AnalyticsBusiness OptimisationSpeech Analytics

Speech Analytics – Inhouse or Managed Service?

One of the hottest new technologies around at the moment is speech analytics. It’s promise is that it can mine ALL customer conversations to unlock insights and identify patterns of behaviour. But does it deliver on this promise and is a managed speech analytics solution better?

The answer is…it depends. The elements of a successful speech analytics deployment come down to:

  • Having the right technology in place (and if you are buying this means you need to the same brand of call recording technology)
  • The right people – neither quality monitoring or IT people have the right skill sets
  • The right structures – so that insights are taken seriously, acted upon and managed

This is not a plug and play technology. We have been working with companies all over the world and found out that huge investments have been made in licence fees, fine tuning the speech terms and actually getting the technology to work only for executives to find that the results are not actionable and do not realise business case benefits.

So why are companies purchasing technology rather than leveraging from a managed service? The answer comes down to people.

Most contact center professionals know the industry is shrinking, digital service is growing and that big data analysis is popular.  They see speech analytics as a way their contact centre can deliver value. And they are right – this technology does raise the contact center profile in the age of digital service by providing insights.

However these professionals don’t know what they don’t know.

It isn’t until the ink is signed on a purchase order and the tool is deployed that these people realise that just getting accurate hit rates  takes a massive effort but is only a small part of the journey.  They focus their efforts into getting accurate hits, but they do this not understanding the massive limitations of the technology. Some technologies have been tested as having a less than 22% successful hit rate – a fact that is not usually communicated by vendors.

Here is a typical transcription from Speech to Text type speech analytics solutions as outlined in the latest academic research.


Source: Kopparapu , 2015

As you can see, the results are not accurate. There are a lot of reasons for this which we can tell you about, but suffice to say it is not a plug and play technology.  Some solution providers offer packages that include phrases relevant to a particular type of business, which means the initial discovery and implementation time is reduced somewhat but even this does not resolve the next challenge.  That challenge relates to this……

Data on it’s own produces no meaningful insights.  That is WHY big data is a hot topic right now.

It may well be really nice to know your customers say x about y, but unless this is analysed statistically to know if this is a meaningful trend and how large the trend is, it is useless information. Without a background in statistical analysis techniques, interpretation and data visualisation techniques, the data produced is useless.

This means having a team working on speech analytics that really understands the business, the technology and can test and interpret insights is critical.  These teams do not come cheaply and their skills are not usually to be found in most contact centers – without taking years to properly train them.   It is also a shock to most executives who have signed the purchase order on the basis that the solution costs are offset by staff savings and don’t realise they are replaced with more expensive resources.

For all of these reasons, many companies are opting to purchase speech analytics as a managed service. The Aberdeen Group report ‘The 2013 speech analytics buyer’s guide’, estimated that in 2013, a quarter of businesses outsourced the management of speech analytics activities.  This figure is growing rapidly.

It is akin to the cloud based technology purchases and moves away from the old school purchase everything IT acquisition mindset.  Rather than going through a long, expensive and arduous journey to acquire, deploy and interpret data, these companies leverage from the insights gained to focus on changing their business.  After all changing business practices is tough in and of itself.

Whatever you decide to do, remember, successful speech analytics deployment is an ongoing journey that requires four parts:

• Understanding the technology and it’s limitations
• Getting accurate results
• Interpreting the data
• Using the insights to effect changes to the business.

Therefore, organisations deploying speech analytics really need to consider what they wish to achieve with this technology and where they are best placed to focus their limited resources – on acquiring and deploying or on making the necessary changes needed in their business.

Speech Analytics – Inhouse or Managed Service?

A managed service solution produces better results faster than an inhouse purchase simply because you focus on making changes rather than getting a technology to work.

Talk to us about our Speech Analytics Managed Service and see how it has been deployed since 2009 across 8 countries and multiple languages.

Leave a Reply

Your email address will not be published. Required fields are marked *