VoiceBase and Tableau Deliver New Insights Through Speech Analytics
Helpdesk phone calls constitute first contact with a company's customers as well as an opportunity to make a good impression, solve issues, or make a sale. This makes the information contained in these calls hugely valuable yet somewhat difficult to access. In other words, voice calls represent a huge opportunity. According to Gartner Research, more than 90 percent of customer conversations are still happening on the phone and generating a staggering amount of valuable data to companies. Speech analytics is surging and is expected to become a billion dollar industry by 2020, according to MarketsandMarkets Research.
Voice conversations can drive better customer experiences as well as generate valuable feedback. Speech is a more nuanced and accurate analytical tool used for gauging customer response. This is especially true in helpdesk environments where unfavorable customer experiences can lead to frustrated clients, erosion of brand value, and lost sales.
Every day, there are 56 million hours of customer phone calls taking place; that's roughly 400 billion words spoken. More importantly for businesses, this data can be a focused source of customer input and business intelligence (BI).
(Image credit: Statista)
What Is Speech Analytics?
Speech analytics is the process of extracting meaning from audio recordings so these can be analyzed using artificial intelligence (AI) to parse out data that businesses can use for deeper insight into the conversation. Speech analytics software can take hours of existing support calls and employ AI to separate multiple speakers on a call, detect the emotional state of callers by analyzing for cues in voice pitch and tone, and uncover and track often mentioned keywords.
"Speech [analytics] in general is pretty mature, having been honed, tested, and refined in call center settings and elsewhere," said PCMag's BI and database expert Pam Baker. "Speech-to-text is common for voicemail messages and that's a highly mature form of speech [analytics]. Once converted to text, the analysis work is pretty much the same as it is for any other text-based input."
From Conversations to Dashboards
Much of the data used for speech analytics comes from cloud-based voice-over-IP (VoIP) systems that have automatically recorded calls and other forms of interactions, including text chats and video conferences. For the most part, this data remains on the servers running the cloud PBX, which makes for a good fit with speech analytics solutions because, as long as these platforms are also deployed in the Software-as-a-Service (SaaS) model, they're easily integrated with the VoIP system or call center.
AI-powered speech analytics vendor VoiceBase recently teamed up with data visualization and BI market leader Tableau. By using VoiceBase's solution, call center audio recordings can now be parsed and then made available as a data source in an enriched text format that Tableau Desktop can use to deliver rich visualizations.
The result is that companies will have access to insights they simply didn't have before. These include using natural language processing (NLP) to surface keywords and topics that make recorded content discoverable. Machine learning (ML) is employed to expand speech analytics and generate conversation metrics, resulting in call drivers and business trends. This information can be used to improve call center interactions, streamline call agent scripts, and highlight product or service areas that could use improvement.
"I would think speech analytics would be a natural fit for BI vendors that are already geared to use natural language querying and audio or video data mining. Other BI vendors might have to do more work to make it fit, but it still makes sense to do so," Baker said.
Once available through a BI vendor's interactive dashboards, users can drill down into their company's calls to understand complaints, competitive mentions, agent interactions, over talk, sales objections, and churn prediction (that is, predicting whether customers will cancel a service or a product). Predictive analytics are used to detect complex events, and predict future customer behavior that's based on past calls and patterns.
How Voice Data Visualization Works
Applying AI and ML technology to voice calls means that conversations need to be turned into quantifiable and actionable data streams. In the case of VoiceBase's solution, these data streams are then categorized into several data feeds. These comprise a wide range of analytics, including Call Predictions, Call Categorization, Conversion Metrics, and Transcription. Once viewed through a BI lens, these analytics can help give users a snapshot of brand health, competitive analysis, the customer journey, marketing campaign analysis, agent monitoring, and sales optimization, to name just a few possibilities.
"We have seen a big trend in our customers' desire to better leverage speech analytics data, which has historically been trapped in the call center, and to correlate it with the massive amount of BI already being served by Tableau," said Jay Blazensky, co-founder and Chief Revenue Officer (CRO) at VoiceBase.
"In the case of speech analytics, the added value for any BI vendor is higher," Baker explains. "That's because this form of data and analysis has historically been limited to call center activities—for example, analyzing phone calls for customer sentiment, complaints, escalations, resolutions, and other things related to customer retention and brand reputation. Adding this call center data to the mix of other data renders more comprehensive and nuanced outputs for businesses to act upon. Further, speech analytics can be expanded beyond the call center so that even more data can be harvested and mined."
This article originally appeared on PCMag.com.