By our CEO Christine Barr for CIO magazine.
The best BPO analytics today operate in real-time, monitoring activities as they happen and flagging potential problems right away.
One of the most overlooked truths about AI’s Machine Learning (ML) is that the ML is only as powerful as its training. And the more datasets from just as many diverse environments it is trained with, the more powerful the results. That kind of ML delivers better answers and the ability to spot patterns and pattern deviations that few enterprises can find on their own.
But some enterprises opt to hold onto their core datasets and only outsource datasets they consider to be non-core. This can undermine any Business Process Outsourcing (BPO) strategy for two reasons. First, core datasets are presumably where the enterprise’s most critical data resides. Why choose to not subject that data to the most extensive analysis possible? Second, by deliberately withholding those critical datasets from the ML analysis, the results are, at best, skewed, and at worst, misleading. Without knowing that other core dataset information, the analysis will only draw conclusions from what it is given, depriving the enterprise of critical but undiscovered patterns.
Business intelligence at work
The best BPO analytics today operate in real-time, monitoring activities as they happen and flagging potential problems right away. Consider a customer talking with a customer service representative and the customer starts to sound agitated. The software can instantly alert a supervisor that there is a potential issue, most likely long before the representative would likely ask for help, assuming they ever would.
This isn’t AI after-the-fact analysis. To truly address the customer service challenge, AI needs to be real-time, constantly listening in on calls for potential escalations. This works extremely well and one of the key reasons is human nature. When customer service representatives take their calls — especially when they know their calls might be monitored by a supervisor — they are going to try and be as impressive as possible. When a customer becomes frustrated on the phone, instead of immediately reaching out to a supervisor, some representatives will instead try to handle the situation themselves in an effort to demonstrate their abilities to their supervisors.
This rarely helps the customer, the representative or the enterprise. But allowing external AI to monitor and take over whenever a potential problem is detected, the situation will likely be resolved more quickly, efficiently and with a better outcome. And the representative can be freed up to immediately move on to the next call. The improvement opportunity doesn’t end there. The data captured can be used to drive enhancements to process and product, boosting revenue and removing customer friction.
Where companies go wrong
This brings us into the area of technical debt, where an enterprise ends up spending more money, time and other resources and enjoys fewer benefits by trying to crunch the data on its own instead of outsourcing the process to the right partner.
Technical debt refers to the cost of additional work necessitated by initially opting for a more cost-effective approach. That cost can be seen as literal costs, in the sense of spending more money to achieve the initial objective. But, with AI in general and ML specifically, it can also be the lost opportunity revenue. More sophisticated analysis of as much data as practically possible is going to deliver better insights, which in turn can be converted into both cost savings and revenue boosts.
What is customer interaction analytics and why do you need it?
“Analytics” is one of the biggest buzzwords in the business world—and now the call center world is buzzing about analytics as well. In BPO, the benefits are not viewed solely in patterns detected and recommended actions. It’s about the tangible results of those actions, as measured in revenue boosts, retention of customers (who might have otherwise left) and greater efficiency in customer interactions. That greater efficiency in customer communications in turn allows the customer service team to take less time to solve the customer’s issue, which means that the same size call center can handle more calls and customers are given shorter hold times. Further, the BPO ML identifies real time feedback to enhance and optimize the end-to-end customer journeys, by client personas – ultimately resulting in proactive identification of the client-preferred channel and enabling a more likely satisfied client and increased ROI.
When attracting call center employees is as difficult as it is today, it can make a huge difference with recruiting by allowing a call center to be operational with fewer people. And those shorter hold times and potential AI intervention capabilities make it easier to keep customers happy.
But the BPO ML opportunity goes well beyond customer service to other areas, such as online transactions and email chat capture.
As efficient as online transactions and email chat functionality can be, they have serious downsides. For a healthy percentage of customers, they work wonderfully by answering the question explicitly and quickly. That happens when the customer has an issue that is common and where the team has crafted a precise and helpful response.
But what about less-common issues? Or instructions that are dated or made irrelevant by a graphical user interface change? For example, an answer tells the customer that the information sought can be displayed by going to the drop-menu under Settings and clicking Payments. But what if Payments is no longer displayed under Settings? The customer gets very frustrated and will likely blame the brand.
What if, however, there was a constant button that said “Is this answer helpful or not? If unhelpful, please click to speak to a representative in fewer than three minutes.”? That’s better, but what if the frustrated customer instead gives up and clicks over to a rival’s site to make the purchase? Not good.
With BPO ML, the system can detect frustration. Perhaps it looks for someone repeatedly asking the same question. Or perhaps it identifies when someone doesn’t follow online instructions, potentially because they can’t? It detects frustration in a wide variety of ways and instantly connects the customer with a supervisor to calmly resolve the issue, keeping the customer happy and preserving the sale.
Now that’s more like it. We believe there is a better way to ensure your outsourced solution is executed with unsurpassed consistency. For more information about BPO ML, visit us here.
