Companies are experiencing an unprecedented labor shortage. There are nearly three million fewer American workers participating in the labor force today compared to February 2020 (U.S. Chamber of Commerce).
Complicating matters further, call/contact centers often have employee recruitment, engagement, and retention issues.
For call center operators, this problem ranges from front-line employees who are handling the surge of pandemic-driven eCommerce sales to industries like healthcare where agents who are clinicians earn $100,000, $200,000, or even $300,000 a year. Regardless, the need for agent productivity has never been greater.
Historically, the best opportunity to address call center labor issues has been to focus on deflecting calls with IVR systems, chatbots, or similar technologies.
However, these platforms are only suitable for simple calls that could otherwise be fielded by entry-level employees. The technology has simply not been available or reliably able to automate other aspects of call center operations, such as on-call support and after-call note-taking, especially for highly trained, expensive agents.
But in the last year or two that situation has radically changed…
Genuine AI to the Rescue
If there is one label that has been over-hyped in recent years it is artificial intelligence or AI. Many people are undoubtedly suffering from “AI fatigue,” as they have been endlessly pitched AI-driven solutions that are nothing more than the previous solutions, conveniently relabeled.
But look carefully through the marketing hype and you will see there has been a step-change in AI performance that enables the ability to automate, analyze, and improve call center workflows in ways never before possible.
Advanced AI technologies can automate the note-taking process and save up to 75% of the total after call work (ACW)…
In many complex call center instances, it is possible to see a reduction of up to 35% in AHT by using state-of-the-art AI tailored to call center workflows – particularly those that involve extensive note-taking.
How Is That Possible?
Let’s look at an example from a medical device customer service center, where patients are calling to troubleshoot an issue with a medical device.
In this situation the calls are typically close to 10 minutes long. The agent then spends nearly the same amount of time documenting and classifying the calls, writing 100-word summaries, and completing other required actions in the company’s system of record (e.g., ServiceNow, Salesforce).
In this instance the agents are highly trained, earning about $100,000 a year. Here, note-taking is approximately 50% of the total handle time.
Advanced AI technologies can automate the note-taking process and save up to 75% of the total after-call work (ACW) by auto-classifying and summarizing the conversations. In many situations where that ACW is 50% of the total handle time, which represents a reduction of up to 35% in total handle time.
Typically, the AI platform will capture the live call center audio stream, automatically transcribe the call, and feed the transcript to a series of AI models that will classify and summarize the call in just 10 to 15 seconds.
When properly configured, the AI feeds those classifications and the summary to the organization’s existing system of record where the agent can review the output and make edits, if needed.
Even with that review time, typically organizations capture several minutes of savings, which will only increase over time as the agents’ edits are used to regularly update the AI model to improve performance.
How good can these summaries be? How can you ensure consistency and quality? What are the downsides?
Even in complex instances the quality can be extremely good. Below are a few of examples of redacted summaries from complex calls.
The first example is from an eight-minute medical device customer service call. The patient was calling to troubleshoot an implanted device.
In this situation the summary the agent needs to write may ultimately make its way to the Food and Drug Administration (FDA). Therefore, accuracy and completeness are critical. Can you determine which summary was written by the agent versus AI in less than 15 seconds?
Next is an even more complex use case, with a six-minute call between two cardiologists who are discussing whether a patient should be approved for a computerized tomography (CT) scan.
This is a process known in the healthcare industry as prior authorization. In this example, the clinicians are earning several hundreds of thousands of dollars a year. Yet they spend nearly half of their time documenting what they just did on the calls.
Notice that what is generated is a fully written note – not a series of bullet points or highlighted suggestions for what to include in the summary. It is a fully formed paragraph or longer that emulates the clinician’s work.
AI-based solution vendors…can fine-tune their models to deliver results that reflect the work of a company’s top
Properly executed, this level of quality can be generated within just a few weeks of configuring models to each organization’s use case – regardless of industry. This can result in millions of dollars in annual savings. But cost savings is not the only benefit. It is also possible to reduce cost while improving quality.
Lower Cost, Higher Quality
As every call center operator knows, agents vary greatly in terms of how they complete their work. That is apparent in how calls are classified and summarized, including the content in a summary, and how long that work takes to complete.
AI-based solution vendors that truly understand call center operations, process improvement, and data can fine-tune their models to deliver results that reflect the work of a company’s top performers.
The market for this type of sophisticated
automated note-taking is emerging.
In the language of process improvement, you recenter and tighten the distribution by presenting to every agent within 15 seconds a draft summary that was written in the style of your best people. In plain English, you get better quality, cheaper, and faster. And your newest agents benefit directly from the quality generated by your best people.
What You Need To Know
The market for this type of sophisticated automated note-taking is emerging. Call center operators will want to avoid investing time, money, and effort into solutions that ultimately cannot deliver the promised value.
What are some of the key questions to keep in mind as you consider this new type of call center automation?
- Can you handle live audio streams? If so, from which suppliers?
- What service do you use for live call transcription, and how good is the quality?
- What type of summaries do you generate? Bullet points, key point highlighting, or full paragraphs?
- How good are the summaries at implementation?
For call center operators:
- How much time do your agents spend on after-call note-taking?
- How much time do your agents spend during the call that extends handle time to complete notes while talking to the caller?
- Do you have call audio streaming available?
- How much could you save by reducing average handle time by 35%?
It’s an exciting time in the world of applied AI and call centers, as best-in-class natural-language processing is now able to automate even complex note-taking. This opens a path to a step change in operational performance, as call center operators can think about shaving minutes off each call, not just seconds.