Telenor's AI virtual agent Telmi exceeds ROI goals in under 12 months, driving customer satisfaction and revenue gains

Background

The American Customer Satisfaction Index found that telecom companies ranked last in customer satisfaction among private-sector industries as of 2019. Telenor faced declining demand for phone-based service and rising customer expectation for instant digital responses. The company needed a solution that could handle complex transactional requests — not just static FAQ responses — through an automated channel.

What Was Implemented

• Engaged Boost.ai in late 2018; deployed Telmi virtual agent in January 2019
• Deep learning and natural language processing enabling advanced conversational understanding
• 20+ unique integrations enabling transactional actions within customer accounts (PUK codes, data plan upgrades, invoice viewing)
• Dedicated AI Trainer team at Telenor to monitor interaction quality and improve the model continuously
• Positioned as an additional sales channel, not only a cost-reduction tool

Results

Telmi exceeded Telenor's defined business and ROI goals within the first 12 months of deployment. It demonstrated value as an additional sales channel. The 20%/15% customer satisfaction and revenue figures frequently cited in connection with this case study originate from a general McKinsey industry study cited by Boost.ai — they are not confirmed as Telenor-specific measured outcomes in the primary source. The record uses only confirmed outcomes.

Lessons

• Virtual agents generate higher business value when built with deep transactional integrations (account access, order execution) rather than designed as answer-bots
• Setting explicit business and ROI goals before deployment creates accountability for commercial performance beyond cost containment
• Sustained improvement requires a dedicated internal team (Telenor's AI Trainers) to continuously monitor quality and refine the model