CX AI Features

// // overview modern consumers expect seamless and efficient digital experiences each interaction contributes to the overall customer experience (cx), and organizations need tools that provide deep, actionable insight into those interactions elevateai, powered by nice's cx ai models and backed by decades of contact center research and billions of customer experience interactions , delivers immediate analysis of voice and chat interactions at scale using pre built behavioral models what's new with elevateai's v1 11 release , all declared interactions – including audio, transcripts, and chat – are now automatically processed using the full range of available cx ai services this update expands coverage across all communication types and ensures consistency in customer experience analysis key capabilities elevateai's pre built cx ai models enable smarter application integration and process automation automated analysis of customer interactions at scale predictive and prescriptive insights, based on behavioral models purpose built models for immediate deployment and value // // voice activity for any audio interaction, voiceactivitysegments returned with cx ai results include the start and stop times ( in milliseconds ) for all voice and silence segments this enables immediate insights into total non interaction time – like periods of silence or hold time – when participants are not actively engaging with one another for audio interactions, elevateai provides speaker diarization for cx ai using both cx and echo transcription models – automatically identifying each speaker, or voice segment, as associated with participantone or participanttwo , empowering you with information related to the total talk time associated with each speaker, as well as any cross talk that may have occured and who initiated note the cx model supports diarization in mono and stereo, while the echo model now supports diarization in stereo only, at this time sample cx ai response { "allparticipants " { "voiceactivitysegments" \[ { "starttimeoffset" 0, "endtimeoffset" 4920, "result" "voice" }, { "starttimeoffset" 4920, "endtimeoffset" 7560, "result" "silence" } ] } } // // enlighten ai bundles nice enlighten scores are generated for each behavioral model and are provided as an overall score, as well as segmented by interaction quartile this allows users to identify behavioral shifts over the course of the interaction behaviors for customer satisfaction // // acknowledge loyalty the acknowledge loyalty score indicates how effectively an agent acknowledges a caller's long term relationship with the organization and expresses appreciation for their loyalty a higher score suggests the agent is proficient at recognizing customer tenure in a natural, conversational way the model was trained to evaluate the entire conversation – considering language from both the agent and the caller – to assess agent performance and determine this score for the most accurate assessment, we recommend using the allparticipants score, which reflects the interaction as a whole // // active listening the active listening score reflects how well an agent listens and responds during a conversation a higher score reflects higher predicted agent proficiency , indicating that the agent is engaged, responsive, and avoids asking the caller to repeat information unnecessarily the model was trained to evaluate language from both the agent and the caller to assess listening behavior because of this, we recommend using the allparticipants score to get the most accurate view of agent performance // // be empathetic the be empathetic score reflects how well an agent acknowledges the caller's issues and shows understanding of their impact or hardship a higher score suggests higher predicted agent proficiency in demonstrating strong empathy during the conversation the model analyzes both sides of the conversation – agent and caller – to assess empathetic behavior to get the most accurate insight, we recommend using the allparticipants score when evaluating this agent behavior metric // // build rapport the build rapport score measures how effectively an agent creates a personal connection with the caller by engaging in friendly, general conversation not directly related to the reason for the call a higher score indicates strong agent proficiency in building trust and comfort through small talk or positive interaction the model analyzes language from both the agent and the caller to assess this behavior for the most accurate results, we recommend using the allparticipants score when evaluating agent performance // // demonstrate ownership the demonstrate ownership score measures how effectively an agent shows they understand the caller's issue and are committed to helping resolve it a higher score indicates strong agent proficiency in providing reassurance and taking responsbility during the interaction the model evaluates language from both the agent and the caller to assess this behavior for the most accurate results, we recommend using the allparticipants score when reviewing this agent metric // // effective questioning the effective questioning score reflects how well an agent asks thoughtful, purposeful questions to better understand the caller's experience, concerns, or needs a higher score indicates strong agent proficiency in using questions to explore issues and uncover opportunities during the conversation the model analyzes both sides of the conversation – agent and caller – to assess the agent's ability to ask meaningful, insightful questions throughout an interaction for the most accurate insights, we recommend using the allparticipants score when evaluating agent performance // // inappropriate action the inappropriate action score reflects how well an agent avoids behaviors that are considered unprofessional or offensive – such as refusing a reasonable transfer request, using inappropriate language, or acting disrespectfully a higher score indicates strong agent proficiency in consistenctly maintaining appropriate conduct throughout the conversation the model analyzes language from both sides of the conversation – agent and caller – to evaluate this behavior to ensure the most accurate results, we recommend using the allparticipants score when assessing this agent behavior // // promote self service the promote self service score measures how effectively an agent encourages callers to use available self service options – such as an ivr system, website, or mobile app – when appropriate a higher score indicates strong agent proficiency in guiding customers to helpful self service tools that can streamline future interactions the model evaluates language from both sides of the conversation – agent and caller – to evaluate this behavior to get the most accurate results, we recommend using the allparticipants score when assessing this agent behavior // // sentiment the sentiment score indicates the predicted likelihood that a customer would give a positive rating in a post contact survey, post interaction the score is based on both spoken language ( what was said) and acoustic cues ( how it was said), including pitch and tone, laughter detection, and cross talk the model was trained to analyze the full conversation – across both the agent and the caller – to determine customer sentiment for the most accurate results, we recommend using the allparticipants score when assessing this customer behavior // // set expectations the set expectations score measures how effectively an agent communicates next steps, outlines required actions, and helps the caller understand what to expect after the interaction a higher score indicates strong agent proficiency in providing clear, proactive guidance that builds confidence and reduces uncertainty the model evaluates language from both the agent and the caller to assess this behavior for the most accurate results, we recommend using the allparticipants score when assessing this agent behavior vulnerable customer behaviors a vulnerable customer is defined by the uk financial conduct authority as "someone who, due to their personal circumstances, is especially susceptible to harm, particularly when a firm is not acting with appropriate levels of care " these customers may require additional actions on the part of financial firms to ensure they receive fair and equal service // // vulnerable customer the vulnerable customer score indicates the predicted likelihood that a caller may fall into one or more of the five defined vulnerable customer subcategories referenced in contact centers, which includes health , life events , resilience , capability , and age related factors a higher score suggests the presence of customer language or behavior that signals potential vulnerability during the conversation the vulnerable customer model analyzes both sides of the conversation – agent and caller – to evaluate potential indicators across an interaction for the most accurate insight, we recommend using the allparticipants score when assessing this customer behavior // // vulnerable capability the vulnerable capability score indicates the likelihood that a customer has limited knowledge or confidence in managing financial matters it also reflects challenges in other areas like financial literacy or digital skils a higher score means the customer may need extra support due to these vulnerabilities the model analyzes both sides of the conversation – agent and caller – to identify this behavior for the most accurate assessment, we recommend using the allparticipants score when assessing this customer behavior // // vulnerable health the vulnerable health score indicates the likelihood that a customer is experiencing physcial health conditions or illnesses that impact their ability to manage daily tasks a higher score suggests the customer may need additional support due to these health related challenges the model analyzes both sides of the conversation – agent and caller – to identify this vulnerability for the most accurate results, we recommend using the allparticipants score to assess this customer behavior // // vulnerable life event the vulnerable life event score indicates the likelihood that a customer is going through difficult situations such as bereavement, job loss, or relationship breakdown a higher score suggests the customer may need extra care and support due to these challenges the model analyzes language from both sides of the conversation – agent and caller – to identify this vulnerability for the most accurate results, we recommend using the allparticipants score when assessing this customer behavior // // vulnerable mental health the vulnerable mental health score indicates the likelihood that a customer is experiencing mental health conditions that impact their ability to manage daily activities a higher score suggests the customer may need additional support or care due to these challenges the model evaluates language from both sides of the conversation – agent and caller – to identify this vulnerability for the most accurate results, we recommend using the allparticipants score when assessing this customer behavior // // vulnerable resilience the vulnerable resilience score indicates the likelihood that a customer has a reduced ability to cope with financial and/or emotional stress a higher score suggests the customer may need extra support due to these challenges the model evaluates language from both sides of the conversation – agent and caller – to identify this vulnerability for the most accurate assessment, we recommend using the allparticipants score when assessing this customer behavior additional models in currently in beta include financial distress high risk transactions