Grammarly achieves 87% deflection and 4.2 CSAT with Forethought
Grammarly's previous chatbot could not understand conversational context, broke when users rephrased or asked follow-up questions, and required the support team to manually author every answer inside a decision tree that was hard to manage and update.
Grammarly's prior chatbot relied on menu-based interactions and canned responses, lacked multi-turn context awareness, and forced the team into constant manual maintenance of a sprawling decision tree.
After deploying Forethought, Grammarly's CSAT tripled to 4.2 out of 5 and deflection climbed from around 60% to a sustained 87%, never dropping below 80%.
API integrations drove an additional 5–10% gain.
Show all 7 reported metrics
Frequently asked questions
What did this team achieve with this AI workflow?
After deploying Forethought, Grammarly's CSAT tripled to 4.2 out of 5 and deflection climbed from around 60% to a sustained 87%, never dropping below 80%.
What tools did this team use?
Forethought, Solve.
What results were reported?
CSAT score: 4.2 out of 5; CSAT improvement multiplier: tripled; Deflection rate: 87%; Deflection rate starting point: around 60% (source-reported, not independently verified).
What failed first in this deployment?
Grammarly's prior chatbot relied on menu-based interactions and canned responses, lacked multi-turn context awareness, and forced the team into constant manual maintenance of a sprawling decision tree.
How is this customer support AI workflow structured?
Customer contacts via email or chat → Agentic AI understands customer intent → Knowledge base and system integration → Ticket deflected or resolved → QA accuracy confirmation → Unresolved query review loop.