“What do you think of when I say the word design?” Margaret Gould Stewart, director of product design at Facebook, is here to talk about the kind of design that you normally don’t think about — the design of digital systems that are used by billions of people each day.
As examples, Steward reminds the audience that Google handles 1 billion searches per day. People upload to YouTube more in a single day than all of the US television networks broadcast in the last 5 years combined. Facebook transmits the photos, messages and stories of over 1.23 billion people, or about 1/6 of humanity.
“What’s really hard at designing at scale,” she says, “is that it requires a bizarre combination of two things, audacity and humility.” Audacity to believe that what you’re doing is important, and humility because it’s not about the designer’s portfolio, but about the people they are designing for. It’s also a new problem. “There is no design school that offers, ‘How to Design for Humanity 101.'”
That said, she and her colleagues have learned a lot about how to design at scale. She shared three lessons.
Lesson 1: No such thing as a small detail
Facebook recently had a problem: They wanted to redesign the Like button — the little blue thumbs-up that lives on outside websites. It seems simple, but it has to work all over the place, in an incredible number of contexts. “Designing this button was a huge pain in the butt.”
The designer who did the work put in an estimated 280 hours to redesign it. Why so much time for one button? Because, “When you’re designing for the world, there’s no such thing as a small detail.” The Facebook Like button and the Share button that accompanies it are seen on average 22 billion times a day and on over 7.5 million websites, Stewart says: “It’s probably the single most viewed design element ever created.”
Lesson 2: How to use data
Facebook has a huge amount of data about how its users behave, “but it’s not as simple as following the numbers.” As an example, Facebook once had a very simple tool for reporting photos for spam and abuse. But they found that only a small percentage of photos reported were actually violating the community standards.
It turns out a lot of them were people asking for photos of themselves to be taken down, “This was a scenario that the team never took into account before.” So they added a tool to request that the uploader take the photo down. But that didn’t work — only 20% of people used it. So they went back again. This time they consulted with experts on conflict resolution, and learned that they actually had to help people express to their friend how the photo makes them feel, not just request that they take it down.
Now when someone goes to report a photo there is an option to take it down because they don’t like it, and it’s embarrassing. The tool then provides them with language to send in a message to the friend who posted it. This had a huge impact. Before, 20% used the tool; now, 60% do. Surveys also showed that people on both sides of the conversation were happier. That same survey showed that 90% of your friends want to know if they’ve done something to upset you. (“I don’t know who the other 10% are,” she says, “but maybe that’s what our ‘unfriend’ feature is for.”)
She is clear that at Facebook, “we use a lot of data, but also iteration, research, testing, intuition, and human empathy. It’s both art and science.”
She also says that the term ‘data-driven’ drives her team nuts. She say, “It would be irresponsible of us not to rigorously test our designs, but the fact is, data analytics will never be a good substitute for design intuition. Data can help you make a good design great, but it will never make a bad design good.”
Lesson 3: Manage change carefully
Stewart also spends a lot of time designing the introduction of change. “People,” she says, “can become very efficient at using bad design. Even if a change is good in the long run, it’s still frustrating when it happens. Especially true of user-generated content platforms, where people can rightfully claim a sense of ownership.”
Years ago, when she was at YouTube, they were wondering how to get more people to rate videos. They had a five-star rating, but they discovered that people mostly rated things with all five stars, a few of them used one star, and almost no one used the three stars in the middle. So they wanted to switch to a simpler up/down model.
They believed the change was right, but knew people wouldn’t like it. So, before the change, the team published the data on their blog and engaged in conversation with the larger industry.
The whole world
These lessons are helpful, “but they won’t mean anything if you don’t understand something more fundamental. You have to understand who you’re designing for.” At some point you run up against the walls of your bubble — such as the bubble around Silicon Valley workers, who get crabby when they lose 4G service for 10 minutes. “What if,” she asks, “you had to drive four hours to charge your phone? What if you had no access to public libraries? What if your country had no free press? What would these products mean to you?”
She shows a photo of several common “feature” phones, non-smartphones, and reminds us that this is what the internet will look like for the next 5 billion people to come online. She says, “If you want to design for the whole world, you have to design for where people are, not where you are.”
She says designing on a global scale can be difficult work, and the humility part can be especially hard. “Almost everything that I’ve designed in my career is gone, and everything I will design will fade away. Here’s what remains: the never-ending thrill of being a part of something that’s so big that you can never get your head around it.”