Closing 65% of payments inside chat with Agentforce

Technology & Services
5 Minutes

Everpeak built an Agentforce agent that takes the payment inside the chat window from start to finish in 30 seconds.

About the Client

A FinTech SaaS company that bills a large customer base on a recurring basis.

01

The Challenge

A customer would arrive in chat ready to pay, and the chat was unable to accept a payment. The chat tool was built to point them somewhere else: into a self-service portal, a knowledge base article, or to trigger an email. The payment often never survived the conversation it started in.

  • The chat could not take a single payment. Its job was to hand the customer off, to a portal or a help article, and hope they finished somewhere the business could not see.
  • Every payment cost an employee five to six minutes. A live agent had to take the details and key them into another system, against 100 to 200 payment conversations a day.
  • Nobody could confirm the money arrived. Even when a customer paid through a link, the agent had no way to know it cleared without going to another system to check.
The goal: let a customer pay inside the conversation they are already in, and let that conversation end with a confirmed payment.

02

The Solution

Everpeak built an Agentforce "Payment Specialist" agent, live on web chat, that keeps the payment in the thread the customer is already in.

  • The agent takes the payment request itself. It recognizes when a customer wants to pay and handles the conversation from the first message, escalating to a person when a request falls outside its scope. It also answers account questions, such as payment history, from the payment provider's records inside Salesforce.
  • The amount comes from the invoice record, not from the customer. The agent pulls what is owed, then takes the customer to a card they have already saved with the payment provider, or to a step where they choose one. Capture happens on the provider's hosted page, so no card number is ever requested or entered in Salesforce. The whole transaction runs in under 90 seconds, and faster still when the card is already saved.
  • The confirmation comes back into the same chat. Everpeak built a listener that waits for the payment provider to confirm the payment cleared, then posts that confirmation straight into the open conversation. Nobody has to go check a second system.
  • The paid invoice goes out on its own. On a successful payment the system sends the customer their paid invoice automatically, so the transaction finishes end to end without ever leaving the thread.
03

The Impact

Agentforce now takes 65% or more of payments itself, start to finish, in the chat the customer is already in. The rest still reach a person, which is what the agent escalates for.

  • 70% faster for a customer to pay. A payment that took an employee five to six minutes now completes in under 90 seconds, and on the payments the agent handles, no employee touches it at all.
  • Fewer payments stall. A customer who arrives ready to pay can actually finish paying, which is money that reaches the business in the conversation rather than in a follow-up cycle.
Technology Used
Salesforce AgentforceThe AI agent running the payment conversation
A
A payment gateway hosts card capture and confirms the payment cleared
Results at a Glance
65% of payments finish in the chat they start in
70%faster for a customer to complete a payment
30 sec full transaction, start to finish
100+payment conversations arriving in chat per day
Solution at a Glance
4 Steps
Ask (customer asks to pay in chat)
Amount (pulled from the invoice record)
Pay (card captured on the provider's hosted page)
Confirm (confirmation posts back in chat, paid invoice sent)
One conversation, four steps, no handoff.

About Everpeak

Founded in 2017, Everpeak helps organizations win more deals and operate smarter by leveraging their technology, data and people.

The Revenue Climber

Everpeak’s bi-weekly newsletter offers revenue leaders free How-To guides, software recommendations, and practical frameworks for scaling operations.