Updates from Mark Angel RSS
10:36:40 am on January 29, 2010 |
Predictive Analytics Part 2: Apply Models to Your Service Processes
To effectively optimize your decisions, you first need to define your customer service processes and ensure that your agents are following them. Next, you need to add a decision step in your service processes that allow the agent to offer a cross-sell. Then you must carefully measure the success of each cross-sell with individual customers and correlate their relative success with the data that you have gathered about these particular individuals. With this information, you’re better equipped to recognize new situations where a cross-sell will work and those where it won’t.
All this may sound difficult, but mathematical models do exist that make this data collection and analysis easy. They’re part of a scientific area of study known as symbolic regression. Symbolic regression allows you to build models based on lots of data and lots of variables whose correlation to the decision may be unknown initially. For more info about building models, check out Evolved Analytics.
10:30:21 am on January 8, 2010 |
Predictive Analytics Part 1: Optimize Your Decisions
How exactly do you go about optimizing your business decisions? Can you determine, for example, whether it’s more advantageous to cross-sell or upsell a customer during a service interaction?
The success of your decision depends on a variety of factors – the persona, or profile of the customer, the issue at hand and the mood of that customer at the time are a few of the critical factors. Since we live in a world of information overload, you very likely have mounds of information relating to these various factors stored somewhere in your service organization.
Look at the persona of your customer. A single trait of this persona can help you answer the question above, to effectively identify if a cross-sell or upsell offer will be accepted.
You must treat information like this as evidence and leverage it systematically in making your decisions. Essentially, your success means predicting the outcome based on historical trends. This is where predictive analytics is a key.
Predictive analytics is based on techniques for symbolic regression, which creates mathematical models used to synthesize even petabytes of data into intelligible answers. They help you better understand true variables driving success. These formulas work by comparing past examples of success and failure, and then using them to predict future outcomes. Models like these can be deployed and run by business analysts, not just statisticians with PhDs.
Such models have been successfully applied to a range of real-world scenarios:
- Adjusting fuel mixtures for optimal engine efficiencies
- Understanding the set of environmental and genetic attributes that correlate with breast cancer survival
- Adjusting oxygenation levels of premature babies to increase their rate of survival
They enable scientists and doctors to efficiently and effectively understand and capture the value from available information and use it as evidence to optimize outcomes. If predictive analytics is applied to fields like chemistry and medicine, then it can certainly be applied to customer service. In my next post, we’ll explore exactly how you can apply it effectively in your service environments.
02:04:09 am on November 24, 2009 |
Top 5 Problems with Service Today
We’ve all been the victims of bad service. We have to repeat ourselves to agents; we get different answers depending on whether we use email or we call a service rep; or worse, we don’t even receive a response. We have a tendency to blame the agents.
But it’s not their fault. They want to do a good job. They just don’t have the tools they need to meet customers’ expectations of personalized, consistent, accurate and fast service. They don’t have the tools because their technology isn’t working.
Thus, I trace bad service to 5 root causes:
– IT organizations have not solved the integration problem. Agents need dozens or often hundreds of un-integrated tools and applications to perform their jobs. Agents must toggle through many applications in the span of a service call, resulting in long hold times.
– IT organizations have not solved the change problem. Agent tools are typically hardwired together. When procedures change in a company, IT organizations cannot quickly respond to the changes in order to give the agents the new tools that they need.
– Knowledge management vendors have not solved the knowledge problem. Corporate knowledge exists on an island; it does not fit into the context in which agents are searching for content. This means that agents need to wade through many solutions in order to find the one that is right for a particular customer.
– Case management vendors have not solved the business process management problem. Today, business or call center leaders can’t drive agents through clear processes. This means that they put the responsibility of following the right resolution processes in the hands of agents, which are not all equally competent.
– Organizations have not solved the agent training and turnover problem. Businesses know how they want service delivered, but they can’t have their best, most highly trained agents handle every interaction.
How do we start solving these problems? We start with technology. We implement service-oriented architectures (SOAs) to begin integrating systems in a flexible way so that information is no longer siloed. This means that when processes change, IT systems can be rewired easily, without a tremendous amount of overhead. We ensure knowledge is it’s shared and applied appropriately. We present in a way that makes sense to agents, and we use it to inform our decisions about what’s working and what’s not working in our service processes. We provide agents with stepped procedures for each and every service interaction that allows them to provide the consistent service that we desire. If we are able to achieve these goals, and provide agents with the tools they need to succeed, then we’re solved our final dilemma. Because every agent becomes as good as our best, most highly trained agent.
10:25:08 am on October 22, 2009 |
The Ants Go Marching
If ants have to choose between two unequal length paths from a source of food back to their colony, they ultimately always choose the shortest, easiest one. Here’s how it works:
1. Ants run around the colony, more or less randomly, looking for food.
2. If an ant finds food, it returns back to the colony, and leaves on the ground a scent trail.
3. This scent trail attract nearby ants, which will follow this path and strengthen the scent of the trail, attracting more ants, strengthening the trail…
Ant colony optimization can be analogous to customer service. Think of how customer service agents answer customers’ questions. They each have their own style of interacting with the information at hand. Typically, they hunt and peck through disparate, unintegrated data, knowledge and back-end systems in a way that is unique to each of them. This often results in inconsistent, inefficient or, worse, incorrect service.
To avoid such consequences, companies must leverage technology to create process flows that guide agents through the same method of discovery. Think back to how ants find food – no single ant finds the best route back to the colony alone, yet a collection of ants are successful in doing so. And positive (stronger scent trail) and negative (weaker scent trail) feedback throughout this quest is used to guide this optimization.
Imagine if you had the ability to design not only a single customer service process that led agents to the correct answer to a customer question, but several variations of this process. Take, for example, varying the step in a service process in which the identity of a customer is verified, or when a specific knowledge article is presented to the agent.
Ideally, companies measure KPIs at each leg of this process. These steps are akin to the resistance that ants experience at each leg of their journey. By uncovering the strongest segments of each process, companies essentially rely on basic ant colony optimization techniques to determine the best customer service processes.