The role of external context to understand the customer mindset & personalize digital channels & chatbots.
The core feature of Anbotux is the usage of external contexts and compare the customer-care legacy channels & chatbot usages and customer segmentation ot understand the drivers of each individual customer or customer-segment when contacting their service provider (bank, insurance company, frequent airline, etc).
Anbotux has created a way to normalize the external contexts in order to process ML algorithms formally in an equivalent way even for very different external contexts. It is the way that we pre-structure different information that can be analyzed globally.
External contexts are whatever phenomena that a company could verify the impact on individual users/customers or in a particular segment. For example for an insurance company it could be needed to test severe weather influence in customer service usage, a telco could need to try to compare marketing campaigned perceived pressure with the use of customer care of their customers asking for better prices and also detect with are the groups or segments more like to do it. A bank could want to measure the real impact of a stock exchange shock or an airline could try to measure how news about ATC or pilots possible strikes impact in customer service needs. Detect reactivities of the customers to those external contexts and more to be ideated, loaded with daily values will help to create special offerings for most reactive groups and also personalized the conversations adding empathy and making the customer feel understood, not only the language but the mindset, key to reduce churn:
The external contexts:
- The external context could be totally objective values (ranges of an stock value change) or totally subjective (economic optimism).
- The external context can be geographically applied. The stock exchange is a national context but the presence or heavy rain is something local.
- The way to normalize the information forces to create a scale between minimum and maximum values, and the number of intermediate steps will determine the granularity and determination power of the context provided.
See a couple of examples:
After proccessing the external context values benchmarked with desviation for average customer care and chabot usage it is possible to create reactivites (total, and channel by channel) for individual customers:
and customer-segments:
Getting the best predictors we could personalize the customer service conversation, the way to provide real time the personalization info is using an API to allow chatbot (and other customer service channels) developers to obtain the information and ellaborate personalization estrategies. It is key to provide a friction-less integration with the technologies in charge of facing the customer requirements :-)
Have a look to the videodemo covering the topic:
Whatever comment, doubt, idea is more than welcomed !
the Anbotux team :-)
