Research-driven · Product Design · AI-powered
Research-driven · Product Design · AI-powered
Salesforce Trailhead is a free learning platform used by millions of professionals to get certified and build careers in enterprise software. Through guided modules, hands-on challenges, and community forums, it takes beginners to certified experts.
Over 10 months, our team partnered with the Salesforce Trailhead product team to research, design, and prototype new features for the platform's most experienced users, the power users who built their careers on Trailhead and then quietly stopped coming back.
.jpeg)
User Research
To understand why expert users disengage, we conducted two rounds of semi-structured interviews with 11 power users across 5 regions - exploring their habits, frustrations, and what would actually bring them back.
What we found about advanced users:
Disengage after hitting major milestones like certifications
Mentioned supplementing Trailhead with other tools (ChatGPT, YouTube, Slack)
Found content to be too shallow for advanced users
Were unhappy with existing personalization recommendations
"How might we drive advanced users to keep returning to Trailhead with deeper, relevant, and career-aligned experiences?"
Ideation
Ideation began with five engagement requirement areas: post-milestone engagement, career visibility, habit-forming systems, community integration, and real-world problem solving. We used these as constraints - not prompts - to find the smallest intervention that would create a genuine reason to return.
Team ideation sessions: whiteboard synthesis, affinity mapping, concept development
Solution
Reimagined as an experience that builds real-world readiness, keeping expert users engaged beyond milestones.
01 Solution 1
An AI-powered workspace where expert users describe a real Salesforce problem in plain language and get a structured, expert answer calibrated to their certification level and Salesforce release. Includes an expertise depth toggle, version-specific flagging, and a one-click community escalation path.
How does it work?
Type your problem in plain language. Get a specific, actionable answer from Salesforce's full knowledge base, calibrated to your expertise level.
Adjust the answer depth to Novice, Intermediate, or Professional so you're never over-explained or underserved.
Still stuck? Solve drafts a Trailblazer Community post for you, pre-filled and ready to publish, so you get an expert answer either way.
02 Solution 2
An adaptive career layer that tells you what changed, where you stand, and what to do next. One personalised action per visit, tied to certifications, recent Solve sessions, and Salesforce release changes. Each visit becomes more valuable than the last.
Trails recommended based on chats with Trailhead Solve.
Agentforce identifies trending skills for a user's role, shows their current standing in the market, and enables them to generate custom Trailmixes to close skill gaps.
Context
The engaged state. Users join Trailhead to learn: completing trails, earning badges, and working toward certifications. Engagement is high and the loop is clear.
The drop-off. Once certified, users hit a wall. The content feels too basic, and when real work problems come up, they turn to ChatGPT, Google, or Slack instead of Trailhead.
Closing the loop. Trailhead Solve brings experts back with accurate, Salesforce-backed answers to real problems. The Personalised Dashboard then nudges them toward relevant trails, turning a one-off visit into re-engagement.
Testing & Impact
We were not measuring task completion speed - we were measuring whether the product creates a new motivation to learn.
We built 12 AI agents from our research data to simulate power user behavior and stress-test prototype flows before any real participant encountered them.
8 Salesforce power users, ~30 minutes each. Where participants would normally use ChatGPT, we asked them to try Solve first - giving us a live substitution signal.
Would use Trailhead Solve over ChatGPT for Salesforce-specific questions
Average confidence rating in Solve's answers
Said the Dashboard surfaced insights they wouldn't have explored on their own
Expressed unprompted return intent after the session
Reflection
Focusing on power users who had already arrived gave every design decision a clear test. If a feature could apply to a beginner, it was wrong. That constraint produced sharper work than any open-ended prompt could have.
Recruiting authentic power users within a semester timeline is genuinely difficult. These are full-time professionals across five time zones who have no particular reason to give a student team 30 minutes.
Scale testing globally with certified professionals. Refine Solve answer quality using structured feedback loops. Pitch an integration roadmap to Salesforce Trailhead's product team.
Demo
