Data is the fuel of organizations. It provides the energy to the various engines that create, deliver, service, fix, and sell its products and services.
Contact centers help provide and refine that data fuel through their interactions with customers. And, in turn, that resource drives those interactions by providing customers with information, including targeted responses and sales offers. Information that guides them on their customer journey.
Therefore, having the right data, and the ability to extract insights from it in a timely manner, is essential for organizations’ survival and growth.
But could the continual rapid evolution of AI, such as with the advent of agentic AI, and with proposed legislative changes on data privacy like in Canada, impact contact center data management and analysis?
To gain insights into how organizations are using and managing data, we reached out to Richard Boire, a leading expert in data science and predictive analytics who is president, Boire Analytics, and an occasional contributor to Contact Center Pipeline.
Q. What are the top challenges that you are seeing with how contact centers utilize and analyze data?
Essentially, we now operate in an analytics environment where data can be analyzed both in a textual format, as well as the traditional nontextual format.
Within the traditional nontextual format, this is essentially about trying to determine who are our best customers, who are most likely to defect, and what are their likely next actions. This analytics information has always been readily available for those contact centers whose organizations are willing to make the necessary investment in both human capital as well as technology capital.
In the textual environment, the advent of AI has now accelerated the development of new tools which as we all know were pioneered with the release of ChatGPT.
These tools use the discipline of natural language processing (NLP) alongside the mathematics of AI to provide whatever script is appropriate based on the input of the user.
For the contact center rep (the user), this could simply be the conversational responses of the customers. ChatGPT would then provide the appropriate response as one option for the contact center rep.
“As organizations move forward with this technology, the key will be training in not only how to use these tools but how the domain knowledge of the contact center rep can best be leveraged within these tools.” —Richard Boire
Of course, the real limitation here is that the tool is only as good as what has been input or fed into the tool. Because of this, many of the bigger or more advanced organizations will build their own custom “ChatGPT” tools, which are referred to as large language models (LLMs).
For contact centers assigned to a specific product line or if they are a business process outsourcer (BPO), dealing with a specific company and their customers, the use of customized LLMs would always be the preferred approach.
This is easy to understand because the customized LLM is being fed specific information based on emails, texts, and prior conversations with only that specific product line or company. It is easy to see that the superiority of this information input will yield better solutions.
Q. What are the top opportunities available now and in the wings for contact centers to help maximize the net benefits, and obtain optimal results, from their data?
If organizations can overcome the challenges as indicated in the preceding, the benefits are happier customers as their needs are being met. At the same time, we can develop these solutions in a much quicker timeframe.
But the key to all this is having better data and better tools that can use this data for a more optimized solution.
I don’t want to look at these tools as mechanisms to simply reduce labor costs. Instead, these tools really allow organizations to work with their suppliers to build more and better solutions within a given cost structure.
Rather than being seen as a cost minimizer, the better perspective might be one of revenue enhancement, as we can now do more within our existing labor infrastructure.
As organizations move forward with this technology, the key will be training in not only how to use these tools but how the domain knowledge of the contact center rep can best be leveraged within these tools. This training would consist of using both generative AI (GenAI) tools as well as CRM tools.
Let’s look at an example. Here I, Richard Boire, call up CIBC [Ed. note, in Canada] because I am unhappy with my current mortgage. John Smith at the contact center picks up the phone call and immediately begins the identification process.
Once I am confirmed as a customer, the CRM system then provides information that I am a high-value customer. It also identifies that I am at risk of leaving the bank entirely.
“The dynamics of the contact center today require a completely different skillset than the contact centers of yesteryear.”
At the same time, it also identifies that the next best service for me would be a line of credit (LOC) for my mortgage.
Based on this information, the rep would input into the GenAI system that I am a high-value customer but also a high-risk defector where my most likely next product should be a LOC mortgage product. The GenAI system then outputs the following prescribed activities, based on its knowledge of CIBC Products and services as well as CRM knowledge:
- Offer LOC mortgage product at the lowest rate possible.
- Review when mortgage comes up for renewal and present options that reduce monthly payments as well as interest payment.
But this is not the end of the dialog. As the customer is responding back to these prescribed actions, perhaps with the comment about lower rates at another financial institution (FI), GenAI can then respond with other advantages of CIBC’s current mortgage product over the lower rates offered by the other FI competitor.
The rep doesn’t have to use the GenAI script. But they can use it as an aid in delivering a more appropriate response to the client.
The dynamics of the contact center today require a completely different skillset than the contact centers of yesteryear.
But it is a more demanding and rewarding one, as a higher level of skillset is required to meet the consumer demands within a technology environment that abounds in data.
Q. So, what is the impact of the AI “revolution” on data volume and the ability to safeguard, authenticate, analyze, and productively use it in contact centers? Are there new models or tools that can help?
We now live in a world where data volume is like oil and gas. More oil and gas (setting aside the climate change challenges) brings forth more control and power.
The same now holds true for data. Data has always been there, but technology has always been the barrier in both accessing it as well as processing it.
This is no longer the case. Data technologies now allow us to consume infinite volumes of data such as what has transpired with the advent of OpenAI and ChatGPT in Nov. 2022.
Because of these developments, the use of AI is becoming an everyday tool, as discussed in the first two questions.
But AI’s authentication will always be suspect as there are two elements here. The first element is behavioral, which is very reliable based on what the consumer does. However, the second element is based on what the consumer says. This is always going to be less reliable because not every person conveys their thinking publicly.
“The key for contact centers is having access to the data regardless of where it is, and then having the right APIs...”
In other words, you may not be capturing the thoughts of the so-called silent majority and thereby incorporating a bias here which is not reflective of the entire population.
Q. There still appears to be an ongoing debate between cloud, on-premises, and hybrid data environments. Where do you see the data that contact centers handle now and in the future residing, and why? Will some data migrate back to on-premises from the cloud? Other data moving to the cloud?
It will really depend on the organization. In the example above, CIBC could have their data on-premises. But TD’s data is entirely in the cloud while Bank of Montreal (BMO)’s data is a hybrid of both on-premises and in the cloud.
The key for contact centers is having access to the data regardless of where it is, and then having the right APIs to move the data back and forth to provide the reps with the right information in real time as they interact with the customers.
Q. Let’s discuss data security. It appears the bad people are becoming more successful in stealing data. Are organizations – and their contact centers – as secure as they should be? In responding to data breaches?
There is no real hard answer to this. The key for any contact center is to hire an outside data security consultant who is extremely well-versed in this area. They need to review their existing firewalls, as well as explore in great detail how they move or transfer data between different locations and platforms. This would require a review of their existing APIs that execute the data transfer process.
Q. Does AI aid or hinder data security?
There is a two-edged answer here in terms of limitations and benefits. On the limitations side, we have more data which increases the likelihood of more data security risks. Because of its insatiable appetite for data, the use of AI has certainly accelerated the level of concern in this area.
Yet, the benefit with more data is that these AI tools can also be used to predict events or processes that are more likely to encounter these risks.
Q. What effect do you see with recent and proposed legislation and regulations in Canada (e.g., Bill C-27 that would have to be introduced in the next session of Parliament, Quebec Law 25: (see this article) on data management, privacy, and security?
The current legislation, the Personal Information Protection and Electronic Documents Act (PIPEDA) in Canada, which was introduced in 2003, attempted to provide regulations or a paradigm on data and its use by companies. There are 10 main principles in the legislation which can be summarized into three main goals:
- How will you use my data?
- Did I give you consent?
- How will you protect my data?
This legislation was very appropriate at the time in helping organizations in how to best manage and use data in creating optimum business decisions. But at that time, the internet was still relatively new and social media was non-existent.
So, the Canadian legislation needs to be revised in light of the current digital environment that is so pervasive in our everyday world. The General Data Protection Regulation (GDPR) legislation developed in Europe in 2018 represents more up-to-date legislation in attempting to deal with the more current realities of our digital environment.
“The new data paradigm requires an individual who is proactive in using data to best serve the customer...”
Right now, Bill C-27 is on hold, but I would expect the legislation to further reinforce the concept of privacy given the digital ecosphere that we all live in. More disclosure and transparency is required to ensure what data from the consumer is readily available to third parties as they operate in this ecosphere.
Q. Agentic AI has recently burst onto the scene with many conversations on it. What are your thoughts on agentic AI and what are its implications for contact center data management and analysis?
Agentic AI is a new level of cognitive reasoning or the increased ability of AI to think. In traditional AI, the machine provides answers to specific tasks or steps that the human provides while in agentic AI, you give the machine a goal and it outlines the steps or tasks as well as options.
A good example of this in the contact center might be an irate caller in Canada who is unhappy with his registered retirement savings plan (RRSP). The goal would be to make this client happy with his RRSP product or some facsimile of this. The machine would outline all the tasks as well as the options that are required to achieve the end goal.
Q. What are your recommendations to contact centers to help them manage, secure, and effectively utilize their data?
The most important is data governance, which is outlined in the three [preceding] goals that can and should apply to customers, regardless of where they live, and organizations irrespective of their locations, not just in Canada.
Organizations should understand the governance policies. In turn, their contact centers need to have a designated area or section and team that manages this issue. I would expect that this area would also use outside experts to ensure compliance across the three outlined goals.
On effective data usage, the goal should be twofold:
- Become familiar with a given CRM tool that best manages the optimization of customer profitability. This would also entail some knowledge about the various approaches and processes that are used to maximize this customer journey towards profitability.
- Become familiar with GenAI, (i.e., ChatGPT or something more customized) that can learn from previous conversations to provide the best script going forward.
Both points 1 and 2 need to be integrated. The days of a reactive contact center rep are over. The new data paradigm requires an individual who is proactive in using data to best serve the customer while making money for the given organization.