Bluemine
THE PROBLEM TO SOLVE
BACKGROUND
USER PROBLEM
My ROLE
As Bluemine’s product owner and design lead, I drove project vision, set our product roadmap and priorities, and directed four design workstreams.
I managed a design team of 14 UXers on a product team of ~50. I scoped and allocated work; pushed our team’s craft through weekly design reviews; and advocated design decisions to stakeholders and partners.
I partnered with Bluemine’s tech lead, two product managers, and two marketing/analytics leads to plan sprints, prioritize features, and make product decisions.
I championed my team’s work internally across IBM Design as well as shared the success of our redesign externally by writing and submitting our work to StepTwo and Nielsen Norman Group for award consideration.
🔍 INSIGHTS
study
RESEARCH PURPOSE
*After launching an IBM survey with 527 completed responses, 40 IBMers volunteered to be interviewed by our design team.
RESEARCH METHOD
HIGH LEVEL INSIGHTS
Bluemine’s experience was inconsistent and outdated.
Many landing pages often got in the way of getting directly to the desired content, especially for pages that were rarely maintained.
Search didn’t always come back with relevant results; at times, reports dating back five years would appear in top results, thwarting user expectations.
Finding market answers was time-consuming. Because of how search was structured, IBMers often had to read through many reports to find a key nugget of information.
Publishing was time-consuming and complicated, which dissuaded many IBMers from contributing to Bluemine.
Publishing took 30–90 minutes to post content that had already been written as a report, and another 24 hours before the content would go live.
The publishing tool’s categorization and selection criteria were out-of date and complicated to use.
ChallengeS
#1: We struggled to figure out the “right model” and design for search.
IMPROVING SEARCH
Because of search’s design structure and outdated algorithm, existing metrics weren’t useful. User interviews supported our assumptions that the rigid design forced people to behave in a way they didn’t want to
Users cited Google as a complementary way of getting market intel, so we used Google search and filters for design inspiration
PRODUCT RISKS
Designing an interface on top of a new algorithm (that didn’t yet exist) without know what results might look like
Simply copying Google’s search design without the power of its search algorithms would likely mislead user expectations
#2 We struggled with politics and relationship management when the redesign included culling and combining content.
CULLING CONTENT
A redesign of our content ecosystem affected 668 pages, 126 tools, 1800 internal active reports, 88k links to external reports, 50 content owners, 350+ IBM publishers, 16 executive stakeholders, and 125k users
Over 30% of pages had not been updated in over two years, but we could not simply kill off unused content
PRODUCT RISKS
Unintendedly killing off valuable content due to Bluemine’s design weaknesses biasing existing metrics
Potentially damaging relationship between organizations who contributed to Bluemine’s repository of insights
Our Solution
SIMPLE, BUT COMPREHENSIVE
WEB-RESPONSIVE, CONTENT FLUID
DESIGN IMPACT
HOME
BEFORE - 2016
after - 2018
SEARCH
BEFORE - 2016
after - 2018
Publish
BEFORE - 2016
after - 2018
🔍 INSIGHTS
10-10-10 STUDY
RESEARCH PURPOSE
RESEARCH METHOD
HIGH LEVEL INSIGHTS
In the first 10 minutes, users were positive about the redesign. They commented on bringing Bluemine to modernity and simplifying the experience. Despite positive feedback, three areas of improvement emerged:
Price confusion: 12 of 12 users were confused about the interdepartmental charge and overall pricing of Bluemine.
Search was not prominent enough: 6 of 12 users did not initially notice the search box on the homepage.
Search relevance: Users who entered compound search phrases such as [topic + geography] did not feel search results were relevant enough.
After 10 weeks, only 7 of the original 12 users could participate due to holidays and scheduling conflicts. Three themes emerged:
Only frequent users noticed feature changes and could articulate those distinctions between Round 1 and Round 2 of 10-10-10s. These users a) used Bluemine daily, and b) primarily focused feedback on parts of the experience that were not meeting expectations.
Users still struggled to start their journey beyond the search bar, and search relevancy was still an issue.
Of the 7 users, 4 had changes in job role, scope, or project work. We noted that Bluemine team must be mindful of the fact that many IBMers change roles within the company, but search behaviors and needs of Bluemine didn’t. The team should minimize categorizing or gatekeeping areas of Bluemine by job role.
“[BLUEMINE] is one of our better efforts as a company.”
— IBMER, 10-10-10 STUDY
retrospect
SUCCESSES
FAILURES
Transparency to the wrong audience
In the first month of launching the redesign, there was confusion around Bluemine subscription costs. The interaction struck fear in IBMers who didn’t understand the back-charge and worried they would get in trouble with their managers.
LEARNINGS
Test microcopy early and often
While we spent a lot of time testing prototypes, microcopy was always the last step to be completed. We learned that writing good copy is an important part of design, which can make or break the user experience.
Simulate real-life scenarios
Testing the new search design was often done in a controlled environment as cognitive walkthroughs or with designated search prompts. Having users search with a small database of only 2,000 reports helped us get a more accurate read on if search design was meeting expectations.