Bespoke Conversation Intelligence Software with Speech Analytics for Call Center Quality Assurance (Automotive Industry)

The company has 8,000 employees and 210,000 m² of automated warehouses. It has 500 branches. Customers can order from a catalog with more than 45 million products, either in stock or available to order.

Over 400,000 calls come in each month. It is very important to ensure that every call is high quality. Otherwise, customers will be dissatisfied and will not make purchases.

The company wanted an AI system that analyzes what callers say during calls and determines if the representative is helpful and professional.

• Transcribe and analyze 100% of company calls.
• Implement conversation intelligence to process all call data and extract strategic insights related to products, service quality, sales, and trends.
• Apply speech analytics for quality control of call center staff.
• Implement the technology inside the client's secure IT environment, with no data leaving internal systems, and provide deep integration with existing IT systems (CRM, ERP, SAP, and reporting tools).
• Use these data sets to build dashboards with KPI breakdowns based on access level: call center agents, supervisors, quality assurance managers, sales department heads, business unit leaders, and top management.

Stage 1: Discovery

• Identify requests for quality improvement (gather requirements from business line managers, sales directors, and the quality team; study current processes; analyze existing calls)
• Review existing evaluation methods and checklists
• Annotate company calls
• Define system requirements, including the necessary metrics and data for analysis.

Stage 2: Design of a Conversation Intelligence and Speech Analytics System

• Figure out how your new software will connect with and fit into the systems and routines the call center staff already use.
• Developed new parameters that were not previously available to extract data from calls and analyze for customer sentiment index, agent productivity analytics, cross-sell and up-sell attach rate, emotional analytics, negative sentiment rate, agent engagement score, and performance trend analysis.
• Build an automated process that takes call audio, turns it into text (transcription), and then checks this text for expected patterns (sales signals).
• Create user screens and dashboards that show the right information for each group: agents, supervisors, sales directors, and executive management

Stage 3: Implementation of a Conversation Intelligence and Speech Analytics System

• Configure the visualization module to collect, process, and display results using charts, tables, and dashboards.
• Test the model with a business unit that covers 20 percent of the company's calls.
• Test the model on 50 percent of outbound warm calls.
• Evaluate whether the system is working as planned.
• Roll out to cover 100 percent of company calls and scale across all branches and employees.

Stage 4: Conversation Intelligence and Speech Analytics System Support

• Keep checking how well the system works after it is live, and use feedback from the clients' staff to make ongoing customization

Technical results (first 3 months)

• Achieved 100 percent transcription of outbound warm calls responding to website leads. Every call was automatically turned into text, including separating each speaker.
• Each call was checked against six checklists, so auditors could see whether the salesperson followed each of the key required steps and rules during the conversation. For complex cases (20 percent), the system enabled double-checking of results.
• Our team successfully delivered dashboards showing relevant call data and metrics for both sales leaders and those responsible for call quality. The system now monitors all calls and evaluates each specialist using the checklists, highlighting weak and critical quality control areas.
• The new conversation intelligence and speech analytics system for analyzing calls was configured to work inside the company's private, protected IT network, so no data could be leaked.

Business Results

• Before, the client's quality team checked only 1 out of every 100 calls. Now, nearly every call is reviewed.
• The company used to review and analyze sales calls once a month. Now, they review and analyze every sales call each week. With this change, they can quickly spot issues and opportunities, make fast improvements, and provide feedback to sales agents sooner. As a result, the sales team started closing more deals. This increased the sales conversion rate by 3.1 percent and brought in an extra 75,000 euros in monthly revenue.
• This quicker feedback also increased sales team improvement (compliance with the checklists) by 10%. The alert system that notifies sales managers immediately when a call is likely to have a conflict or issue with the service level agreements helped reduce the number of customer complaints by 12%.
• Moreover, each sales agent now handles about 76 calls per day instead of 70 (9% increase). Their After-Call Work (ACW), which is the time agents spend finishing up tasks after each call, went down by 38%.
• The quality team responsible for checking calls now spends 70% less time reviewing calls.
• Because the process is faster, the company was able to save 480 working hours each month. They were able to cut staff by two supervisors without affecting call analysis.

Looking to Implement a Similar Solution?

Contact us. We deliver this platform as a white-label solution and provide you with full rights to a custom version of this software tailored to your company.