Most companies are always on the lookout for the next ‘Game changer’ for their contact center. When you talk to experts, you’ll hear a lot of buzzwords these days, such as ‘Omnichannel’, ‘Net Promoter Score (NPS)’, ‘Chatbots’, and ‘Artificial Intelligence (AI)’. It can be challenging to sift through all of these terms to see what actually matters to you.
In this blog post, I want to demystify a few things about contact center artificial intelligence (AI) and share how AI is used to drive better results, to boost efficiency and improve customer satisfaction. I’ll focus on AI in workforce management technology but I will also talk about other ways AI is used in contact centers, because there are actually several interesting applications.
The term Artificial Intelligence was first introduced by John McCarthy during the 1956 Dartmouth Artificial Intelligence (AI) conference where leading researchers of the time discussed advanced research topics such as complexity theory, language simulation, neural nets, and learning machines.
Today, AI is seen as a a sub-field of computer science and is defined as “the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages” (Oxford Dictionary).
In fact, AI can be regarded as a substitute for human work, as it is based on the premise that machines imitate human intelligence. The objective of AI is to have processes in place that improve automatically without you having to intervene for each adjustment. AI technology generally uses machine learning to observe patterns and the impact of various inputs to continuously improve an output. As the inputs and the conditions change, the system adjusts itself.
For customer-interfacing, chatbot technologies are powered by AI to handle support requests. A chatbot allows customers to get answers immediately to relatively simple or common questions. The more the chatbot gets used and the more input it gets on whether the answers were helpful or not, the better it calibrates future responses. So continual input and feedback-driven continuous improvement without any manual intervention and human support from the contact center.
This of course changes the role of the customer service agent. As customers increasingly use chatbots to self-serve, simple and quick transactional calls no longer reach the contact center. Instead, the calls that go to your agents become more complex, and longer, and may focus more on managing relationships with your customers.
As you plan for handle times you will likely see an increase in the average handle time. However, if the technology is effective and your customers are getting their issues resolved without talking to someone, your total handled minutes will go down. When you implement AI or self-service technologies for customers, be sure to see if the number of inbound calls goes down. If you only measure the effectiveness by handling time, you’re not getting the whole picture. This is often missed by organizations as they roll out and integrate self-service options in their contact center in the quest to provide omnichannel customer service.
Workforce management systems can benefit from and leverage contact center artificial intelligence as well. SaaS and cloud-based WFM tools, like injixo, make use of AI technology such as machine learning to remove a lot of the manual and repetitive work from forecasting and scheduling while improving accuracy and increasing the overall effectiveness of the system. This can provide significant efficiency gains and cost reductions through more efficient operations. Moreover, service level goals are met more consistently because your staffing better aligns with the demand at the interval level. Less human intervention is necessary and you need fewer people to maintain the models. That doesn't mean that the job of the forecaster is redundant. The time freed up by automation can be used to enable planning team members to focus their time on more advanced, analytical work to add value to both the company and its customers.
Several WFM tools use AI in their forecasting and scheduling modules. injixo is an industry leader in cloud-based WFM technology and pioneered the use of AI forecasting. So I asked them to help explain how AI is used in their workforce management system. This is the answer I got:
"injixo Forecast consists of two fundamental models. The first one focuses on the time series of daily totals, while the second one drills down into intraday volumes. The daily totals model consists of two components: The first one detects seasonal variations, weekly, monthly, and yearly patterns, together with long-term trends such as growth or decline. The second component models short-term interactions within the time series (e.g. the impact of recent days on the upcoming days). The same approach is used for volume, average handling time (AHT), and other properties. injixo Forecast produces an interval-level forecast for 365 days into the future. And it works perfectly for all channels: email, chat, back office, and social media, as well as calls.
In our curve-fitting algorithms, we use techniques from machine learning, such as regularization. This prevents the model from fitting the historical data too closely, which is known as 'overfitting'. That results in inferior accuracy on data that the model has not seen before.
We use self-learning algorithms (machine learning) that are specifically tailored to the contact center domain. Over 20 years of experience in the call center field provided us with rich insights that helped us to deploy contact center artificial intelligence that generates extremely accurate forecasts for our customers. The AI evaluates a pool of different models and selects the one that works best for each individual workload (set of queues), for each individual customer. It is the opposite of 'one-size-fits-all' and ensures that every customer gets the most accurate possible forecast with their data in their center. We don't disclose the exact algorithms used by injixo Forecast - and in fact, we are constantly evaluating and deploying new techniques. I can tell you that we do use neural networks to guarantee customer-specific modeling and continuous improvement in order to minimize error margins.
injixo Forecast also lets users apply business intelligence such as holidays and events. The algorithm learns about the impact of previous events and applies this learning when forecasting future events.
An important benefit of injixo Forecast is that it is fully automatic. Minimal manual effort is needed - from ACD integration to generating the first forecast. Another nice feature is that customers can build new test scenarios on the fly with sample data (e.g. for an upcoming campaign, new product, or a new client), without touching the original data imported from the ACD.
As well as improving accuracy, injixo Forecast dramatically reduces the time and effort required for forecasting. Now, customers can generate forecasts of such quality that would previously have required a team of mathematics boffins, completely automatically.
We won't rest on our laurels. We will further improve injixo Forecast, potentially leveraging 'big data' from external sources such as weather, traffic, and event data, thereby exploiting this collective intelligence to make the forecast even faster, better, and more accurate.“
To learn more about forecasting in general - check out my recent 3 part blog post series on Contact Center Forecasting Fundamentals.
As you can see, leveraging the power of artificial intelligence in workforce management can be a huge asset for your business. I’ve used it successfully in a few different industries. However, if you also want to take advantage of it, you should consider a few things.
First, machine learning and AI adjusts based on the information that comes in. So if you have external events in your data that impacted workloads in the past, you need to consider if you exclude those events before an AI tool uses that data to create a forecast. This might be necessary if these are one-time events. For example, when I was in the travel industry, there was a hurricane in early September 3 years in a row. However, that doesn’t mean there will be one this year. But the technology sees the pattern and adjusts seasonality accordingly. Now, because we’ve had that incident the last 3 years when I’m doing long-term planning, it’s not a bad assumption to plan for it to happen again. But as you get closer to the date, you will be able to clearly see where there is any potential for a hurricane. If the weather is clear and expected to stay clear you need to manually override this.
You’ll see examples of this in other areas as well. If your business runs a marketing campaign or has an ad play in the media, this may drive an immediate response. This will impact your overall volume, your day-of-week distribution, and your interval arrival pattern. As your data comes in, normalize it by removing those events that you don’t expect to be recurring. After the base forecast has been created leveraging the power of AI, these events can be overlaid again if necessary.
As you implement workforce management technology with artificial intelligence, make note of all of the anomalies that can impact your forecast. And don’t be afraid to make manual adjustments. Over time, you’ll have the best of both worlds - technology that continues to learn from actuals, and the right level of human interaction to make sure you get the best possible forecast. You’ll continue to reduce the amount of time you spend engaging with the technology and reinvest that time into driving more value.
The worst thing you can do is to expect contact center artificial intelligence to take care of everything. Like any tool, it assists you and reduces the heavy lifting. But you want to always understand how it’s working and where it’s making improvements. The best thing you can do to maximize the quality of an AI-driven forecast is to ensure the data used for the forecast is clean. It’s like that old saying 'Garbage in, Garbage out'.
Here is a brief overview of some other use cases and application options of AI technology in the contact center landscape compiled by Call Centre Helper and a dedicated panel of experts. For more detailed explanation, check out their article here.
1. Replacement / complementation of the IVR (Interactive Voice Response) processes: AI makes use of more sophisticated mechanisms based on Natural Language Processing and Machine Learning techniques to better understand statements and to provide a broader set of choices that are more accurately tailored to the user or customer.
2. Enhanced analytics and customer intelligence based on chatbot interaction data: With AI technology incorporated in chatbots, more customer interaction data can be captured and analyzed accordingly to improve customer satisfaction and overall contact center processes.
3. Optimized self-service experience through virtual assistants: Virtual assistants provide a great contribution to self-service options in helping customers/users to navigate websites, and find the information they are looking for without involving human support.
4. Robotic process automation to benefit from big data across platforms & systems: Robotic process automation (RPA) attempts to consolidate, analyze and share data across channels and platforms to facilitate access and personalization of service delivery.
5. Predictive analytics in customer service: With AI-powered analytics, customer behavior can be identified and predicted at a faster pace which eventually leads to better decision-making in both planning and operations.
6. Cognitive systems accelerate automation and reduce human intervention: With the help of AI-powered technologies, the former rules-based system will become a cognitive system, allowing for automation and increased optimization potential while reducing the need for human intervention in activities such as forecasting and Skills-Based Routing.
7. Better and enhanced Self-Service capabilities: AI advances self-service capabilities in the everyday business of the contact center, thus freeing up time and effort for planners and agents to focus on more sophisticated tasks to create a better customer experience.
8. Customer interaction via robots: Robots such as chatbots empower customers by reducing complexity and allowing for quick self-service solutions that do not require human intervention - thereby also reducing costs for contact centers by cutting off repetitive, time-consuming tasks.
These are just a few of many application examples for contact center artificial intelligence technology and there is certainly more to come. Yet, many contact centers are still at the fledgling stage in adopting AI tech to increase operational efficiency and streamline customer service. It is an exciting time to watch out for new trends and observe recent technological developments to stay ahead of the competition and continue to provide superior customer experience.
Do your research on how Contact Center Artificial Intelligence can help your workforce management team. In addition to reaching out to workforce management technology vendors, you can also talk with consultants to discuss what options exist in the marketplace and what may best suit your contact center.