UIE Web App Master’s Tour – Seattle, Washington – May 23, 2011
Facebook likes the start-up culture. They believe in small teams: photos, engagement, etc. Each team is treated like a small company with a product designer, researcher, engineers & a product manager.
Facebook uses data to form a lot of the decisions that are made. Data helps understand how users use product and how they can be optimized.
Facebook had great ideas, built products and then left them to move on to the next thing.
Were using an old photo uploader tool built in Java, didn’t work well and was poorly designed. Built their own tool, which required a plugin. Tested it and only 34% users successfully uploaded photos. Went back to the drawing board. Research showed that people were bailing at the plugin install step.
They re-did it, using Flash, but decided to completely re-do the whole process. Just changing the flow (removing plugin) raised the success rate to 45%. 85% of users were only uploading 1 photo. When users uploaded only 1 photo, they presented an interstitial with instructions on how to upload multiple photos. 1 photo users went down to ~40%.
Use data to Sanity check.
text box to input status at the top of the feed.
Added links for different options. Status updates went down by 2%.
They did a lot of changing and testing, but nothing changed user interaction significantly.
Qualitative testing can help understand the user’s intent and how they feel. What do people think about the company, the web site, the process, etc.
Facebook users ramp up their usage during a life change, like getting married or having a baby.
Facebook is very interested in local businesses. That’s why they have the pages products. Interested in small businesses who don’t have social media teams and web resources. Current focus – talked to owners on well-known small business section of San Jose. They don’t want to make a lot of money, but they want to influence their community and create a space. This will influence the path of their pages product.
Focus groups helped them understand users’ concerns with privacy. These focus groups helped them create settings and set defaults.
Use data to evaluate success of certain metrics-oriented projects.
In 2007, facebook created a team that was solely focused on growth. One of their most successful experiments.
Findings: First group to utilize data to create road maps and direction.
Friend invitation form: not too bad, but could use some change. Separated the form into two steps, the second being the captcha. Increased conversions 3% (9,000,000 users).
Deactivation page: very clinical, wanted to make it more emotional, new design reduced deactivations by 7% (1,000,000 users). Have always offered deactivation vs. complete deletion.
Be wary of being overly data driven. Facebook tries to be agnostic. They don’t want to have a personality or a voice because they want users to associate facebook with their friends, not the company itself.
Facebook health dashboard – track all kinds of data real-time throughout the day.
It’s difficult for a set of metrics to fully represent what you value. Consider qualitative data, quantitative data, strategic goals, user interests, network interest, competition, regulatory bodies and business interests.
Modeled after the Growth team. First attempt at engagement was reads and writes. Added like button on comments. 85% of reads and writes are generated by 20% of users.
Adding the application menu increased clicks, but not as much as the goal. Facebook was afraid the like button would cannibalize commenting, but commenting actually went up, as well as user interaction with “likes”.
Facebook believes that real innovation invariably involves disruption. Looking at data isn’t going to help.
When Facebook first started, everyone bounced from the home page. No one was using the data on the page. That’s when they came up with the newsfeed. Everyone hated it. 10% of users joined a group called “we hate the newsfeed”. Over time it was a valuable change and is now accepted as a standard.
People didn’t realize that things were going to be published on their pages.
When you take big risks, you make mistakes, but you also can make great decisions.
Facebook has always taken risky decisions, but when you have 700,000,000 users you study the data a little bit more carefully. The greatest risk is taking no risk.
Quantitative Data helps you understand how users use product, how they can be optimized and whether certain projects are successful or not.
Qualitative Data helps you understand and validate our assumptions about the user’s intent and how they feel.
Pure metrics don’t capture all values. Over-reliance on data leads to micro-optimizations. True innovation doesn’t come from being driven by numbers.
Tutorials – user education – more users – let’s users know when there is something new.