Customer support · Production

tado° uses Fin AI Agent to maintain CSAT stability through 400% peak-season volume spikes

The problem

tado° faced seasonal support volume spikes of up to 400% and a growing, multilingual customer base across six languages. Historically, CSAT scores dropped during the busy winter heating season due to longer wait times, while the team needed to scale without driving up overhead costs.

First attempt

For years tado° consistently experienced CSAT drops during its busy winter heating season, with longer wait times frustrating customers dealing with heating problems in cold weather.

Workflow diagram · grounded in source
1
Proactive at-risk customer guidance
ai_action
“identify areas in its platform where customers are at risk of running into difficulty and guide them through the process of getting set up using its products to ensure a smooth experience”
2
Customer contacts support
trigger
“whenever someone gets in touch, Fin is able to instantly identify”
3
Fin identifies customer segment
ai_action
“Fin is able to instantly identify: Which customer group the individual belongs to. Whether they're interested in automating their heat pump to use only the cheapest and greenest energy. If they have questions about the new tado° EV Smart…”
4
Route and resolve with help content
routing
“Fin can quickly guide customers down the right path and pull in relevant information from tado°'s help centers to provide them with the information they need and resolve their query”
5
Collect article feedback
feedback_loop
“workflow to request feedback from a user who reacts negatively to an article. According to the team, this is an invaluable channel that they can use to drive direct and continuous improvement of their help center content”
6
Tag and extract conversation topics
output
“the Support team uses tags to classify conversation topics and automatically extracts these to analyze the types of issues their customers are experiencing”
7
Share insights with product teams
feedback_loop
“Armed with these insights and the feedback that emerges from support conversations, they can work more closely with cross-functional development teams within tado° to help improve the product”
Reported outcome

With Fin AI Agent, tado°'s CSAT scores remained stable and even improved year-over-year during peak season, despite conversation volumes spiking up to 400% and reaching an average of 11,000 customers per week.

Reported metrics
Peak-season conversation volume increaseup to 400%
Weekly conversations during peak season11,000 per week
Monthly conversations outside peak season10,000 per month
CSAT during peak seasonremained really stable and improved compared with the same period last year
Reported stack
IntercomFin AI AgentWorkflows
Source
https://www.intercom.com/customers/tado-fin
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

With Fin AI Agent, tado°'s CSAT scores remained stable and even improved year-over-year during peak season, despite conversation volumes spiking up to 400% and reaching an average of 11,000 customers per week.

What tools did this team use?

Intercom, Fin AI Agent, Workflows.

What results were reported?

Peak-season conversation volume increase: up to 400%; Weekly conversations during peak season: 11,000 per week; Monthly conversations outside peak season: 10,000 per month; CSAT during peak season: remained really stable and improved compared with the same period last year (source-reported, not independently verified).

What failed first in this deployment?

For years tado° consistently experienced CSAT drops during its busy winter heating season, with longer wait times frustrating customers dealing with heating problems in cold weather.

How is this customer support AI workflow structured?

Proactive at-risk customer guidance → Customer contacts support → Fin identifies customer segment → Route and resolve with help content → Collect article feedback → Tag and extract conversation topics → Share insights with product teams.

Related customer support cases