Case Study: 

Forging a Warlord

Playtest 1 (n = 30) | Playtest 2 (n = 27) | Playtest 3 (n = 29)

What was the motivation behind the product pairing push?

League of Legends is a live-service game with 100 million global users and over $1 billion in annual revenue. This project focused on testing a new character, Ambessa, for League's team-based multiplayer strategy game, where each player picks a unique champion to play.

For users

A new champion with a high skill ceiling and multiple complex mechanics, intended to freshen up gameplay and give advanced players a new playstyle to master.

For the business

Champion releases are League's primary driver of player engagement, and a significant share of League's marketing spend goes toward them. Each champion carries the potential for millions in new revenue and reinforces the connective link to the broader IP and adjacent media.

The underlying assumptions

Ambessa's multiple complicated mechanics will appeal to advanced players, and they will find that complexity rewarding to master.

Research questions

• Can Ambessa succeed as a new character release?

• Is her combat loop intuitive, and is she accessible to lower-skilled players?

• Does the intended gameplay style match player expectations?

• Are her abilities visually clear, and can players distinguish between abilities in a sequence?

Mei's role

• Project lead: owned study design, analysis, and presentation across all 3 playtests

• Supported by a 5-person team: 1 assistant researcher (lab moderation and note-taking), 2 operations lab assistants (check-in and logistics), and 1 team lead (department sign-off)

• Directly responsible for research design, stakeholder alignment, analysis, and final reporting

• Final reporting was presented directly to the lead designer and executive producers after each lab

The challenges

• A first-time stakeholder relationship that started badly: This was my first lead-researcher role, on one of League's biggest activations of the year, working with a new lead designer who had little experience with the research and feedback process. My first readout went poorly enough that he became visibly emotionally distraught, and I had to rebuild trust before my findings could make an impact on the character’s final design.

• Research goals shifted between playtests: The team's understanding of what Ambessa needed to be evolved playtest over playtest, from general combat-loop questions to an explicit character identity framing by Playtest 3. I collaborated with the developer team after each playtest to refine the research framework so it tracked their evolving goals instead of measuring against a moving target with old questions.

• A structural problem hidden in execution-level feedback: Early player feedback read as "the auto-dash is too hard," but the real issue was that Ambessa's gameplay mechanics didn't match the assassin identity players expected from her. I had to help the team see past the surface level difficulty complaint to the deeper identity mismatch driving it.

Goal of this research

Validate gameplay sentiment and provide actionable solutions to achieve developer goals.

Playtesting

Live Observation & Mixed-Method Surveys (total n = 86): League players piloted Ambessa live at 3 separate milestones across a 7 month development cycle on in-development builds. Sessions were measured through direct observation (onsite and via video review) and a post play mixed-method survey covering metrics such as gameplay understanding, enjoyment, ease of use, learnability, and visual clarity. Each playtest ran as a single 6 hour day across 4 games (30 - 45 minutes each). 

Reasoning behind the method

• Pros: Combining quant KPI trend-tracking across 3 playtests with qualitative thematic coding let the team see that design shifts had objective backing, not just a researcher's subjective read. Direct observation also caught real usability breakdowns that self-report alone would have missed. Because these were live playtests on in-development builds rather than post-launch feedback, findings could shape her mechanics while changes were still cheap to make.

• Cons: Research questions evolved between playtests, so results aren't perfectly comparable playtest over playtest. The subset of players actually piloting Ambessa each playtest was small (n = 7 - 9), so scores carry limited statistical power on their own. These issues are mitigated by triangulating with the larger "fair to play against" sample (n = 20 - 22) and with qualitative data across playtests.

Studies I used to triangulate results

• Playtest 1 (n = 30): recruited lower level players and focused on measuring general gameplay feedback (understanding of playstyle, accessibility). 8 participants were chosen to play Ambessa, whereas other players were either on her team or fought against her.

• Playtest 2 (n = 27): recruited high level players with specialties in fighters, measured gameplay feedback more in depth (engagement methods, ability combos, auto-dash usage). 7 participants were chosen to play Ambessa, whereas other players were either on her team or fought against her.

• Playtest 3 (n = 29): recruited high level players with specialties in assassin characters and measured design concepts (visuals, narrative, alignment with character fantasy) as well as specific gameplay feedback (types of item builds, visual clarity of attacks). 9 participants were chosen to play Ambessa, whereas other players were either on her team or fought against her.

↪ Recommendation to team

Continue refining remaining rough edges (dash readability, ability distinction) while preserving the assassin-identity direction that resonated with players.

What was assumed

Ambessa's complex, multi-ability mechanics would read as rewarding depth for advanced players without further validation of whether it matched her intended assassin identity.

What we found

Early playtests showed players liked and understood her mechanics far less than they should have, largely because her character identity didn't match her gameplay style. After the team realigned her mechanics and goals around that identity, Playtest 3 showed a significant, measurable improvement across headline metrics.

What was decided

The developer team pivoted Ambessa's gameplay identity to lean into an assassin playstyle, landing her in a state the executive producer was comfortable releasing alongside her reveal in Arcane.

Downstream wins

• Ambessa's in-game release was tied to her reveal on League's Netflix show, supporting engagement goals for both launches

• Significant improvements made to player-facing visual clarity and combat usability across her mechanics

• Reinforced a stakeholder-alignment approach that shaped how I approach research with less experienced stakeholders going forward.