- September 26, 2018
- 35 min to read
Photolemur's Red Dot Award experience by Nazar Begen
Author: Nazar Begen
Photolemur UX. Lessons and Red Dot Award.
For me, the goal of interface design is to help users get results from software or a device in the easiest possible way.
This is the first, most basic level of interface design: solving a user’s task so that they don’t even have questions like What should I do? or Where should I click?
The second level is the individuality of the design: things which help users distinguish between brands.
A good design starts with solving the first-level tasks and then moving to the second level. Unfortunately, the world of products and interfaces is full of design experiences where being different was prioritized and basic tasks weren’t solved.
Last week I became a father. My friend gave me a baby car seat. When it was a time to bring our baby home, I decided to install the seat. There were no instructions. I spent 40 minutes trying different options. None of my friends, two of whom are fathers themselves, could figure it out. They had experience installing ISOFIX seats. But this seat needed to be secured with seat belts. We didn’t manage.
The seat is cool and it’s made by a good manufacturer. For two days, I was terribly angry at it and even thought that there weren’t enough parts and went to the children’s shop to buy something :-)
Later, I spent an hour and a half and eventually figured out how to install it. It turned out that it’s super easy but not obvious.
This is a vivid example of a good product with bad onboarding.
Now, think of every office coffee machine you’ve used that can do cappuccino. What was your first experience? Did you know where to press when you first used it? I believe that there’s a huge unoccupied niche for coffee machine interface designers. You can just start making user-friendly screens for coffee machines and make millions!
For me, my first UX lesson took place in 1998 when I went to the Physics and Technology Lyceum. There was a mathematics lecture on the fifth floor of our university. From the windows, you could see the pathways (neat, paved with tiles) and the beautiful lawns (rectangular, planted with grass). The only problem was that all the rectangular lawns had paths worn on the diagonal from all the foot traffic. When we studied geometric formulas, our teacher pointed out the window and said that this diagonal was the hypotenuse. And then she added that people are always looking for the fastest way. The path through the hypotenuse is the shortest.
Later, when I learned about user experience, or UX — the concept introduced by Apple to describe the whole experience of human interaction with a product starting with the purchase and the first use — I immediately remembered this story with paths and the shortest way.
UX is an absolutely relative concept that depends on what experience your user has already had. A good UX assumes that you know something about people and offer them the most obvious way to work with your product.
“We’ve created a new user experience” is the most ridiculous phrase in the design field. You can’t create an experience. You have it or go through it.
This understanding leads us to the problem faced by all innovative products on the market. If we create something new, we have to create a way to work with it. And here one more principle comes into play.
For nearly all living beings, anything new is stressful. A new job is stressful. Meeting new people is stressful. Installing a new baby seat is stressful. A macOS or Windows interface update can be a great source of stress. Where are my photos now? Even a birthday surprise at work can be stressful. Everyone copes with stress to a greater or lesser extent, and sometimes we don’t even notice that we’ve experienced stress.
But when we work on product interfaces, it’s critical to remember that new experiences are stressful. It’s also important to answer these questions: How can I make the user experience stressless? How can I make sure that working with a product, whether physical or digital, comes with a minimum of discomfort?
Let’s move now to the real case study.
At the end of 2016, I joined the team of amazing specialists at Skylum (formerly Macphun). They already had tremendous experience in the software market for professional and semi-professional photographers and had developed very cool technologies for photo processing as well as a bold idea: to create a program that would automatically process images and make them look great. We’re talking about the mythical make awesome button.
At the very beginning of my career, 15 years ago, I had experience working in photo studios where I processed photos in Photoshop. We handled dozens of photos per hour. Hundreds and sometimes thousands of photos per day. In those days, I dreamed about such a program.
Understanding the potential for this software, I joined Photolemur as a project manager.
We all realized that we were creating not just a new photo editor but a new type of product in general. For us, it was a problem even to figure out what to call this type of software: an automatic enhancer? an anti-editor?
At that time, the first algorithm and a very standard interface, which resembled the average photo editor, was ready. That’s where we decided to start.
Note: Many companies have UI/UX departments, product specialists, and development departments. And they all exist in parallel realities. Worst of all, the marketing and sales departments, in general, have no relation to the development department. I believe that when it comes to a commercial product, the company must have one clear goal that all departments are helping to achieve, each with its own tools. Products are created to be sold, and they aren’t bought by a target audience but by real people. People buy products that help them. To create a product that helps people better than others, you need to listen to people, get feedback and then properly digest it; this can be done even by the development department. You need to make sure that people who work directly with users have a direct role in product development.
In our case, we decided to launch the product with the interface that existed solely for internal testing. We did this to get quick feedback and understand what to do next. We focused on sales and marketing first, then on a new UX and other improvements.
Thanks to the first quick campaign, we got thousands of real users (who paid money!)and got real feedback.
Work in depth
Problem 1: Standard interface
The interface had standard controls like a photo editor. Visually, users wanted to find certain settings, and when they didn’t find them they were unhappy because they thought that the program was incomplete.
Problem 2: Processing speed
The first algorithm was several times slower than the current algorithm. But even so, processing was achieved several times faster than manual processing, for example in Lightroom. We conducted tests to confirm this. The problem was that in Lightroom during manual processing, a user doesn’t count every minute: the time is spent in active work, so it goes unnoticed. In Photolemur, you had to wait for 20 to 40 seconds. This was annoying users.
Problem 3: Cool technology that isn’t visible
There are super cool things happening inside the Photolemur software during processing: definition of objects, face recognition, the application of twelve different algorithms to different photo zones depending on what needs to be corrected. But for the user, all this was just a progress bar. And the wait time seemed too long for many users.
So we decided to make the 2.0 version of Photolemur.
When we started work on the new product, we knew that we owned a unique technology for automatic photo enhancement that needed to be packed in a super laconic interface that would show the value of our algorithms and leave the user satisfied after working with Photolemur.
Designer Dima Novikov (the art director at MacPaw) suggested the concept of a so-called non-interface design. Since the idea of our technology was to make users stop editing photos and just start getting good results, the actual interface of the program became the photo that the user drags into the program. Accordingly, the size and frame of the program is the photo itself.
And the first window of the program that a user sees after starting the program is a simple window with the Photolemur logo and an Import button. It’s impossible to do something wrong!
To keep the photo the main element, all buttons and auxiliary elements of the interface were designed in black and white. There are no elements that distract.
The next step is the process of photo enhancement, which is complicated on the inside but not visible to the user. How could we show a user what happens to a photo? Dima had the idea to simulate a surface where a photo sinks for processing, like a bath for chemical photo processing. The photo goes into the bath and the magic processing begins. However, we kept the idea of a plexus, or a network of nerves, and decided to simulate neurons in the brain. The neurons give a clear understanding that what’s happening is some kind of complex technological process, and you can also play with them! Move the cursor on the screen and the neurons will follow it. The important point is not to leave the user lost and guessing what’s happening. Therefore, over the neuron plexus we listed the process that was occurring within the program: face recognition, finding the sky, improving the white balance, etc.
We removed all additional features that were present in the first version: cropping, processing power settings, and export settings. Now there’s automatic processing and nothing else. The focus is on what we do uniquely.
We launched a new version of the program and received a bunch of positive reviews. But this was just the beginning. Together with the good reviews, we got negative ones.
One of the things that we decided to do differently in the 2.0 version was the pricing model. While the first test version was sold as a license, 2.0 was launched as a subscription SaaS app with a limit on the number of photos you could process. Formally, this is part of the business strategy and we should talk about it in another article. But if we look at this from the user’s point of view, it’s part of the user experience — the experience with our product. It’s just as important as the graphical product interface.
In the first month, we made thousands of sales; in the second, hundreds; and in the third, we got about 100 new subscriptions. We had to do something. We conducted a survey among existing users and among 50,000 users of the free version. Here’s what we found:
1. People who wanted to use the product didn’t want to purchase a subscription for a desktop program. What had succeeded for Adobe wasn’t working for us. At the same time, people were ready to pay a higher price than the one we had charged for Photolemur 1.
2. People wanted control. We gave our users the coolest result by processing with artificial intelligence, but they wanted to have tools to influence the final result. This was a very important discovery that can help you if you’re working on automated intelligent products and systems. When you solve a difficult task of automating some kind of habitual process, leave the person the right to make the final decision. Otherwise you’ll create stress!
So we released Photolemur version 2.2 Spectre (2.1 was a mini update with bug fixes and additional languages).
Funny story: A week after the launch of Photolemur 2.2, someone in the office said that Photolemur was featured on the Mac App Store. Once a month, Apple selects programs which are placed in the section Loved by Apple — a block in the App Store that all users see when they come to the store. The funny thing is that at that time we only had the first version of the product available on the App Store — in fact, it was the beta version. So we appeared on the main page and our sales started to grow sharply. We decided that in future we would no longer support the 1.0 version. But if we removed the program from the App Store, we would lose the opportunity given by Apple. In less than a week, we released version 2.2 on the App Store. Version 1.0 was still available for sale. But all Photolemur 1.0 users who contacted support could get Photolemur 2.2 for free. After three months, Apple featured us again in the high season — Black Friday and Christmas — and for almost eight weeks our program was among the first three programs that users saw when they came to the Mac App Store.
Let’s go back to the Photolemur 2.2 release and the UX. One of the best decisions at the launch of 2.2 Spectre was getting rid of the subscription system and selling the product as a one-time purchase — simply put, selling licenses. The number of sales immediately increased. Moreover, people in the comments on our social pages wrote about how cool our company was for listening to its users! This was an incredible result for me. And although this part of our experience has nothing to do with the theme of interfaces, it has a direct relation to the UX. People gained experience with our company. Now they knew that we listened to them, thought about them, and wanted to make their experience with our product better.
The most important innovation in the product was a new slider that controlled the strength of the artificial intelligence.
We had many discussions about whether we should implement this slider in the program, which should do everything automatically. But after surveying and analyzing user behavior, we realized that people should have control. And we made it very simple and easy to work with.
Here are other things that we changed in the interface and our reasons for doing so:
1. Our CTO and the person who came up with the idea for Photolemur, Dima Sytnik, is focused on literate and intuitive design built on Apple’s guidelines. We repeatedly redesigned the system and context menus so that they met the guidelines. This also applied to the placement of items and their names. Next, we made the version for Windows based on the Mac version and, as a result, the program turned out to be much more stylish and intuitive than is usually the case with Windows products.
2. Version 2.0 had onboarding: four consecutive screens that taught the user how to work with the product. When we were working on version 2.2, Dima Novikov commented:
“All people are like children: they don’t want instructions. Look at the kids. When they get a new toy, they start playing with it right away.”
We decided to remove the onboarding from the program altogether. But the open question was how to make it so that anyone could start playing with the program right after launching it. Not everyone has a photo they want to process or improve right away. So we added the Open Sample Image button to the startup screen. We placed several sample photos in the program that are randomly opened. This lets people play with the software immediately after they install it.
During a survey, we found out that 85% of our users were mostly photographing nature. Therefore, the test photos are mostly nature shots and show vivid examples of sky enhancement, a technology which defines and automatically improves the sky. This feature works just fantastically, and we didn’t miss the opportunity to show it off.
3. In Photolemur 2.0, we actively collected analytics including analysis of program errors. Photolemur allows you to process photos that are at least 600 x 800 pixels. Technically it’s possible to process smaller photos, but super small photos contain too little information to substantially improve them. And it’s often possible to even worsen them. Therefore, we added this restriction. When loading small images, the program gives an alert about the impossibility of processing such a small photo. After analyzing the data, we found that 95% of all errors were about importing small images. In version 2.2, we added brief information about the minimum size of the image under the Import button.
4. Another important update that we made because of our analytics was better technology for JPEG processing. Our first algorithms worked perfectly with RAW photos. We used the same algorithms for JPEG and BMP, and for these formats they worked well but a bit worse than with RAW. Our analytics showed us that with nearly 1,000,000 versions of photos processed in Photolemur 2.0, 880,000 were JPEG. This was unexpected for us because we thought that owners of DSLR and mirrorless cameras (like those from Sony and Fuji) would use the product the most. After receiving this data, our technical team developed new algorithms that now work differently for different formats. So for example for JPEG, additional algorithms for cleaning artifacts (pixels, dots) are applied.
After all these changes, conversions from the free to the paid version increased from 3.5% to 7%, i.e. a 100% increase.
The second level — individuality.
Our designer came up with a couple of things that distinguish Photolemur:
1. The lower right corner of the program is atypically rounded. All other programs have either straight angles or the same curvature for all angles.
2. Photolemur uses black and white navigation elements. The color is only in the photo. This is especially noticeable on the slider. It’s like a yin-yang symbol. The Photolemur logo in the first window also has a special monochrome version.
3. The plexus that simulates neural connections is also unique. At conferences in San Francisco and Hong Kong we met users who love watching our enhancing process!
UX lessons that I’ve learned from this year’s work on Photolemur
1. We don’t know much about the experience of our users until we let them try the product. You have to create sandboxes, let users play, and closely monitor the results.
2. Good UX and design require not only experiments but also a great experience. Try to work with the best specialists. It sounds inefficient, but try to hire highly skilled people for such a job as UX even if you can’t afford it.
3. People are like children. They want to get the result right away, never read your instructions, and always want more. Think about how to solve this problem in your product. Especially think about onboarding.
4. People want to have control. Even automatic super-intellectual systems must have a certain level of control. People are afraid of Skynet. Give them at least an illusion of control over Skynet.
Note: The control slider is actually used by no more than 12% of Photolemur users. All the others are just happy that it’s there.
The UX itself doesn’t solve all product issues. What’s inside is important. On the flip side, cool technology alone without a thoughtful UX and interface design won’t find any users. A good product is a coordinated team effort, with each part of the team solving common tasks.
This is a critical point: sometimes designers can offer fantastically beautiful solutions that can’t be implemented technically. I pay special attention to this because I’ve faced the situation when designers have said: I thought of this amazing design but these idiots can’t implement it!
But interface designers aren’t responsible for drawing on paper or making sketches but rather for the final product just the same as developers, testers, and managers. If a designer draws or designs something that can’t be implemented by a technical team, then it’s not their problem but the designer’s problem as well. In that case, everyone has to sit down together, discuss everything as it is, and maybe come up with everything from scratch. Together.
With Photolemur, we had everything. We had a cool idea that the technical team knew how to implement. We worked with a cool designer who, in addition to bringing lots of experience, was able to listen to the marketing and technical teams. And we worked on the product iteratively, in each subsequent version solving specific user problems.
Just don’t think that everything was simple all at once. We have repeatedly fought and redefined the product. But the result is worth all the nerves expended.
Lastly, Red Dot was for me personally a very important award, one which I have long wanted to receive.
This is not my personal award. The Red Dot Interface Design Award is given for the interface design. This award is primarily thanks to our designer and development team that together managed to implement an awesome product.
But as the one who manages the process and has been communicating with users from the very first version to the award-winning version, I’ll be proud to stand up for the award.
P.S. Last week we released Photolemur 3. There are even more interface updates and cool things!