Customer support · Production

Freshly achieves 50% conversation containment and 48% cost reduction with Conversocial automation

The problem

A Covid-driven spike in customer messaging volume starting March 2020 threatened to overstretch Freshly's agents and risked losing valuable new customers, requiring a scalable solution for high-volume recurring inquiries.

Workflow diagram · grounded in source

Messaging volume spike triggers need

trigger

“the continuous messaging volume soared in 2020 through the success of social ad campaigns and the global pandemic”

Automated flow handles high-volume intents

ai_action

“worked with the Freshly's team to develop an automated flow to address their highest volume intents. The objective was to handle recurring questions, enabling customers to self-serve”

Complex requests routed to agents

routing

“enabling customers to self-serve and leverage agents for more complex requests”

In-channel resolution delivered

output

“50% of conversations resolved in-channel with no need for human intervention”

Reported outcome

Freshly achieved 50% conversation containment in-channel with no human intervention, cut average first response time by 80% from 30 minutes to just six minutes, and reduced cost-per-contact by 48% by shifting customers to Facebook Messenger.

Reported metrics

Conversation containment rate50%

First response time reduction80%

First response time (before automation)30 minutes

First response time (after automation)just six

Show all 8 reported metrics

conversation containment rate50%

first response time reduction80%

first response time (before automation)30 minutes

first response time (after automation)just six

messaging agent conversations per hour20

phone agent conversations per hour9

Facebook Messenger cost-per-contact$0.84

live-chat cost-per-contact$1.24

Reported stack

ConversocialNLUZendesk

Source

https://www.verint.com/case-studies/freshly/

Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Freshly achieved 50% conversation containment in-channel with no human intervention, cut average first response time by 80% from 30 minutes to just six minutes, and reduced cost-per-contact by 48% by shifting customer…

What tools did this team use?

Conversocial, NLU, Zendesk.

What results were reported?

Conversation containment rate: 50%; First response time reduction: 80%; First response time (before automation): 30 minutes; First response time (after automation): just six (source-reported, not independently verified).

How is this customer support AI workflow structured?

Messaging volume spike triggers need → Automated flow handles high-volume intents → Complex requests routed to agents → In-channel resolution delivered.