Defining the Damn Data
@nstopUX
Jen Matson Sr. UX Designer, RealSelf
Photo: https://www.flickr.com/photos/rh2ox/9989876925
We are:
The world's largest community for learning and sharing information about cosmetic surgery, dermatology, dentistry, and other elective treatments.
In 2014, our users:
• Numbered 51 million (unique visitors)
• Viewed half-a-billion pages
• Spent 5 million hours in their research
• Are predominantly mobile: 72% and growing
Core team: UX designer, product manager, developer(s)
Supporting players: Business analyst, community manager(s)
Team
Agile process: 2-week sprints, PM and design at daily stand-ups
Bite-sized work: Build and launch many smaller things, test, iterate, improve
Project approach
Ongoing data sharing
Weekly meetings:
• Business analysts present Google Analytics data to full product team
• Community managers share user insights
How do we interpret the data?
• What’s wrong, or where can we do something great?
• Why do we think this might be worthwhile?
• Do we have any data to support this?
Step 1: Define the problem / opportunity
• High traffic, but high bounce rate - most visitors view 1 page, leave
• We have highly relevant content, but it’s not easily discoverable
• Issue identified via analysis using Google Analytics
Step 1: Define the problem / opportunity
• What’s the profile of auser visiting this page?
• What motivates this user, and how does that intersect with what we have to offer in our product?
Step 2: Define your users
• Someone Googling for answer to a specific question about a medical treatment
• Motivated to the get that answer first and foremost
• Unaware of RealSelf brand
Step 2: Define your users
• Reduce bounce rate, converting some portion of casual users to active users
• Active users contribute content that increases value of site for all
• Increase awareness of RealSelf brand
Step 3: Define the desired business result
• Focus on getting the user their answer quickly, easily
• Provide relevant related content for those motivated to dig deeper
Step 4: Identify the highest-value work
• What content, interface and functional changes will we make?
• How long will it take to design and build?
• What can we deliver quickest, of highest value to the user?
Step 5: Define what to design and build, and how
• Remove irrelevant content/UI so we can move valuable content further up the page
• Tweak existing design for initial experiment
Step 5: Define what to design and build, and how
• What data will we use to determine success or failure?
• What do we have the ability to measure now?
• How many people will see this and on what devices/in what regions?
Step 6: Define metrics, measurement, launch
• Show reduced bounce rate % in Google Analytics
• Increase clicks on specific links to related content
• Target mobile for initial test
• Expose to a small % of users to start
Step 6: Define metrics, measurement, launch
• Create final visual design
• Build design in Optimizely using HTML, CSS and JavaScript
• Set up click and other goals in Optimizely keyed off of defined user behaviors
Step 7: Design, build and instrument
• Someone Googling for answer to a specific question about a medical treatment
• Motivated to the get that answer first and foremost
• Not engaging with doctors at this point in their journey
• Photo upload was clunky, so users created multiple review entries for multiple photos
• We made uploading photos easier, data showed sharp drop in reviews
• Initial data was misleading - we didn’t break anything, we made the experience better
Always test before you launch to all
1. Define the problem / opportunity
2. Define your users
3. Define the desired business result
4. Identify the highest-value work
5. Define what to design and build, and how
6. Define metrics, measurement, launch
7. Design, build and instrument
8. Launch, observe, learn