, App Development
Framework & Methodologies
The framework for this project combines a number of methodologies that were necessary to appropriately structure each step of the process. The overall project framework is that of design thinking and follows the standard five stages. Within the last two stages and for the chatbot side of the project, it was essential to use more specific methodologies such as Agile and RAD to achieve the best results. The project also closely follows the user-centred design framework.
It was found that shisha smoking is the highest between males of the age 18 to 24. It is particularly popular within Middle Eastern and Asian ethnic groups, although it is being sought out by White British groups too. This can be attributed to the fact that this social activity originated from the Middle East and is a cultural norm that is commonly practised even at home. In fact, it is normally introduced by families as it is a shared activity that occurs during gatherings. Other motives behind smoking shisha is socialising, networking, for stress-release especially after work or university, or as an adjacent activity while eating, watching TV, conversing, or playing games. This in turn led to the increase of smoking time from an average of 45 minutes to several hours.
The majority of the research conducted around shisha lounges barely brushes over their demographics. The reason is because the focus of the studies is the impact of smoking on health and not the establishments. Therefore, a lot of the data presented previously is generally about shisha smokers and not the actual demographics of lounge frequenters. To get more insight on the users, shisha lounge frequenters were interviewed.
It was found that those who visited shisha lounges were of an older age group, early twenties to early thirties, but stayed true to the male dominance. Some enjoyed visiting lounges after work with their colleagues or friends as a way to relax and have a good time, while others kept it as a weekend activity as a night out. Also, the activity was popular with university students, but more so with those with jobs, particularly low to medium incomes. This was because the general cost was very affordable while offering all that was wanted, from food to entertainment to smoking and having a great time.
Since the introduction of the Smoke-Free law in 2007, shisha lounges grew by 210% by 2012/2013 and their popularity does not seem to be slowing down. While the smoking ban does apply to tobacco and shishas, there are ways around it. Shisha lounges began opening outdoor spaces or semi-roofed areas which legally allowed smoking activity to continue, and even made it more desirable. However, there is no doubt that this will all be affected due to covid-19 as well as cut-downs and raids by councils on lounges.
Aside from the booming industry, it was important to research the mobile app market. Current existing restaurant reservation apps are OpenTable and TheFork; both apps are children of famous parent giants Booking.com and TripAdvisor, respectively. Both companies have a long history in booking and reservations, and so they used this experience to roll out mobile apps and even a restaurant POS system. Yet, neither include shisha lounges or any of similar nature, nor other apps on the market.
The prototype consisted of five main pages which were the home (discover) page, bookings, saved, notifications, and profile. The prototyping stage was heavily directed by user research and feedback through continuous testing. Gestalt’s and Gutenberg’s principles led the design process and can be seen throughout the app.
App Information Architecture
In recent years, businesses have become more aware and are engaging more with the thought of deploying chatbots and AI. In 2019, chatbots saw a 136% growth rate with 53% of service establishments expecting to use chatbots within 18 months (Sweezey, 2019a). This comes as no surprise when the advantages are considered, the top being 24/7 service and instant response time improving customer service, increasing customer engagement, and if it is an AI chatbot then complex problems could be solved without the need for human assistance. According to a research conducted by Mathew Sweezey on SalesForce (2019b), “58% of customers say emerging technologies such as chatbots and voice assistance have changed their expectations of companies”. Moreover, when offered a choice between filling in a form (in the case of this project there will be 2 forms) or asking the chatbot directly, 86% of customers chose the chatbot. The Voice Report (2019) by Microsoft found that 72% of respondents used a personal digital assistant for voice search, and 57% actually preferred using voice over texting. This shows a promising adoption journey of the technology despite it being in its relatively early stages.
The chatbot for this project – named Ashes – is a simple AI chatbot that is mainly used for the booking process. The reason it is an AI bot instead of a menu bot, is because it is much better at understanding contexts, differentiating between user intents, and most importantly it can be trained. The bot was developed using Dialogflow and exported to Facebook Messenger.
Chatbot Training Phrases
Chatbot Entities - Detailed
Besides the Facebook page and the chatbot on Messenger, the main marketing strategy is actually for the Instagram page (linked on the cover). Considering the age range of the target audience and their levels of engagement with lounges on the app, the impact of Instagram and the importance of having an account cannot be disregarded. In fact, as of June 2020, there were 26.54 million Instagram users, with the majority being from 18-34; 18-24 counted for 23.4% of all users, and 25-34 were 30.5% (Statista, 2020). In addition to the standard content, the main form of marketing will be through Instagram Stories. Statistics show that half of all monthly users use Instagram stories every day, that’s 500 million people, and 62% of users grew interested in a business due to the stories (Hootsuite, 2019). There are a number of different ways to utilise the potential of stories, whether that is through the use of hashtags, location, or using polls.
For this project a set of original AR Instagram filters were created to be used on stories and will be available to all Instagram users. All the media material strictly follows the branding and overall theme of the app. The filters were built using Spark AR which was refreshingly simple to learn and use. The filter comes in two different colours and is triggered by the user’s mouth movement; it is inspired by smokers’ obsession with making smoke rings. The user can control the speed and size of the rings using their mouth.
Testing was done at nearly every stage of the development process and different methods were used depending on the function that needed feedback. Due to covid-19 and the lockdown, user testing was quite complicated as it all had to be done remotely and through online tools and therefore it was unmoderated; the average sample size was 10-15 users that included both the target audience and potential users.
When the general idea and direction of the app was settled, a quick interview session with the target audience was conducted. Since the user was the centre of the project, it was important that it was developed to their needs. Conducting such interviews before designing and building provides valuable insight into what the user expects, needs and wants, and a better understanding on how to create the best user journey for the best user experience. The interviews were completely open, and the conversation was led by the user so they could fully express their thoughts; they were only given a description of the app idea and what it would offer.
A/B testing was done with two main design elements; theme mode and the menu. The A/B testing for this stage focused more on the qualitive feedback from users rather than quantitive. For the app theme, two versions (light and dark modes) were sent out to the users, and the feedback was almost immediate; nearly everyone wanted to use the dark mode.
The second test was done to determine the kind of menu users prefer. The first version was a burger menu that slides in from the left, and the second was a bar menu at the bottom of the page. The test showed that the bar menu was preferred because for one it was visible, it highlighted where they were in the app, and it really just simplified the process.
User Testing – Open Analytics & Poll
While working on refining the prototype and development simultaneously, another round of testing was conducted, and it was not limited to a specific audience. Using XD and Useberry, the app was shared through social media for users to freely navigate and use. The most common feedback from XD was on the weight and size of the font, followed by colour contrast in the inner pages of the app where more white is used. This was all changed and increased according to the feedback.
The results from Useberry on the other hand were extremely insightful. One of the concerns was whether a profile was necessary for this kind of app, and if it was then how do users go about this. When the results got back, it was found from the user flow that the first thing nearly everyone did was navigate to the profile page to sign in or sign up. This immediately solved the first concern. The second was which method and providers do users prefer. The heatmap of the profile page showed that a lot of the engagement was with Google. To further solidify this, a quick poll was done through social media, and the results showed that Google was indeed the preferred sign-in method.
Once there was a chance, a few in-person interviews were done to discuss the test. The biggest benefit of such interviews is the ability to see the users’ body language and facial expressions when using the app. It uncovers a new level of user understanding because it physically shows whether the user is frustrated, having difficulties, or is satisfied. Moreover, testing in-person provided the opportunity to get unfiltered feedback because it is immediate.
User Testing Part 2 - Tasks for Target Audience
This time round, the testing was focused on five subjects that were frequenters of shisha lounges, up to 4 times a week. They were of different genders, age range from early twenties to early thirties, different social statuses, etc. Through Useberry, they were given four tasks to complete. These were the most relevant to their user journeys. The tasks were to book a table at a lounge, join the waitlist, find the shisha menu, and signing in to view their past bookings. the results showed that were no failures to complete the tasks, and they found it quite easy and quick to reach the goal. The feedback given was to add features that allow the users to edit or cancel a booking from the booking page. After this was implemented, two extra features were added which were the ability to call the place and to rebook using the same details as before.
After all the app features were developed, the APK was sent out to five users so they could download the app and test it on their own devices. This is where the limitations regarding the Facebook SDK were discovered. All the other features worked just fine, although there were also some slight UI issues with the prefabs instantiated from APIs. The users enjoyed using the app and were happy to see the results after all the previous prototype variations.
Before deciding on Dialogflow as the chatbot building tool, Amazon Lex was in fact the start of the chatbot. However, after working with Lex for a while and testing it, it was found that it had weak intent detection which posed as a major issue. In order for the chatbot to work, the phrasing of the request has to very specific and near identical to the sample utterances of the intent otherwise it would not pick it up or confuse it with another intent. Another tool that was used was Chatfuel. Though it is quite limited, it was a good contender for the win but eventually fell short to the benefits of Dialogflow. Due to the limitations of the Facebook SDK, the Dialogflow Messenger bot could not be tested by anyone else.