Use speech analytics to unlock customer insights and opportunities for
Contact centres no longer
carry the badge of “transactional cost centre” – they are a rich source of
customer insights. One of the most valuable ways of unlocking these insights is
through analysing calls. But manual
review of calls is a costly and time-consuming process, and most contact
centres only review 1-2% of customer conversations – the insights in the other
98-99% of calls are left hidden. Speech analytics, which unlocks customer
insights from mass volumes of calls, has changed all that, bringing big data into
the hands of even small and mid-size contact centres.
Speech analytics is a
complicated, highly specialised technology hence many implementations fail to
deliver on the promised ROI. (A six-figure budget is generally required to
purchase licences, pay for professional services and hire staff to run the
technology.) The team in charge of running a speech analytics program needs not
just a technical, or quality monitoring skill-set, but an understanding of the
psychology of why consumers react the way they do, how they make decisions, and
how to deal with model errors inherent in the hit rates of the system.
Getting the most out of speech analytics with a managed service
The good news is that speech
analytics is also available as a service
– whereby companies don’t need to purchase expensive licences or hire
specialist resources, thus removing the risk.
The only prerequisite for using
a managed service is you need call recordings, metadata and a provider that can
read your call recording type.
While large enterprises with
thousands of contact centre seats may be comfortable they have the requisite
knowledge and talent to buy speech analytics licences and deploy internally,
the as-a-service option is generally more suitable to small and medium-size
contact centres, from both a risk and budgetary perspective.
The as-a-service model also
provides an affordable entry point to pilot speech analytics and build a business
case for longer-term usage.
Types of speech analytics
There are three main types
of speech analytics tools – phonetics, text-to-speech and grammar-based. Phonetics
is the fastest approach to speech processing and requires the least amount of
data storage; text-to-speech converts the entire call recording to text, and takes
longer to deploy.
There is also a difference in when the analytics is conducted –
post-call or real time. Post-call or historical analysis is the most
common, and is suited to deep-dive analysis of trends and mining of calls to
identify customer service and compliance issues. If, on the other hand, your contact centre
wants to set up supervisor intervention when certain words are mentioned (for
example, swear words or words with compliance or legal implications), then
real-time is a good option.
Things to consider
when procuring speech analytics
Some vendors have
few experts on the ground – ensure you can get the local support you need on
an ongoing basis.
Broad accent support
Find out how the supplier’s technology deals with
a range of accents in the local market. The impact of accents is
underestimated in analytics deployments and is especially important in
Determine if the
tool works in your environment and perform your own compatibility testing.
Be sure to start with a small scope of work, test
and retest until you have level of accuracy and then build upon that success.
The most successful analytics engagements start with a proof of concept that
delivers an ROI before moving on to identifying more complex issues.
Buyers considering a managed
service or cloud-based version of a speech analytics solution need to check
whether data and call recordings are being stored or analysed locally or
offshore. Some managed service vendors are not based in the countries they sell
to, and offshoring data is something that can affect compliance with local privacy
Contact Centre Analytics-as-a-Service (white paper)
Send me more information on speech analytics as a service or contact us.
Did you enjoy this article? Please share by clicking the LinkedIn Share icon below!