The spreadsheet cell nobody can explain

• How to systematically read an AI system’s unresolved queries as market research instead of a model-training backlog, using the categorization approach Ingka’s data pointed to

• The specific organizational mechanism that converted a customer-service cost center into an accountable, targeted revenue channel

• Why reskilling programs succeed or fail based on whether they target an evidenced capability gap, rather than a generic “AI upskilling” mandate

• How to report automation savings and new-revenue metrics separately, so a bigger opportunity doesn’t get buried inside a cost-savings press release

• A specific 7-day pilot you can run using your own support system’s escalation logs, with no new AI tooling required

• What this implies about where competitive advantage is shifting in an AI-saturated market, and why “best model” is the wrong thing to optimize for

1. Executive Summary

2. Case Study in One Sentence

3. The Organization

4. The Challenge

5. The AI Strategy

6. The Implementation

7. Results

8. Why It Worked

9. Lessons for Readers

10. Replication Framework

11. Mistakes to Avoid

12. Steal This Idea

13. The Strategic Insight

14. Key Takeaways

15. Closing Thought

• Ingka Group, the largest IKEA franchisee, deployed an AI chatbot called Billie for customer service starting in 2021and from 2021 to 2023 it resolved approximately 47% of customer enquiries, translating to 3.2 million interactions and nearly EUR 13 million in savings.

• Rather than treating the unresolved majority of queries as a bot-training problem, Ingka analyzed them and found a concentrated, unmet demand for interior design consultation, a service its call center wasn’t built to sell.

Keep reading with a 7-day free trial

Subscribe to Business Analytics Review to keep reading this post and get 7 days of free access to the full post archives.