What is Amazon Connect Wisdom? How
does it work?
When Jimi Hendrix said:
“Knowledge speaks, but wisdom listens”, he had no idea that he would be summarizing the functionality of Amazon Connect Wisdom. (Wisdom).
Wisdom combines speech analytics (the “listens” part) with a database search and retrieval tool (the “speaks” part).
Speech analytics processes what is said and identifies key phrases. These are used to search various databases to find appropriate materials.
By automating the search for these materials, it reduces Average Handling Time and improves customer satisfaction.
Traeger Grills, an Amazon Connect Wisdom user in the US said: “So far, agents using Amazon Connect Wisdom have seen an increase in customer satisfaction and first contact resolution of roughly 15%. They have also decreased their call handle times by roughly 15%.”
Here are some scenarios where Wisdom, integrated with the Omningage Agent Desktop, might be used to improve productivity and CSAT while cutting costs.
Scenario 1: Agent’s channel only, provides information to agent only
The system is set up to monitor what the agent says. Agents are more likely to use key phrases consistently than customers.
Speech recognition system identifies agents’ key phrases and intents more easily and with less training, and can retrieve and present the appropriate materials to agents faster.
These “appropriate materials” include user guides, price lists, mandatory terms and conditions or even troubleshooting scripts, with the prompt for the next script item based on key phrases an agent might say. In this way, the system helps the agent to navigate their way through the workflow.
Scenario 2: Agent & customer channels, provides information to agent only
The system can be set up to react to what both agent’s and customer’s utterances.
When agents ask the customer “How can I help you?”, the natural language processing (NLP) behind Wisdom recognizes the customer’s intent and brings up the appropriate materials or script prompts for the agent.
It is trained to recognize and react to the customer’s answers and bring up further script items, without the agent having to type in any instructions on the keyboard.
Scenario 3: Integration with a bot – no agent required
This scenario is more likely for digital text-based channels. Customers write their requests, Wisdom recognizes the customer’s intent, then it presents the customer with further questions.
These responses are then recognized and prompt further questions to take the customer through the workflow.
This scenario will dramatically improve productivity since no human talent is required to handle these requests.
Reporting and management
Managers want to monitor and manage how these tools are used to make sure that they get the most value from them.
Managers want to know how good Wisdom is at assisting agents and customers.
There are a number of options:
Percentage of calls where manual override used: Agents are the best people to judge if materials selected are relevant or not. A report showing how many times agents manually overrode the automated searches will be a good indicator of how well Wisdom is selecting the appropriate materials. This can be analysed by topic / category, queue, team and by agent.
Analysing First Contact Resolution (FCR): If failing to achieve FCR can be defined by a customer calling back from the same number within, for a specified interval, then Wisdom’s effectiveness can be measured, in part, by the FCR percentage achieved. These numbers can be analysed by topic / category, queue, team and by agent.
Customer sentiment: Amazon Contact Lens can analyse customer sentiment in response to the advice provided by Wisdom. Positive and negative sentiment can be analysed by topic / category, queue, team and by agent to find out what the system is doing right and wrong.
Customer reactions: Phrases where customers express gratitude or acknowledgement that their problems have been solved can be detected and reported. This can be used as a back-up check to customer sentiment. These data can be analysed by topic / category, queue, team and by agent. They should be broadly consistent with the customer sentiment figures.
If you’re using, or planning to use, Amazon Connect as your PaaS, then contact your Amazon Connect partner or Omningage’s Sales team to find out more about how we can integrate Wisdom with your contact centre to improve your productivity and customer satisfaction.