Page 20 - AeM_May_2020
P. 20
RESEARCH
ANALYSIS
TRENDS
Why AI-based conversational analytics is
the future for first-party marketing data
As regulations like GDPR and CCPA put a stranglehold of customer phone calls every year and have tens of
on the use of third-party consumer data sources, thousands of call recordings banked — just imagine the
making sure you have access to actionable first-party kind of customer data that is in all those calls. You will
data is more important than ever. While you probably learn why they are calling, what makes a purchase
have access to data like customer purchase history, happen, whether they are calling more often for service
website and other digital journey information like paid or sales, whether they are happy or mad — the
search interactions, there are limits to how much insight possibilities are nearly endless. Then imagine manually
marketers can gather from these sources. wading through all those call recordings to gain insights
Conversational analytics represents one of the last on those calls. It is just not possible and as it turns out,
bastions of precise first-party insights into how it was challenging for computers to do it, too.
customers interact with your brand, how they think of
your product or service, and perhaps most importantly, On the computer side, much of the difficulty in
exactly how they talk about it. analyzing conversations lies in the many nuances of
human speech. Unlike the formulaic equations and
Conversational analytics is the process of extracting coded strings of commands computers usually deal
usable data from human speech and conversation with, human speech follows only a loose pattern and
using natural language processing (NLP) to allow logic. Even if we are only talking about analyzing the
computers to “understand” speech (e.g. phone calls English language, there are hundreds of different
and voice assistants) and typed speech (e.g. customer accents, inflections, phrase patterns, varying word
service chatbots), as well as artificial intelligence (AI), usage, slang, and colloquialisms that even other people
to extract and organize data from it. Read more on have a hard time understanding. New research shows
Invoca’s website how it is used in call tracking software that some elements of speech are hardwired in the
that allows marketers to understand call context, human brain, but what really makes people different
predict outcomes, and apply the data to optimize from machines is our ability to instantaneously process
marketing campaigns and improve customer all of this variance of language. Creating a machine
experience. learning algorithm that can “learn” how to process
human language is a whole different ball of wax.
One of the reasons that conversational AI is seen as a
great first-party data source is that today’s consumers When it comes to processing conversations in phone
expect more than the purely digital “point-click-buy” calls, which is what Invoca Signal AI conversational
transactions and are demanding blended experiences analytics does, things get even hairier. “Phone calls are
that bring conversations into the mix. In fact, 70% of idiosyncratic in the world of natural language
consumers feel frustrated or angry when they do not processing,” said Invoca data scientist Mike McCourt.
have the choice of contacting a human representative. “They can be repetitive, can contain both recorded
This means that businesses need better ways to listen. messages and human speech, and often suffer from
In order to hear and understand what your customers bad connections.” On top of that, phone calls also
are saying at any sort of scale, you need contain both full conversations and sequences of
conversational AI to make sense of and take action on simple yes/no answers, hold music, keypresses,
the data. silence, and many other variables that you do not see
in textual communications. This makes it difficult to
Companies that frequently have conversations with design AI software that can juggle so many competing
their customers on the phone are sitting on a goldmine needs.
of customer data. They may have thousands of hours
May 2020: Natural language processing & conversational analytics: data quality beyond reproach 20