Most customer support teams are built to react. They reply to frustrated customers, escalate their issues, and hope to smooth things over. But the damage is often done by the time they’re looped in.
The problem isn’t their effort or skill—it’s their ability to proactively solve problems before the customer feels frustrated. Things like a feature that keeps confusing new users, a billing question that surfaces week after week, or frustrating phone trees aren’t just one-off issues. They’re signals, and the right AI can catch and solve them before they turn into churn.
Research proves bad customer service costs you customers
We surveyed customer service leaders and professionals and found that you spend five to 25 times more money acquiring new customers than keeping current ones, yet many slip away without a peep.
One reason is that 37% of customer interactions still end without a satisfying resolution. These unresolved conversations often lead to silent churn. People don’t complain; they just stop coming back. That alone is costly, but these moments also drive up support spend. Cost per resolution has climbed to an average of $15, up from last year, with little to show for it when customers leave anyway.
Plus, unsatisfied customers often tell 9 to 15 others about their experience. That word-of-mouth actively shapes how others see your brand. With 77% of consumers saying service quality drives brand perception, those stories can limit your future pipeline before you even see it.
When you can spot unresolved issues early and respond with consistency, fewer customers fall through the cracks. Companies that use AI effectively see 64% improvement in customer satisfaction, compared to just 49% for those without it. That means less churn and lower support costs.
This is why your customers are walking away
When support feels impersonal, people leave. They want fast, effective help without repeating themselves or jumping through hoops. Our benchmark report shows that 60% of people still turn to the phone when they want resolution because they’re looking for faster human touch. Yet, voice is where experience breaks down most, with long hold times and clunky menus.
They want answers fast and expect human agents to know who they are without needing to re-explain everything. In fact, 47% say that repeating themselves is one of their top frustrations.
Preferences vary by age. While everyone prefers the phone, younger generations feel most comfortable chatting and texting. But across all groups, what drives people away is having to start over when they switch channels. That inconsistency doesn’t just frustrate them and shapes how they see your brand.
And too often, a resolution never comes. Over a third of customer service interactions still end in dissatisfaction. People will try to help themselves first, but when that fails and support doesn’t follow through, they don’t complain. They just leave.
Smart companies are moving beyond basic automation
The first wave of support automation included decision-tree chatbots and IVR phone menus. These systems followed strict scripts, handled only simple issues, often frustrated customers, and burdened CX teams with more work than they were worth. According to our benchmark report, only 8% of companies rely solely on these tools, and that number is shrinking fast.
The next step was retrieval-augmented generation (RAG), especially when paired with generative AI. This brought more natural conversations and better access to information. But it still puts the burden on customers to act. These systems helped—55% of companies saw improved CSAT, with an average of 6.4 NPS and 37% reporting lower resolution costs—but they didn’t go far enough.
Agentic AI marks the next shift. It doesn’t just respond; it takes action.
These systems resolve issues on behalf of the customer by issuing refunds, checking order status, or updating account details automatically. This shift to resolution-first automation drives better outcomes: 64% see improved CSAT, NPS rises to 8.3, 63% report reduced costs, and 54% see better retention.
Our survey confirmed that customers rank “solve my issue” and “connect me to a human” as top priorities. Systems that only answer questions but don’t resolve issues are no longer enough. Customers expect outcomes, not explanations.
AI that takes action delivers dramatically better results
Most customer support interactions end quietly. But every now and then, a support experience is so smooth, fast, and helpful that people actually talk about it publicly.
That’s what happened when a Grammarly user was charged for an auto-renewed subscription. Instead of getting redirected or delayed, an AI agent refunded him on the spot over chat. No human agent intervened, no escalation, no back-and-forth. He got the money back in under two minutes and then posted about it on LinkedIn—not to vent, but to say the experience made him more likely to resubscribe.
Agentic AI understood what he asked, confirmed refund eligibility, and completed the transaction immediately. This kind of outcome isn’t just luck; it’s what Forethought was built to do.

Forethought is an AI platform for customer support teams. It helps companies resolve issues faster and more accurately across chat, email, and phone. Two of its core products—Solve and Voice—handle customer conversations on the front lines.
- Solve is Forethought’s AI chat experience. It doesn’t rely on scripts or rigid flows. Instead, it uses generative AI and retrieval-augmented generation (RAG) to understand customer questions and deliver clear, relevant answers across a wide range of topics.
- Voice brings those same capabilities to phone support. Instead of forcing customers through phone trees or long hold times, it uses natural language to understand what someone needs and respond directly without an agent on the line.
Both Solve and Voice can surface helpful information, answer common questions, and reduce the load on human agents. But the real shift happens when they’re powered by Autoflows.
Autoflows is Forethought’s agentic AI layer. It connects the conversation to action. When a customer asks for a refund, change, or update, Autoflows interprets the intent and runs the right workflow automatically, like in the Grammarly exchange. That AI agent didn’t just explain the policy; it processed the refund.
According to our benchmark report, companies using agentic AI report 64% improved CSAT, 63% lower resolution costs, and 54% stronger retention trends. When automation resolves the issue instead of passing it off, customers notice and stay loyal.
Here’s how to build support that keeps customers loyal
Support leaders talk about improving CX, but the teams that actually do it share a specific mindset: they treat every conversation like part of a system. They know loyalty depends on the structure behind the response—how requests get routed, how knowledge gets surfaced, and how actions get triggered.
This is the real work of operationalizing loyalty. It’s not about tone of voice or adding emojis to messages. It’s about finding where support breaks down, choosing tools that resolve problems instead of deflecting them to human agents, and implementing AI in a way that actually fits your workflows.
Finding the moments that drive customers away
Customers lose trust when they have to repeat themselves, start over after switching channels, or wait too long for help. According to our benchmark report:
- 63% of customers have had to restart a conversation because a chat ended or the context was lost
- 70% say support quality directly shapes how they feel about a brand.
These problems often happen quietly, without setting off alarms, but they add up. When support can’t solve the issue or connect the customer to someone who can, trust slips.
Selecting technology that matches your business needs
Findings from our benchmark report show that companies using dedicatedAI platforms see better results across customer satisfaction, resolution cost, and retention than those relying on helpdesk add-ons or fragmented solutions.
Forethought is built to be a complete system for modern support teams. Solve and Voice handle conversations across chat and phone using generative and agentic AI, but Forethought goes beyond chat and voice.
- Discover helps teams find automation opportunities by analyzing ticket data
- Triage automatically tags and routes issues based on context
- Assist supports agents during live cases by surfacing relevant answers and knowledge
All of this runs on a shared AI foundation, which means everything works together. Forethought also connects to hundreds of external tools (think CRMs, billing systems, and order platforms) so your AI agent can understand what the customer needs and take action right away.
Your customers’ expectations are already changing
The most dangerous thing about poor support is what it silently costs you. Every unresolved issue represents a customer who walks away without saying a word. And every friction point creates an opportunity your competitors are eager to claim.
What’s surprising is that most companies already have the data they need to fix these problems. What they don’t have is a system that connects the dots across voice, chat, and email to see them in the first place. That’s where Forethought comes in.
See how our newest solution, Forethought Voice, transforms support calls from a liability into a differentiator. Schedule a demo to hear it for yourself.
Hashtags blocks for sticky navbar (visible only for admin)
{{resource-cta}}
{{resource-cta-horizontal}}
{{authors-one-in-row}}
{{authors-two-in-row}}
{{download-the-report}}
{{cs-card}}

