A growing number of customer service and IT leaders are signing off on investment in speech analytics, encouraged by the technology’s increasing maturity and the continued spread of best practices and use cases.
However, while this combination of factors has ensured that speech analytics now represents a much safer investment for risk-averse businesses, a number of obstacles still remain.
And with organisations increasingly aware of the range of benefits that speech analytics can deliver, it is important that their eyes are wide open to the potential problems that they can encounter, before undertaking what can be a very costly project.
With this in mind, experts have identified the most common challenges that companies can experience.
Because of the costs, building a business case can be tough
“These aren’t cheap investments - they can quickly run into hundreds and hundreds of thousands,” warns Jim Davies, research director at Gartner. “And when you’re building a business case a lot of the time you’re dealing with theoretical benefits. All the use cases are theoretical, the only reason you get those outputs is after you have done the deployment and you’re six miles down the line, and then you can say ‘yes it has helped us with our collection and helped us reduce churn’. So this is a challenge that companies have to go through, particularly in light of the high cost of deployment.”
Mark Pritchard, customer experience specialist at Kcom, adds: The technology is complex and real-time analysis requires expensive processing so cost of ownership is high. Export of audio streams or recordings is considered to be a security or data protection issue. Speech analytics is not top of mind for organisations when budgets are set. The Return On Investment relies on unlocking value which is often linked to customer care and process improvements that are difficult to quantify.”
Speech analytics is anything but plug-and-play
“Many organisations mistakenly believe that if they just plug-in speech analytics and have it analyse their calls, the software will immediately deliver insights about their business. Unfortunately it’s usually not that easy,” notes Sean Murphy, director of product marketing at Genesys.
Donna Fluss, president of DMG Consulting, says: “The challenge to making it successful is that this is not A Field of Dreams where you build it and they will come. You put it in and then it requires a significant amount of work and resources to make the applications work. You have to fine-tune the application, which means you don’t put it in and walk away and it works – you implement it and then you improve your definitions and then you enhance your definitions and you work at it until you get it to work for your operational environment.”
Speech analytics requires specific skills
“The organisation has to have a degree of expertise in-house – there has to be the analytics guru that is using the system as their day job. And if the company hasn’t got that time and expertise internally then maybe having a managed service is the way to go, rather than an on-site or Cloud-based deployment,” says Davies.
Murphy adds: “Speech analytics, like any analytics software, needs to be driven by the specific business requirements of each organisation at each specific point in the organisation’s business cycle. In order for this to happen, skilled business analysts need to always be closely involved with every speech analytics project - it cannot be treated as another IT project in which the IT department installs the software and the project is done.”
“So the biggest obstacle to successfully using speech analytics is people related - the need for people who are very familiar with your business who are also willing and able (and available) to become familiar enough with speech analytics to be able to recognise opportunities to leverage speech analytics in your business.”
Speech analytics requires change management
“The issue is more the application of the findings rather than the accuracy of the applications,” explains Fluss. “More often than not, organisations are not positioned to use the results – speech analytics can give you the results but the challenge is being able to apply those results not just in the contact centre but throughout the enterprise.”
Murphy agrees. “Speech analytics can identify opportunities for improvement that are extremely valuable, but often business processes and/or people’s behaviours need to change in order to capitalise on those opportunities. Such changes are never easy to make, but when proper change management processes and procedures are in place it becomes much easier to implement such changes. A very big obstacle is the willingness of the organisation to build an enterprise-wide change management process to support the application of the findings from speech analytics.”
Selecting the right speech analytics technology isn’t straightforward
“The technologies are so different. There are different fundamental approaches in how you get to that insight – the phonetics, the transcription and the key phrase matching – and trying to work out what is going to be the best underpinning technology for what use case is one of the biggest challenges,” says Davies.
“That lack of understanding of what is going to be the best approach for the organisation is compounded by the vendors, who do a very good job of upselling what they’re good at and down playing what they’re not so good at based on the approach they have chosen. So from an external customers point of view it is a very difficult landscape to navigate because everybody appears to be good at everything, regardless of what technology they are using.”
Technology system integration can be difficult
“There are a lot of different systems involved in the contact centre, where you have multiple technology systems, for example, your IVR, your agent desktop, your CRM and perhaps your social media monitoring tools,” notes Omer Minkara, senior research analyst in the customer management technology practice at the Aberdeen Group. “If those systems are not speaking to each other, you will not have a single view of the customer and that will reduce agent productivity. When using speech analytics as part of such disconcerted activities, agents would need to navigate multiple screens by logging into a different application for speech analytics while they’re on the call, or after they finish the call, and that obviously is not the best use of an agent’s time. While companies are focused on integrating the agent desktop with speech analytics to overcome this challenge, integrating speech analytics insights with other systems such as CRM remains to be an area more companies should focus on to maximize the benefits from their speech analytics investments.”