Go BackYour customers are calling. But do you know why?
To be successful, contact centres must be both effective and efficient. 'Effective' at meeting key objectives (from delivering high quality service, to resolving a high percentage of customer queries first time, to resolving technical issues) as well as 'efficient' in terms of their use of resources. Life would be easy if their only objective was to be effective - but no organisation has a bottomless pit of money and resources to throw at its service operations.
How organisations meet these dual objectives of efficiency and effectiveness is the million dollar question. Offshoring and IVR (i.e. automated call handling) appeared to offer many advantages in terms of cost saving but their introduction has sometimes been at the expense of quality. Conversely, quality management and call recording manufacturers may have created systems that can improve quality, but the data generated often requires painstaking manual analysis - tying up the time and energy of Team Managers and Call Centre Managers whose time could often be better spent training Advisors and new recruits.
What's clearly required is a combination of the two approaches: the cost-effectiveness of automation, with the power of systems that can analyse and improve contact centre performance and quality.
Step forward speech analytics solutions.
What is Speech Analytics?
Speech analytics solutions enable organisations to reign in unstructured information contained within customer contacts, to organise and categorise that information and, via automated data mining processes, to deliver instant insights into service quality, customer satisfaction and agent performance.
From a system point of view, they sit on top of contact (i.e. call and email) recording systems - and work in conjunction with other quality management tools in the following manner:
Step 1: Contact recording solutions will generally record a sample of customer interactions (although some systems record 100% of interactions) and then organise and store those contacts as users dictate, for example, by existing customers/prospects, sales/service/retention/technical assistance contacts, Gold/standard customers etc.
Step 2: Quality management (QM) solutions then enable supervisors/ managers to manually listen to stored customer contacts and to analyse/score them. Automated quality management tools can go further, identifying events within calls (such as pauses within conversations, raised voices, advisors accessing certain pages within a CRM system, callers being transferred to another department etc.) that may indicate positive/negative customer experiences, effective agent performance, regulatory compliance and more. Until recently, however, QM solutions have not been able to analyse 'what' is being said during interactions.
Step 3: Speech analytics solutions take QM a stage further by using speech recognition techniques to search within recorded phone contacts. They work by ingesting recorded calls, indexing and cross-referencing them, and then evaluating words and phrases to create a picture of common themes between recorded calls. Organisations can hence identify anything from specific products or competitor names spoken during customer calls, to swear words indicating customer dissatisfaction, to key words such as 'buy', 'close my account', 'thank you' and more.
First generation speech analytics solutions are now used by a large number of quality and process managers to quantify the impact of specific influencers on their business, their customers and their employees - and to find the proverbial 'needle in a haystack.' However their use is now being superseded by more advanced and powerful solutions.
Second generation solutions
The latest generation of speech analytics solutions takes the capabilities of this exciting new technology to new levels, guiding managers into areas where they previously had no specific knowledge or visibility. Second generation systems typically enable organisations to:
Analyse customer interactions across multiple communication channels (i.e. voice, chat and email) and languages to form a conceptual understanding of all customer interactions that take place in the contact centre
Create search criteria that can be saved for repeated use, easily updated and trained to search for specific types of interactions and alert users as new content is available - providing instant access to customer issues or coaching opportunities
Categorise and organise data based on 'related' concepts - and instantly notifying managers and supervisors to provide an immediate pulse on hot customer issues
Detect sentiment, emotion or stress on the customer and agent side of the conversation - marking each speaker with a different colour during playback and automatically marking areas that contain cross-talk and heightened emotion
Using these advanced tools, system users can select time periods, specific Teams and set their own search criteria. The speech analytics system will then search contacts and present a graphical view (or 'cluster') of common themes present in them.
Users can create sub-views of that data based on themes that are deemed to present the most opportunistic returns. By data mining 'clusters', managers can also quickly get to common themes impacting their business - and then dissect calls further to understand specific drivers or levers.
In theory, the outcome is specific actionable goals that can be focused on increasing customer retention, improving quality, beating the competition, improving existing processes, identifying training needs, identifying and understanding specific customer attitudes and behaviours and more.
But, as every contact centre manager knows, technology theory and practice are often two very different things.
Solving everyday work issues
Implementing speech analytics has been a steep learning curve at Garlands. However, to date, it has been a very rewarding experience.
As a leading UK outsourcer, being proactive and delivering best practice to our clients is a key element in our partnership proposition. And with the introduction of speech analytics, we can now take that to the next level. As well as delivering call quality statistics to clients based on ACD and quality management systems, Garlands now has the capability to tell clients 'why' things are happening - based on an understanding of the actual content of calls.
It isn't a 100% automated solution. Key calls still need to be listened to in order to draw the right conclusions and make the best decisions - that's simply the nature of effective quality management. But in our experience, there are around a dozen elements (or events) within a typical call that need to be analysed to indicate specific training needs, upsell opportunities, missed 'future complaint' opportunities etc. and what speech analytics does very effectively is automate several steps of this painstaking analysis process.
The key areas where Garlands - and its clients - are starting to gain COMMERCIAL benefit from second generation speech analytics solutions are:
1. Identifying the reasons behind unexplained events
We've found that speech analytics is not only useful at finding a needle in a haystack; it's also extremely effective at finding the needle you never knew was in the haystack. Let's say you have a high incidence of people calling to cancel their accounts in a particular week. Have all these customers simply decided to churn at the same time - or could it be that a competitor has come up with a cheaper offer? Only be delving into the content of calls will you find the answer.
2. Analysing the root cause of issues
Just because a customer calls up about a 'payment' issue doesn't mean that the root cause of their problem is a payment issue. Payment could be more of a symptom than the root cause of a grievance. Speech analytics solutions will look within categories to spot trends (such as the high incidence of particular words and phrases) and hence identify the root cause of issues - as well as find 'cause' and 'effect' relationships.
3. Understanding contact centre performance
Speech analytics can be highly effective at identifying the reasons customers are calling. It can be equally effective at identifying why your Teams are performing in a particular way - or why your contact centre personnel are frustrated. What connects people? What drives Team performance? These are all issues we are starting to explore using advanced speech analytics tools.
4. Finding common themes in call types
By analysing call clusters, we are starting to find common themes occurring within specific call types. Within billing calls, for example, we've spotted the odd words that we believe could be driving dissatisfaction - and common 'item codes' associated with pauses in conversations and long call durations. Identifying these common themes and acting to resolve problems are key actions that will ultimately eliminate caller frustration, enhance satisfaction and reduce average handling times.
5. Planning for the future
Speech analytics has a valuable role to play in tactical and strategic planning by identifying consumer needs and desires. What are customers talking about in the run-up to Xmas? What do they think about the latest mobile phone and video games console? What are they talking about that we don't currently sell? These are all questions than can be analysed - as well as 'related concepts' identified.
6. Answering the question 'How effective was our product launch?'
Using speech analytics to analyse the day's calls, an answer can be given the very next day - without the need to send out customer letters, or do costly extra research etc. This can create a massive ROI advantage and enable clients to make quick and timely decisions.
With the introduction of second-generation solutions, speech analytics products have evolved from being 'nice to have' to being commercially viable, with their benefits now increasingly measurable in pounds and pence. Despite the media hype, it's a science in its early stages - and Garlands is very much in the vanguard of developments in this space. What our early steps have shown, however, is that speech analytics can be both 'efficient' and 'effective' - enhancing quality and performance at an acceptable price.
