Customer support · Production

Fetch Achieves 26% More Customer Support with Same Workforce and 3.9x ROI Using Forethought

The problem

Fetch's rapidly expanding user base drove recurring ticket surges whenever new app features launched. About 30% of tickets were easily answered FAQ-type questions and another 20% required agents to gather additional information from users, creating a high reopen rate. The team was deeply skeptical of chatbots after poor experiences with clunky keyword-based tools.

First attempt

Fetch initially deployed Forethought Triage to predict ticket content and send automated email responses, but this approach only addressed less than 1/3 of the basic ticket types it was meant to deflect.

Workflow diagram · grounded in source
1
Customer submits ticket via email
trigger
“They mainly communicate over email, interacting with thousands of users a day.”
2
Triage predicts ticket content
ai_action
“Fetch initially used Forethought Triage to predict the content of a ticket.”
3
Automated email deflects simple tickets
output
“Fetch would send automated email responses based on that prediction, effectively deflecting many simple tickets.”
4
Scout answers wide range of questions
ai_action
“Since deploying our Forethought AI agent, Scout, our users have been able to get a wide range of questions answered.”
5
Human assistance for complex cases
human_review
“massively expand the scope of issues that Scout can handle with and without human assistance”
Reported outcome

After deploying Forethought Solve as their AI agent Scout, Fetch achieved 26% more customer support capacity with the same workforce and a 3.9x ROI, automating 316,000 ticket actions for $90,000 in 11 months, with CSAT scores for fully automated chats as good or better than those of human agents.

Reported metrics
Customer support capacity with same workforce26%
ROI3.9x
Ticket actions automated316,000
Cost of automation over measurement period$90,000
Show all 12 reported metrics
customer support capacity with same workforce26%
ROI3.9x
ticket actions automated316,000
cost of automation over measurement period$90,000
automation deployment duration for ROI measurement11 months
accuracy in head-to-head testing93%
accuracy advantage over tested competitor28%
CSAT for fully automated chats vs. human agentsas good or better than those with human agents
share of tickets that are basic FAQ/navigation questions30%
share of tickets requiring additional info gathering from user20%
share of basic tickets deflected by initial Triage-only approachless than 1/3
customer support team sizemore than 100
Reported stack
ForethoughtForethought SolveScoutZendesk
Source
https://forethought.ai/case-studies/fetch-achieves-triple-roi-forethought
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

After deploying Forethought Solve as their AI agent Scout, Fetch achieved 26% more customer support capacity with the same workforce and a 3.9x ROI, automating 316,000 ticket actions for $90,000 in 11 months, with CSA…

What tools did this team use?

Forethought, Forethought Solve, Scout, Zendesk.

What results were reported?

Customer support capacity with same workforce: 26%; ROI: 3.9x; Ticket actions automated: 316,000; Cost of automation over measurement period: $90,000 (source-reported, not independently verified).

What failed first in this deployment?

Fetch initially deployed Forethought Triage to predict ticket content and send automated email responses, but this approach only addressed less than 1/3 of the basic ticket types it was meant to deflect.

How is this customer support AI workflow structured?

Customer submits ticket via email → Triage predicts ticket content → Automated email deflects simple tickets → Scout answers wide range of questions → Human assistance for complex cases.