CHALLENGE

Replacing rigid navigation with a unique natural language interface to deliver intuitive, AI-driven discovery for HSBC customers

OUTCOME

Journey Optimization

Achieved a massive 64% reduction in journey steps and a 61% reduction in journey branches, drastically simplifying the user path to resolution.

Containment & Resolution

Increased predicted chatbot containment to 70% (up from 23–35%) while driving the channel mix toward 70% digital adoption

Advanced Personalization

Integrated 11 new APIs across 26 journeys to enable "on-the-fly" personalization. Customers now receive greetings and solutions tailored to their specific time zone, name, and banking segment (e.g., UK Premier) without manual data entry.

Context-Aware UI

Replaced a concealed chat entry point with a prominent, context-aware interface across all mobile hubs. The new UI features predefined search prompts and "smart replies" that educate users on conversational capabilities based on their current location in the app.

IMPROVEMENT

New Natural Language Understanding models
are expected to increase understanding to 97%

Today’s chatbot

• Manual language identification

• UK: 20-30 of utterances per intent (+ 200 per domain)

• HK: 5-1000 utterances per intent

• 85% NLU understanding accuracy

• Single intent understanding

• Sometimes frustrating experience

• Universe of intents manually defined

North Star Chatbot

• Automated language identification

• 10s of utterances per intent

• BERT integrated – NLU to understand meaning

• 97% NLU understanding accuracy

• Multi intent understanding

• Improved experience

• Increased set up efficiency

LOCALISATION

Achieved a 21% reduction with consolidated
Global Standard journeys with localisation
where sensible to drive consistency and efficiency

Local market deviations

Markets have the flexibility to deviate from global journeys due to:

• Product differences

• Business process differences

• Regulatory differences

• Market context / culture