Get best of the week

Contact Center AI: third party in conversation is fine


Our customers and blog readers are aware of Dialogflow, a platform for creating smart bots. It is Dialogflow that underlies Contact Center AI (hereinafter CCAI) from the corporation of good. Schematically, the technology stack is equal to DialogFlow + WaveNet (text-to-speech) + voice recognition (speech-to-text), at the output it expands the capabilities of a conventional call center, namely it allows you to provide more personalized support services. How exactly? We talk under the cut and give examples of successful integration, welcome!

Concept


Virtual operator


Actually, the very coveted “automation of a normal person”, when 24/7 people respond not with a pre-recorded IVR, but with a voice assistant instead of a live operator - using slot-filling, the operator understands the problem / intention of the called customer, expressed in free or semi-free form. For the caller, this means that you don’t have to get through the gap in the recorded messages in order to get help faster as a result. However, we keep in mind that requests are very unusual, in this case, the conversation will still be transferred to the employee who will help ...

Live Operator Assistant


The bot will “listen” to the conversation of two people and help the operator, giving advice / links during the conversation. The operator can use these materials at his discretion: peek at the articles from the knowledge base himself or share quotes and / or links with the called customer, etc. It is planned that such information support will support the operator in any situation - it sounds sensible, because "one head is good, and two is even better."

The assistant has several modes of operation:
  • article_suggestion - the bot recommends links to articles;
  • smart_reply - several answers to one question; options are generated based on predictive analysis;
  • FAQ - the bot recommends answers to frequently asked questions;
  • dialogflow_assist - prompts in the spirit of "as if the bot answered itself";
  • by default - the bot gives the most relevant hints of all types.

An interesting point: while the bot conducts the conversation alone, it not only transcribes the conversation for its needs, but arranges it as a visual chat - this chat then sees a live operator who picks up the conversation after the bot. Thus, the employee immediately gets into the context of the conversation; how it looks can be seen here:

The important thing is that the employee can always either accept the proposed option, or reject it (and write his own) - so the assistant can retrain directly “in battle”.

Planned outcome


It is expected that CCAI will help to understand the emotional mood of users and the reasons for their calls in order to improve their service. As a result, companies using CCAI:
  • It will be easier to build trust with customers;
  • Increase user satisfaction
  • will be able to make each employee an expert;
  • will receive rapid introduction and entry into the market.


Who is already using


GoDaddy registrar, Hulu streaming service, easyJet low-cost airline, PolicyBazaar aggregator and Marks & Spencer retailer. By the way, about the last two - Antony Passemard from Google shared (and we translated) how the technology helped the call centers of these companies.

Policybazaar: conversational AI is profitable


A painful choice of the best life insurance, car, property, etc. can bother anyone, because there are a lot of offers, each of them has its own nuances and tons of descriptions. policy aggregator PolicyBazaar is trying to eradicate this problem through AI in its call centers. The aggregator was one of the first to use CCAI, because the company was looking for a way to give users the best recommendations and optimize the consumer experience.
When a user calls support, he is greeted by Dialogflow, which allows developers to create conversational interfaces for websites, mobile applications, etc. Using natural language processing, Dialogflow understands the user's intentions and interacts with him in the most natural way. Policybazaar reports that their revenue has grown since the introduction of the Virtual Operator. The company associates this with the fact that, thanks to CCAI, recommendations have become more relevant, and user experience in interacting with support has reached a new, exceptional level. To be precise, now the company closes more than 13 thousand transactions per month with total revenue of more than $ 2 million.

Marks & Spencer: Reduces Processing Time


In addition to immediate and personalized self-service 24/7, the Virtual Operator also automates the processing of basic requests and seamlessly transfers the conversation to a live operator in more complex cases - all this thanks to real-time speech recognition.

Marks & Spencer was also one of the first to adopt CCAI and is already reaping the benefits of this implementation. The retailer began using the technology even before it became publicly available, but this did not stop them from accurately redirecting more than 8 million calls from automatic service to the most suitable operators. As a result, the virtual operator was able to identify hundreds of user intentions.

As a result, there is a double benefit: the operators are satisfied, because they don’t need to spend days transferring calls to their colleagues - instead, they deal with the pressing problems of users and issue a solution during the first call; customers are also happy, because service has become faster and better. As a proof, the company cites a figure: the average time for solving a user request decreased by 10 seconds.

Brand loyalty enhancement


It doesn’t matter whether CCAI is used to enhance chat support (PolicyBazaar) or to correctly route to the best operator (Marks & Spencer), AI capabilities help build brand loyalty. What kind of company wants in the current market of services, one bad experience of interaction with support forced the client to go to competitors?

More articles:


All Articles