When a player churns, ask them.
Invite players to an AI interview when they stop logging in or skip a purchase. Findings from the interviews are saved as user tags in Agentic Engine.
- Failed level 3 ×3
- Closed it without buying
“I couldn't tell what the stuff in the bundle actually does, and I didn't want to waste my money.”
Written back from the interview
Start with the drop-off.
Select players who left at the same point, such as after level 3, and interview them.
Behavior data alone
Level 3 loses 38% of players and few buy from the bundle page. Nothing here says why.
Traditional interviews alone
You get players' own accounts, but an interview study can take weeks, and the answers stay separate from behavior data.
Interview study · weeks
Behavior data and interviews together
- Player 88213Closed the bundle without buying
- Player 10422Finished level 5
- Player 77105Closed the bundle without buying
- Player 30981Bought the starter pack
- Player 56470Closed the bundle without buying
Turn churn, refunds and skipped bundles into interview invites.
Keep a trigger running to add new interviews to the same study. Set quotas by player segment and cap how often each player is invited.
- RunningNo login for 7 daysChurned players
- RunningRefund requestPaying players
- RunningViewed the first-purchase bundle without buyingNew players, first 7 days
See which findings held up.
After a change ships, your team checks each finding against Agentic Engine behavior data and marks it confirmed, unproven or overturned.
“I couldn't tell what the stuff in the bundle actually does, and I didn't want to waste my money.”
Opened the first-purchase bundle page 3 times without paying.
Players can't tell what the bundle items do
The bundle price is too high for new players
New players skip the bundle because they never visit the shop
- Answers next to behavior
- Checked after launch
- A status on every finding
- Into the knowledge base
Five steps take a churn signal to a tagged finding.
The agent sends invitations and runs interviews. Findings become user tags and go into a knowledge base of past findings. Your team decides what to change.
Behavior signal
Churn, a refund or a skipped purchase triggers the invitation
AI interview
The agent works through the guide and asks for details
Reasons and quotes
A researcher reviews each finding
Findings saved
As user tags and in the knowledge base
Your team decides
Review the evidence, choose the next move
Live operations teams
Group players by the reasons they gave, such as players who skipped the bundle because its items were unclear.
Product teams
Look up past findings in the knowledge base, each with the player quotes behind it.
Your own agents
Your own AI agent can start a study and fetch findings through MCP, an open standard for connecting agents to tools.
A researcher reviews every finding
The agent runs the interviews and pulls out reasons with quotes. A researcher reviews each finding, and your team decides what to change in the game.
The interviews run on heari.ai
Connect heari.ai to Agentic Engine to pick participants by behavior and review findings next to your play data.
- Participant quotes support every finding
- Interviews in each player's own language
- Works on its own or connected to Agentic Engine
Keep exploring
User Interview Agent
It holds one-on-one interviews with many players at once and asks follow-up questions as each conversation unfolds.
Meet the agent
Total Context
Agents read your player data, community posts and your team's own documents together.
See Total Context
Engagement Agent
It drafts player campaigns from your data. Your team approves each one before players see it.
See the Engagement Agent
Ask players before you change the game.
Connect AI user interviews to your behavior data in Agentic Engine.
