User testing as a design driver
How Looksery created a product for users, not designers
You may have recently seen an abundance of puking rainbows and bug eyes on Snapchat. You can thank Looksery for that.
Looksery launched last year as an entertainment app based on face recognition technology and special effects.
Founded in 2013, Looksery went on to raise money for its face modification app on Kickstarter. After successful crowdfunding, Looksery launched in October 2014.
They were acquired by Snapchat in September 2015.
I (JC) took some time to speak with lead designer Lidiya Bogdanovich (LB) about their design process and how user experience testing transformed their app from good to GREAT.
JC: What was the original concept for the app?
LB: Looksery was always meant to be an entertaining app that demonstrates the technology of face detection but in a fun way, like a video-messenger for you and your friends.
JC: What was your role? How big was your team? What type of team members did you have?
LB: Since the very beginning of Looksery I was the only designer. On later stages (post launch) our design team grew up to three people.
We were a small team and we were doing everything possible for our product to blossom, so we didn’t have very strict responsibility borders of or clearly defined roles.
Startup members usually need to be able to perform multiple tasks and we were lucky to have the “crème de la crème” of design society.
JC: And how did you go about designing the product?
LB: At first we on speculation and past experience, relying on our professional knowledge and more than ten years of design experience.
We wanted our app to appear as innovative as possible so we used trendy UI features. It looked awesome.
It was design made by designers for designers and that is not what we wanted.
It was design made by designers for designers and that is not what we wanted.
The first user testing showed that people didn’t get how to use it. At the end of the day we wanted all users to enjoy our technology without even thinking about UI.
JC: Talk a little bit about user testing. How did you administer the user tests? Did you use different methods? Did any methods work better than others? Worse? How many rounds of testing did you do?
LB: We tried a lot of approaches at first. But step by step we figured out what our mistakes were, and proceeded with our “polished” method.
First we defined our audience. If we’re an entertainment app, who do we entertain?
We decided that the main audience would be teenagers, students and young people who are active in social media.
Every week we were communicating with our audience in both private testing (when we are talking about UI) and general testing (we were gathering people all together and showing them our special effects, to define which ones they’d use)
While one on one testing we were trying to create a friendly relaxed atmosphere, giving our user simple tasks and asking them to comment everything they do while drinking coffee and eating cookies.
That way you really see what your user is about, not when they feel like they are a lab mice.
We did at least small testing every other week, that gave us flexibility to react, in case we are doing something wrong.
Testing started long before the first launch and proceeded till the very end and that was very important.
We ended up changing the UI completely after we started testing. At first we wanted to be like all those apps with awesome design, which screenshots people post like that much on design web sites. But it appeared that those apps come and go, and really awesome ones, that people like and use on an every-day basis, those ALWAYS have a very simple, easy to use design.
And this is what we wanted to be. Not a screenshot on a UI design gallery, but an app that you open every day to chat with your friends.
JC: Could you talk more about your user test findings?
LB: The main problem was that people just wouldn’t hold camera the right way, so the face wouldn’t detect and the technology would reveal itself — users needed strong lighting to capture their face sufficiently.
We needed to simplify the experience — instruct the user on what to do without being annoying or making the app look complicated (because no one wants to play with a complicated app).
We tried tons of concepts and none of them worked.
But after enough testing we figured out that the best results were using walkthrough screens. So we decided to work in that direction.
We hired a model and a photographer and created new screens. The results were much better this time, but there were still room for improvement.
What did we do then? We reduced the length of the titles, and added a “hint” arrow, showing what to pay attention to.
We still were getting some funny results but it was working much better.
JC: Any other major problems that you uncovered from the testing?
LB: Another problem was with a placeholder-picture that illustrated the photo-filters — the instagram-looking ones that are currently present in every camera.
After looking through lots and lots of pictures we decided to go with a parrot one because it had the best color scheme and the most optimal contrast and detalization.
That brought up other issues. We started getting “why I am I pressing it I’m not getting a parrot-look” complains but that is part of the learning process!
JC: What are your major takeaways from the process?
LB: A few things…
1. Choose the right audience. If you are making app for professional taxi drivers, testing it on your teenage sister or your grandma (unless she is professional taxi driver) is not an option.
2. Create a real environment where the app is meant to be used. If it is a navigation app test it while walking on the street or riding in a car. If it is the kitchen recipes app, try to test it while cooking in the kitchen.
3. Install the test app on a familiar device. Try to find something very similar to what they use vs testing on a foreign device — people get used to their devices and other operational system and new phones can cause a lot of stress and irritation.
In our case, people didn’t want to take selfies or test the sharing option on our devices, because they were afraid that those pictures can go public or used in the wrong way.
4. Find and work with the optimal amount of people. With too small of an amount the result are not representative and a huge amount of people might be time consuming or expensive.
Our optimal amount of users is 5–7 people and certainly no more than 10. If you have major problems in your app, you’ll realize it after 4th or 5th person.
5. Identifying the problem doesn’t necessarily mean the identification of its cause. You have to dig deeper. This is probably the most important point.
Why are people are not sharing? Is the sharing button hidden? The button doesn’t look like a sharing button? Or people just don’t want to do it on your device while you are watching?
JC: Wow, that is extremely helpful. What is the next step for the product? Besides making the initial changes, how do you hope to use your learnings moving forward with product development?
As of a few weeks ago we are now part of a Snapchat team. Lot’s of celebrities are now using the app, such as Miley Cyrus (Miley Cyrus using Looksery Technology), Kim Kardashian, John Legend, Jimmy Fallon, models of New York fashion week. But I talking about Snapchat is a whole different story
My personal goal is to share our design experience, showcasing our mistakes and attempts to make it better, to help others in their design process.
People may or may not do user testing just rely on their own design experience, but I hope my experience will help someone in creating new awesome projects.
