Skillshare
Conversational UX
Skillshare offers thousands of creative classes, but getting to the right class and resolving account questions often sends users to static FAQs. During discovery, my team and I saw no first‑party AI assistant on core help or browse surfaces. We hypothesized that a conversational assistant could reduce “hunt-and-peck” navigation and offload simple support tasks. Read through this case study to discover how we tackled the question, "How might we help learners quickly self‑serve membership tasks, find relevant classes, and get timely help—without leaving the conversation?"
New and returning Skillshare users juggle three common needs: sorting through thousands of classes, fixing account issues, and finding clear answers. In benchmarking, we didn’t find a first‑party AI assistant on key browse/help surfaces, suggesting room for a conversational layer that reduces navigation effort and improves self‑serve resolution.
Design challenge: How might we help learners complete common tasks in a single conversation—quickly, safely and with a tone that feels on‑brand for Skillshare?
Manage Membership: plan changes, billing info, pause/cancel clarity.
Browse Courses: topic search, refine chips (level, time, tool) and quick saves.
Help Center: policy answers with clear next steps and optional human handoff.
Research Kickoff: Heuristic review of Skillshare’s public surfaces, lightweight language sampling (queries from learners) and a competitive scan of education chat experiences. We clustered intents into the three pillars above and mapped edge cases before drafting copy.
Users worry about surprise billing and unclear outcomes when changing plans.
Learners need fast, relevant classes without deep menu digging.
Articles answer what, but not what to do next.
Summarize impact before change (dates, access, next charge) and require an explicit “Yes.”
Start with a short probe (topic, level, time budget), then present top 4 topics followed by "browse all courses" with quick actions: find courses in ____. Offer further exploration chips: search specific skill, search for a specific creator
Retrieve answers (articles/policies/how‑to) and pair each with a link. Guide users with next steps.
Browse → find course
User: “can you help me find a course.”
Assistant: “Of course! I can assist you in finding the perfect course.”
User: “I'm interested in art.”
Assistant: “We have a range of courses in the Art & Craft Category, here are a few of the top options.” (offer different courses with quick actions and load more chip)
Membership → safe cancel membership
User: “cancel my membership”
Assistant: “We’re sorry to see you leave! To proceed with your cancellation, I need to verify your email address. Could you confirm your registered email address?”
User: “sample-email.com”
Assistant: *confirm email address* *cancel membership*
Help → policy + CTA
User: “i want to buy gift cards”
Assistant: “A gift membership is a non-renewing membership, giving the recipient unlimited creative learning on Skillshare during the membership term.Skillshare Gift Memberships are available in 3-month, 6-month, and 12-month durations. At this time, gift memberships can only be purchased if you're located in the United States. Would you like to purchase a gift card?" (yes or no chips)
Outcome: A working Voiceflow prototype that demonstrates end‑to‑end completion for the three pillars and a repeatable repair pattern.
If I had more time… I’d run moderated tests with active Skillshare members, refine the help retrieval prompts, and expand edge‑case handling for billing.