OK, I admit it. I have a natural prejudice against Artificial Intelligence, Machine Learning, call it what you will. Or I used to…
As an experienced market researcher, I thought a machine was incapable of reviewing telephone interviews, calls, focus group transcripts or any other type of written and audio data, with the same degree of nuanced insight and analysis that a human could bring to the project.
I give you my French to English Google Translate example from my food and drink research days. “Why are all these consumers commenting on the fishy taste and smell of this yogurt?” I mused.
On reference to the original transcripts: “Oh, they mean they can detect a strong peach flavour!” That one was easy to spot, but how many other nuances were slipping through? (pêche = peach, and pêche = fishing, for those linguists amongst you.)
I still believe you need the human layer of intelligence to sift through the machine output, and set-up and fine-tune the machine in the first place, but there really are some amazing applications available now. Imagine being able to analyse every call being made and received (and even missed) for multiple and fully customisable criteria.
This is already an active tool in a growing and highly sophisticated toolkit of AI-powered conversational analytics. Each audio file can be automatically transcribed and assessed for whatever metrics you want to measure including compliance, sales process, customer experience, call outcome/content and positive or negative sentiment of both the call handler and the customer. This is powerful stuff indeed.
The end uses for this kind of analysis are manifold. When deployed in large call centres it reduces staff churn and improves morale. Managers can look at the ‘always on’ dashboard analysis and clearly see which call agents have patterns of low efficiency or negative emotion over time; step in and provide support or training before the situation escalates.
It can be used to monitor customer reaction around new products and services. The model is constantly evolving and learning and can be seeded with specific phrases. Learn quickly, and at scale, which themes and topics are prompting high volumes of calls. This insight can feed into complaint handling, updating FAQs on websites and social media, even new product development.
Maybe you want to check whether your sales team are remembering to promote additional value-add services; whether they remember to introduce themselves at the start of the call; check the customer’s GDPR preferences or any other part of the sales process. Conversational analytics will give you that data. The conversational and text analysis can be combined with additional data sources about the interaction to provide a complete customer journey.
The beauty of the technology is its speed and volume. Robust sampling (or all your calls if you like) offers cost-effective and reliable insight in almost real time so you can action and implement process changes or training modules quickly and effectively.
Even with a more open-ended approach, the conversational analysis shines a light on trends and patterns within your call data. This can then be explored in more detail, down to the individual call transcript, giving you access to rich, qualitative insights. You will discover answers to questions you didn’t even know you had, helping you to drive business operational efficiency, a high quality customer experience and innovative products and services.
Yomdel offers a suite of AI-powered conversational analytics tools. And if you need it in French, we do that too. Au revoir for now.
For more information please feel free to contact me anytime, email@example.com