Customer support · Production

ActiveCampaign deflects more than 60% of chat tickets and launches a full AI program with Forethought

The problem

ActiveCampaign's legacy chatbot relied on rigid decision trees requiring manual updates, making it clunky and hard to scale as ticket volume grew. Most AI vendors evaluated were building AI as an add-on with unproven features, leaving the team without a production-ready solution.

First attempt

The existing chatbot was built on rigid decision trees that required manual updates for every change, making it impossible to keep up with growing and varied customer needs.

Workflow diagram · grounded in source
1
Customer submits help center chat
trigger
“The AI-powered chat solution lives in the help center”
2
Solve AI understands and responds
ai_action
“Solve uses generative AI to understand what customers are asking, respond in natural language, and improve over time”
3
Routine tickets deflected or routed
routing
“gives agents more time to focus on the 30% of tickets that actually need a human”
4
Autoflows takes end-to-end action
ai_action
“Autoflows is Forethought's agentic AI reasoning engine. It takes the next step after a question is answered. Instead of routing to an agent or requiring a customer to take the next step, Autoflows can take action directly inside tools li…”
5
Assist supports agents in Zendesk
ai_action
“Assist, a browser extension that lives inside Zendesk. It provides agents with fast access to ticket history, summaries, and suggested actions”
6
Solve in Slack serves internal teams
integration
“The ActiveCampaign team set up Solve for Slack, because Slack is where most of the team already spends their day. That decision mattered. It meant people didn't have to go looking for help.”
Reported outcome

ActiveCampaign achieved over 60% sustained chat deflection that has not dropped since deployment, reduced agent-handled tickets by 40% from baseline, and saves the equivalent of more than five full workdays per week through internal Solve usage.

Reported metrics
Chat deflection rate+60%
Zendesk chats reaching agents+50% fewer
Tickets reaching agents vs baselinedown 40%
internal Solve chats per week1,300
Show all 9 reported metrics
chat deflection rate+60%
Zendesk chats reaching agents+50% fewer
tickets reaching agents vs baselinedown 40%
internal Solve chats per week1,300
Solve chats in Slack per weekroughly 1,000
full workdays saved weeklyover five full workdays every week
Solve in Slack vs internal widget volume ratiofour times the volume
internal chat resolution timeunder two minutes
routine customer question rateAbout 70%
Reported stack
ForethoughtForethought SolveForethought AssistAutoflowsAction BuilderZendeskSlack
Source
https://forethought.ai/case-studies/activecampaign-deflects-more-than-60-of-chat-tickets-and-launches-a-full-ai-program-with-forethought
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

ActiveCampaign achieved over 60% sustained chat deflection that has not dropped since deployment, reduced agent-handled tickets by 40% from baseline, and saves the equivalent of more than five full workdays per week t…

What tools did this team use?

Forethought, Forethought Solve, Forethought Assist, Autoflows, Action Builder, Zendesk, Slack.

What results were reported?

Chat deflection rate: +60%; Zendesk chats reaching agents: +50% fewer; Tickets reaching agents vs baseline: down 40%; internal Solve chats per week: 1,300 (source-reported, not independently verified).

What failed first in this deployment?

The existing chatbot was built on rigid decision trees that required manual updates for every change, making it impossible to keep up with growing and varied customer needs.

How is this customer support AI workflow structured?

Customer submits help center chat → Solve AI understands and responds → Routine tickets deflected or routed → Autoflows takes end-to-end action → Assist supports agents in Zendesk → Solve in Slack serves internal teams.

Related customer support cases