Interaction Analytics converts unprocessed conversational data into a structured format where it can be analysed by an AI language model to extract important intents, behaviours and sentiments which are then used to improve customer experience (CX) and staff productivity. Interaction Analytics has become significantly more effective due to the recent improvements to language models such as GPT4, Anthropic, Google Bard and the Alpaca variants. This because it can now comprehensively understand everything contained within a phone call, email, SMS message, social media post or rating site feedback comment and classify in order of importance.
Example AnalysisInteraction Analytics is installed on AWS servers which connect to transcription, machine learning, message routing servers directly via their high speed internal network. Media files are typically stored in S3 buckets or Azure Blob storage where they are encrypted at rest. During the transcription process personal data can be redacted if it is not needed this means that it will not be processed by the LLM systems. AWS is accredited to ISO27001, ISO27017 & PCI DSS which are three of many accreditations that are most relevant to Interaction Analytics. Hostcomm is PCI DSS Level 1 compliant and its technical staff treat personal data in the same way as card data.
Hostcomm Interaction Analytics service uses Amazon AWS for message routing, voice transcription and machine learning. GPT3/4 and Anthropic are used for LLM processing as well as a few open source language models when fine tuning and customisation is required.
Process Step |
Description |
Advantage |
|---|---|---|
Data Collection |
Calls are recorded and transcribed, capturing agent-customer conversations, usual with speaker identification | Text is now formatted can now be used by AI & language models. |
Pre-processing |
Transcribed text is cleaned, removing information like personal data and background noise, focusing on the interaction. | Processing is more efficient and free from personal data (personal data can be kept when using a custom, secure LLM). |
Text Analysis |
Advanced AI techniques analyse the text, including tokenisation, part-of-speech tagging, sentiment analysis, and more. | Preparation for next step. |
Behaviour analysis |
AI extracts insights on behaviour, detecting patterns like tone, intent, active listening, upselling attempts, empathy, responsiveness, adherence to compliance, and customer satisfaction. There is no limit to what can be detected. | Performance & CX measurement improvement. |
Metrics and Scoring |
AI generates metrics to measure performance, including customer satisfaction ratings, handling time, resolution rates, and compliance adherence. | More effective management of service agents relating to performance. |
Feedback and Scoring |
Metrics provide feedback to agents and supervisors, highlighting improvement areas and strengths, supporting training and coaching efforts. | Faster improvement in competence of customer service agents and hence their ability to deliver a better experience for customers. |
Automated Improvement |
The AI model learns from feedback, enhancing its ability to measure behaviour, adapting to company needs, and keeping up with evolving customer service trends. | Ensures that customer service is continually monitored across all touch-points with minimal administrational input. |
If you would like to get more information, or request a demo, please leave some contact details opposite and we will get right back to you.
Alternatively, give us a call: 0808 168 4400
