Kogan.com
Salesforce Agentforce customer service agent
Kogan.com's Agentforce agent took about half of customer inquiries and raised true resolution from 6.2% to about 20%
Problem
Kogan's earlier rule-based bot resolved only about 6% of inbound inquiries. Where-is-my-order questions made up roughly 60% of requests.
Approach
Kogan rebuilt the bot as a Salesforce Agentforce agent, prototyped in an internal hack day and live in production about six weeks later, launched in March 2026. It reads order status from SAP, handles cancellations and address changes, recommends products, troubleshoots TVs and runs a seven-step TV return flow with photo upload and duplicate-case checks. Unresolved inquiries go to a live agent with a summary, or the customer is offered a callback after a 15-minute wait.
Results
Takeaway
Kogan began with its most common question, order status, and added returns and troubleshooting after. Salesforce reported a 20% sales volume increase in the same period and did not attribute it to the agent.
Categories
- Industry
- Retail and e-commerce
- Business function
- Customer service
- Innovation type
- Process innovation
- AI technology
- AI agentsConversational and voice AI
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Salesforce's customer story covers the Agentforce agent Kogan launched in March 2026. It reports about 50% of inquiry volume handled, true resolution up from 6.2% to about 20%, and about 5,000 cases a month. The headline's 67% does not match the article body, so it is not used here.
iTnews reports that Kogan's first bot resolved about 6% of inquiries and the rebuilt agent covers 10 use cases. It describes the SAP order lookups, recommendations across 16 million SKUs and image checks on return photos. It gives no resolution results for the new agent.
Full details are in the original sources.