Speech Technology: Automatic Speech Recognition

Introduction to speech technologies.

Contact centers have been using speech technologies for several years now. Significant breakthroughs have been made in areas such as Speech Recognition, Speech Synthesis, and Voice Verification. These options have resulted in more robust IVR designs and intuitive voice user interfaces that are easier to navigate through. These technologies will be presented in more detail in a 3-post series.

Automatic Speech Recognition basics.

Automatic Speech Recognition (ASR) is a technology that allows a computer to interpret human speech and convert it to text. Speech Recognition should not be confused with Speaker Recognition (which refers to recognizing who is the speaker as opposed to what the speaker said).

The ultimate goal of speech recognition is to be able to convert speech to text with 100% accuracy, regardless of who are the speakers, their accent fluctuations and the language being used. The environmental conditions should not be a factor either (this relates mostly to noise and distortion issues). Despite significant breakthroughs in automatic speech recognition during the past few years, we are still quite far from the ideal situation. However, when imposing some constraints on the recognition process, accuracy can be high enough to make its application feasible in many situations.

How speech recognition engines work.

Early attempts to implement speech recognition were based on template matching. The words to be recognized would be recorded, and a waveform would be produced for each of them. Then the uttered words would be converted to similar waveforms and compared to the archived ones to produce a match. This method was rather inefficient in terms of both required storage and processing power needed to perform the matching (especially as the size of the available vocabulary increased). Template matching was also heavily speaker – dependent and continuous speech recognition was also impossible.

Newer engines are based on phoneme recognition. The utterances are split into small pieces (usually several pieces per second) and the respective waveforms are processed for each piece. The result is a set of frequency bands for each piece, which is then matched to a phoneme. Combining these phonemes, the engine tries to reconstruct the uttered words/phrases. The advantage of this approach is the fact that the total amount of phonemes a human can pronounce is extremely small (compared to the infinite amounts of words and phrases). Phonemes are also common for everyone, despite accent flavors and even languages (though different semantic rules apply on each language – the basic sounds are the same). The matching algorithm is also extremely faster than in the template method.

Phoneme recognition is still far from being perfect. The accent flavor of each human is only one of the issues that affect its success rate. Another important problem is ambiguity. Consider for example someone saying “forty two”. He could mean “42” or “40 2”. Or compare “eye” with “I”. Ambiguities like the one in the first example, concerning numbers, can be solved by introducing rules (constraints), like, for example, that numbers should be spoken one digit at a time. However this constraint causes problems when the IVR asks you to speak your 12-digit customer number. The second type of ambiguities can be addressed by taking into account the context of a phrase. All these inherent problems (and many others that go beyond the scope of this text) limit the accuracy of speech engines. However, using several constraints, accuracy can still be well above 90% in many applications.

The grammar.

A key concept in phoneme-based ASR engines is the grammar. A grammar is a set of rules written in a specific format (often XML-style) that contains all allowed utterances for a specific part of an application. Grammars contain information such as keywords that match to a specific meaning (within a specific context that applies in the part of the application it refers to) and other rules that help solve the aforementioned ambiguities.

The recognition process.

The recognition process involves several steps. When the user speaks to an ASR application, the first step is to define when the utterance starts and stops. This is done to avoid overburdening the engine with analyzing silence and/or noise. The utterance is then analyzed and matched to all allowed phrases in the specific grammar applying to each part of an application. Typically, a result is returned along with a recognition probability. As long as the recognition probability is higher than a predefined threshold, the recognition is considered a success. 

Call recording: when compliance becomes an opportunity

Everyone that uses call centers of any kind of company often runs into a standard prompt sounding like: “This call is recorded for your own protection” or something similar. Most often than not, call recording appears as a result of legislation that companies have to comply with. It started with “sensitive” industry sectors such as banking and insurance, but it increasingly expands to include various other types of contact centers also.

A lot of companies, as a result, tend to view call recording as an obligation, and a costly one too. Compliance with regulations incurs costs related to both the technological products needed as well as the human resources required. These regulations can also come from different sources also (state, organizations etc). And since the costs of non-compliance tend to be far higher, almost all companies opt to adhere to the regulatory frameworks and record all calls.

However, all this recorded material is usually dumped into oblivion, as soon as the regulations allow. By doing this, many companies waste a very good opportunity to extract important information from such data. While small companies with few customers have a level of intimacy and direct contact that makes getting feedback quite easy, large corporations cannot afford such luxury. The vast majority of interactions is funneled through the contact center and recorded on appropriate loggers. These logged calls can (and should) be analyzed to provide information for every customer interaction. Information that can be gathered this way includes agent performance, customer satisfaction and possible preferences, quality of products and services and even free material for market analysis/research. All these can be analyzed and used to provide competitive advantage.

This type of information can be gathered quite effectively with the usage of speech analytics (or audio mining) suites. Speech analytics is a rather complex solution that combines a variety of information such as who the speakers are, what keywords they used, and speakers’ communication historical data and so on. There are several techniques that are widely used for recognition and subsequent classification of data, with various degrees of accuracy and processing requirements.

These solutions are usually offered as complementary products by vendors that sell voice and speech related hardware and software. An example of available speech analytics product is Nuance Care Analytics. Recently NICE systems, a leading vendor of call logger systems among other things, announced a real-time speech analytics product.  

Unified Self Service

A large amount of contact center inbound traffic is related to information queries or simple interactions that involve few steps. This kind of traffic can be often served by using self-service solutions, automated procedures that do not involve human intervention from the part of the company. Self service is an extremely efficient addition to contact centers, since the cost of an automated interaction can be around five times cheaper than using live agents.

Self-service technology can be implemented on a variety of contact center channels. Web-based self-service is the most common and usually easiest to implement and use. A visual web interface is very robust and intuitive, and large amounts of possible interactions can be offered in a small amount of pages. Forms are easy to use and the http request-reply model fits extremely well to the self service model. The only problem web interfaces have is that they require a computer and internet connection, which is not always available or convenient to have.

Therefore, in the context of enhancing contact centers with additional channels, many companies opt to integrate self-service options for additional communication methods which are more readily available on the move. From all non web-based self service methods, the most commonly used is telephone self-service, powered by IVR systems. A Voice User Interface (VUI) is typically developed to guide the users through layered menus, until they retrieve the information they need. Traditional VUI using DTMF inputs only is rather inflexible for large scale systems and menus. The introduction of speech-enabled IVRs that came along with the significant breakthroughs in speech recognition and text-to-speech technologies has alleviated this issue a lot (the quality of speech technologies is still heavily dependent on the language itself – a lot more money and effort is naturally invested in English language than, for example, in Greek). The users can now speak a specific keyword and be instantly directed to the appropriate submenu, skipping all the intermediate steps and getting the information they need quickly. Well designed VUIs can be very intuitive to use and it is probably the best option to use when web interfaces are not available. An option to transfer to a live agent also helps a lot, in case the users encounter problems with the automated systems.

SMS is another convenient channel for self-service that is widely used. It is especially convenient for 1-step queries. The user can send a text message to a number, using predefined formatting to pass through some information about his query. The query is then processed on the server side and the results are returned back to the user via SMS.

The different communication channels can be integrated above a common database, to provide information independent of the channel being used.

Contact Center Workforce Management

What is Workforce management?

Contact center running costs heavily depend on the human resources required to operate the center. Therefore, companies are trying consistently to reduce these costs by increasing the efficiency of the centers and maximizing productivity out of a certain number of employees. For large contact centers with a lot of employees, this kind of management cannot be performed manually without missing out a lot. Appropriate software has been developed to this end, called human resource workforce management software (WFM).

Traditional workforce management in a contact center has been performed by using spreadsheets. However, especially for larger numbers of employees, spreadsheets are inefficient and time consuming. The WFM software that has been developed today can automatically perform complex tasks such as scheduling, satisfying staffing requirements, forecasting, managing real-time changes and a lot more. These tasks are essential for large contact center management, to ensure smooth operations and maximize staff productivity.

Workforce management functionality: forecasting, scheduling and more.

Forecasting is probably the most important benefit of using workforce management software, and the key reason that justifies buying such software over using a spreadsheet. A good schedule can substantially reduce costs by using the correct amount of staff at any given time. This can be achieved by analyzing historical data, to predict traffic spikes and low traffic periods. The forecasted volumes are then converted to actual number of agents needed to serve the traffic at any given time. Better seat management can be achieved and part-time agents can be hired to meet potentially increased demands while keeping costs low. In the case that part of the contact center is outsourced, WFM software can also help with creating the schedules for these resources also.

Scheduling is also important from agent point of view. Historical data help in making fair decisions about vacation planning (business rules, history of schedules, shift bids etc). Manual interventions and adjustments to the automatically produced schedule can help satisfy specific agent needs and preferences in many occasions. Correct utilization of workforce management software can thus reduce agent over utilization and eventual burnout (experience has shown that good workforce management systems can satisfy up to 70-80% of agent preferences).

Workforce management software includes other helpful features such as reporting tools. It gathers a lot of real-time information about operations and provides useful reports. Based on these reports, micromanagement of agent performance can be also heavily improved. Workforce management software monitors their adherence to schedule, arranges breaks in a way that they don’t overlap etc.  Reporting also helps in quick reactions when something needs to change urgently. For example, in a customer support contact center, when a major service outage occurs, the software can predict how many extra agents have to be called in to serve the traffic spike (based on real-time data as well as historical data of similar situations).

Workforce optimization: the next step after workforce management.

Workforce optimization (WFO) is an evolution of the concept of workforce management. It is a unified software suite of WFM software along with tools for quality management, recording, coaching, performance management, speech analytics etc. All these tools combined can be used to improved overall contact center performance, by providing information that the center supervisors can act upon quickly.

The WFM/WFO industry still has a lot of growth potential, since the majority of contact centers currently still use simple spreadsheets to manage their staff. A market research report on the WFM industry by DMG consulting shows 30% growth for 2009 which is the biggest growth for the particular industry ever. The offering of WFM/WFO suites on a Software as a Service (SaaS) model as well as the increased number of vendors offering such suites could potentially make WFM/WFO components affordable even for smaller call centers.

Unified Communications design failures

Unified communications tends to become the long-term goal of a large amount of enterprises. Correctly implemented UC can significantly contribute to business process efficiency in various ways. Intra-company communication becomes more robust and effective, while also customer interactions across multiple channels offer convenience and increased transaction speed.

However, in most unified communications systems deployed currently, the experience is far from the ideal. Companies trying to implement UC often fail to understand the requirements of deploying such solutions. Buying and deploying some equipment that smoothly integrates communication channels is important, but it is not enough.

The most common problem that UC deployments fail to address is the customer representative training required to operate across different channels. A very illustrative example of unified communications fail is this chat discussion. The agent serving the customer in this example shows pretty typical behavior observed in carrier (and other sector) contact centers: he uses scripts that don’t always fit the discussion, he is not well-trained in his field (which becomes more obvious because the customer in this case is rather knowledgeable about the nature of his problem) and he is also quite inflexible. The fact that he is consistently urging the customer during the chat session to use phone support to continue troubleshooting is a typical example of horrible UC implementation. The channels in this example give the impression of being an extra tool the company uses to manipulate the customers rather than serving them.

More hilarious situations seem to appear lately, as some companies attempt to use social media to interact with their customers. I heard another story a couple of days ago from a friend, trying to contact a company by twitter. He expressed his problem and he was promptly asked to contact via the call center where the waiting time was more than 30 minutes. In both these examples, the company failed in the selection of appropriate human resources to support the new channel and/or usage of appropriate processes for the specific channel. The agents assigned to the alternative channels should be both well-trained in the company products, services and business processes, as well as in using the particular channel. In the ideal scenario, some of the most skilled agents of the contact center should be able to use more than one channel, providing a free load balancing mechanism among the UC channels in the process.

Unified communications success depends largely on intra-company culture. Every single member of the company (not only the agents) should be trained to appreciate how UC systems work and how to make the most out of them. UC unavoidably create a new environment, and everyone should adapt and finds their place in the new ecosystem. Teamwork is also necessary for success. There are many UC implementations that simply fail because the company employees cannot use them as they should.

Another common problem that contributes to unpleasant alternative communication methods is bad understanding of the technological characteristics and requirements of each channel. Most often than not, this leads to bad interface designs. There are countless examples of this issue, from “multi-lingual” websites that provide completely different information on the same issue depending on the language to “spaghetti” self-service interfaces. This is tightly related to the end customers themselves. The delivered system should be simple and intuitive to use. If one of the channels is far more efficient than the rest, users/customers will end up over utilizing it and ignoring the rest.

A lot of the UC failure deployments can be also attributed to vendors. Many vendors attempt to sell unified communications as a technological product, while it should rather be a solution. Unified communications is not about building a couple of gateways to integrate the communication channels. It is a far broader concept that relates to company culture, business processes, human resource behavior, technological products, strategy and a lot more. This is documented in more detail in this blog post.

The conclusion is that, if a company wishes to incorporate new communication channels, it should first get ready to actually use them in a constructive way. And this is not an easy task. Otherwise, it is better to just stick to the phone. It will be better for both the customers and the company.

Aspect announces the release of Unified IP 7

Aspect announced on Tuesday 11/11/2011 the imminent release of its next generation Unified Communications platform, Unified IP 7. It is a next generation customer contact platform based on a unified, multi-channel product architecture that allows an organization to:

• Transition to a next-generation platform while preserving key best practices and processes.

• Improve customer experience while containing costs.

• Leverage unified communications infrastructure and collaboration software to enhance customer communications and increase one and done interactions by eliminating communications silos in contact center and across enterprise

• Deliver a rich, consistent experience across a complete set of contact channels, including short messaging systems (SMS) and instant message (IM).

• Support sophisticated processes across multiple channels.

• Virtualize contact center operations with centralized administration and routing.

• Tightly integrate with Microsoft Lync to improve productivity within the contact center and to connect the contact center to the rest of the enterprise.

• Migrate its environment in an evolutionary way to the next-generation unified contact center.

• Improve reliability, efficiency and usability

• Reduce the total cost of ownership

Read the full press release about the product.

Contact Center Outsourcing basics

Contact center is the interface between a company and its customers. In the consumer-centric era we live, customer support is a decisive factor affecting customers’ opinion of a company and subsequently loyalty. Contact centers consist of two major facets, the human resources and the automated systems. Customer demand for increasingly more options and features increases the pressure to deliver better support, and as a result the costs of operating and maintaining a contact center.

In this context, many companies decide to outsource a part of their contact center or even all of it. Full outsourcing includes everything, from technological infrastructure up to the human resources needed to operate the center. A more frequently-used model is partial outsourcing, frequently also called hosted model or SaaS (Software as a Service). In this model, the company retains the human resources that operate the contact center and parts of the software, and uses a service provider for the infrastructure. The infrastructure facilities are then usually accessed via web interfaces. There are also some blended models where contact centers are split between in-house and hosted depending on their functions (customer service for example can be kept in-house, while telemarketing can be fully outsourced).

The choice of whether to outsource the contact center depends most often than not on cost factors. Partial or full outsourcing offers significantly reduced costs, both upfront and operational. In the outsourced models, the pay-as-you-go method is used for billing. Cost depends on capacity and functionalities being leased. There is no need for risky large investments. Operational costs are also smaller (the cost of leasing is smaller than the cost of support contracts and in-house specialized personnel to manage the systems). The cost of upgrading and keeping up with technology advances is also eliminated or severely reduced. Finally, some parts of the contact center, like IVR, tend to have inherent cost benefits in outsourced environments, due to their nature. 

Other than cost factors, there are also some other issues to consider, related with technology. Insufficient knowledge of relevant technology is one of these issues. It is very frequent for a company to buy large in-house systems investing large amounts of money, without actually using them fully afterwards. Outsourcing eliminates the risk of becoming victims to cross-sales and up-sales efforts of vendors.  Launching time of services is another issue that leads companies to outsourcing. In-house deployments are time consuming and bug prone initially. Using outsourced services ensures shorter launch times.  

On the other hand, there are situations where outsourcing is not the best strategic decision. For customer service representatives especially, outsourcing yields many risks, especially when the outsourcer resides on a different country. Insufficient knowledge of the industry and the products of the company, as well as different accent and non-native language might prove unpleasant for the customer experience. Therefore a lot of the companies prefer to maintain in-house agents for at least the more critical parts of the contact center, such as customer support.

Outsourcing can prove cost inefficient also in some situations. If a company already has skilled engineers that can deploy and support the infrastructure, while at the same time they can decide exactly what they need to buy, it can be cheaper to buy their own equipment. This is usually true for large companies, where size of the contact center justifies spending of resources to employ and train their own human resources to operate it. The additional benefit these companies have is that they have complete control over the systems, including security issues (which in some cases, like bank contact centers that manipulate financial data, are critical).

The final decision utterly depends on the special characteristics of each individual contact center. Cost factors should be balanced with customer satisfaction to end up with the best solution. In general, outsourcing should be a good choice for most startups and small businesses that require few resources combined with a large variety of functionalities. Large companies which depend on customer service quality are usually better off with in-house solutions.

Unified Communications basics

The term Unified communications (UC) refers to a set of products designed to integrate various communication channels, both real-time and non-real-time, under a unified interface.  UC allows a message to be sent via one communication channel and then be retransmitted over other communication channels. The purpose of unified communications is to optimize multi-channel communication, reduce processing overheads and nullify dependencies.

Unified communications came into being as a natural consequence of the deployment of IP telephony. With telephone data being transmitted over IP (same as other types of communication like e-mail or instant messaging) it gradually became apparent that it would be very efficient to unify all communication methods under a single banner.

UC integrate a large variety of systems and technologies such as telephony, multimedia communication, voice mail, conferencing, instant messaging, e-mail, fax, collaboration tools and so on. All these systems are connected through various gateways to a single IP network. For example, an email sent to another person could be transmitted through the IP network, arrive at a gateway that converts it to a fax, and then be sent as a fax to the recipient. Various protocols have been developed to manage this communication, with the predominant one being SIP (Session Initiation Protocol). SIP has the advantage of being independent of the communication method used; it initiates, manages and terminates sessions, leaving the users to negotiate the mode of communication in the process.

In a company using UC, every employee can be connected to the network. Depending on factors such as location and time of day, each employee can be connected via one or more communication channels. For example, when he is at the office in the morning, he can be reached via his office landline, e-mail or fax, while when he is at home in the evening he can only be reached via his mobile telephone. To manage availability of various communication modes, the concept of presence was developed. Presence information includes a person’s location and availability of communication channels. Information of whether they can be contacted in real-time or they are currently busy (thus an asynchronous communication method such as e-mail should be preferred) is also a part of presence. This information is what makes unified communications stand out, by enabling very efficient business processes to be deployed. Presence servers are usually tightly packaged with SIP servers.

UC have become an integral part of today’s contact centers. They naturally fit in the multi-modal communication modes employed to interface with the company’s customers. UC in a contact center are most useful for internal communication of employees and associates; for example, when a customer service representative talking on the phone wishes to get some help from an expert on an issue, he may check presence information to see who is available and get the information via instant messaging, while keeping the customer away from unnecessary transfers. 

CTI basics

CTI, standing for Computer Telephony Integration, is a technology that allows data integration between telephony systems and computer/network systems. CTI is an important part of contact centers, providing enhanced capabilities and functions to the PBX. Through CTI, a computer user can control his attached telephone system. The telephone system can also send information to be displayed on the computer.

In a CTI powered contact center, the CTI subsystem usually assumes centralized control over everything else. When calls enter the system, information such as called number and dialed number are used to decide how to handle the called, based on predefined routing strategies. For example VIP customers may be recognized by their caller number and directly forwarded to a preferred representative. Or, assuming the company has two different inbound numbers, one for technical support and another for general issues, the callers will be directed to the appropriate representative group, depending on the number they called.

CTI can have access to various databases whereupon it will extract data required to enhance its business logic further. A typical scenario that is widely used in inbound contact centers is the following: The call enters the CTI system, and it is then diverted to an IVR application that performs user authentication (asking a couple of questions like name, or a PIN number). After the caller is successfully identified, the call is returned to CTI along with identification token. The token can then be used to extract information from a database, and the decision on what to do with the call is taken afterwards. In this scenario, the routing possibilities are virtually unlimited, since the entire CRM information about a particular customer can be used for the decisions. Parts of this information can also be forwarded to the customer representative that is going to receive the call, as a screen popup, giving him all the details he needs to better serve the caller. This architecture example is depicted in the schematic below:

Another system that CTI usually handles is a call logger, a disk-based system where calls are recorded.

Due to its centralized position in the contact center, CTI is usually a good point to gather statistics. Since it interfaces with almost every other system in the center, it has access to a lot of relevant information that can be gathered and analyzed.

Customer Service Representatives and Company Culture

The customer service representative (often also called “agent”) is clearly the cornerstone of a contact center. Well designed automatic systems always help of course, but they cannot really compete (at least not yet) with the flexibility a human representative can offer and the warmth of a friendly human voice.

There are several qualities that a customer service representative must possess in the new breed of tech-supported contact centers. The first and foremost is, of course, sufficient knowledge of the subject(s) the agent deals with. This can be facilitated by usage of software (CTI, reporting tools) that provides personalized information for each customer, but the representative also needs to have a very firm grasp over the products/services and business processes of the company. Sufficient knowledge of how to use the automated software tools is also required. Another important characteristic is the agent’s language skills (speaking and writing fluently). And of course multi-tasking capabilities that allow the representative to communicate with the customer while taking the necessary actions to serve him simultaneously. All of the aforementioned skills can be developed to some degree with sufficient training.

And then, the personality of the agent comes into play. All the knowledge and technical skills in the world are not going to help if the representative does not like what (s)he does. A positive, outgoing and people-oriented attitude is crucial. The best customer service representatives, the stars that we occasionally encounter and we are all very happy to do so, shine mostly because they want to shine. Like Mary, for example, from this very nice story. (which was the inspiration to write this post in the first place).

These crucial agent personality traits are also trainable to some extent. But the development and manifestation of these traits is also greatly influenced by the environment (s)he operates. Being a customer service representative is often a high pressure job that requires quick reflexes. Situations where queues increase quickly (when something has gone wrong big time) are common in most contact centers, and supervisors press for even quicker issue resolutions than the agent can actually handle. Other contact centers walk on the edge with representative staffing, to decrease costs. This can lead to constant agent over-utilization. I happened to work in such an environment for a couple of months and the experience is far from pleasant. Keeping one’s composure in such a situation is not easy.

In this fast-paced and often emotionally taxing environment, some key factor in sustaining a good morale among the contact center personnel are how the management treats its representatives, the relationships between people within the contact center, specific performance incentives documented in business processes known to the agents and so on. All these conditions collectively define the culture of the contact center itself. And this culture can be defined and influenced by the top management in many aspects.

Allowing the representative some breathing air in some situations may be as important as following the rules. With the workforce optimization software packages that every large contact center utilizes, sometimes being an agent feels like being a slave rowing in an ancient Roman Galley, with the whip lurking above your shoulder.

In very large contact centers though, flexibility harder to achieve. In the outsourcing call center company I worked for as a customer service representative, there was a huge building with agent farms on every floor working for many different customer-companies. The supervisors were very strict and adhered to the “letter of the law” because their own managers were also enforcing rules in detail and so on. Individual effort and performance assessment were mostly based on the average call handling time. As long as the inbound queues were small, everyone seemed to be happy, regardless of the ridiculous problems many of our customers had.

Not long after I quit that job (which I would have never really fit into anyway) I had the chance of visiting, as an engineer, a couple of small contact centers with less than twenty agents. The difference in the atmosphere was breathtaking. The agents in the small contact center seemed more like a family. Other large contact centers I visit are more often than not like the one I was working on.

Of course the size of the contact center is not the only factor that affects the culture, but it definitely is an important one. Another equally important contributor to a good atmosphere is a sufficient number of agents to handle the average traffic without burning out. Cutting down on human resources budget too much can have a devastating effect to the morale of the representatives.

The company culture and its effect on how agents behave can make the difference between a successful and a failed contact center. At the end of the day, the agents are humans, not machines. The way their are treated by their superiors is reflected to their behavior towards of the customers. And it is surprising how often we all forget that, on either side of the phone.

Note: The above opinions and beliefs are based on the current situation in Greek contact centers which I am familiar with, but most of these should apply in contact centers around the world. If you are experiencing things differently, please be so kind as to leave a comment.

Outbound contact center basics

Outbound contact centers and their goals.

Outbound contact centers are focused on performing campaigns. Communication is initiated from somewhere within the company to deliver a message to a current or potential customer. The vastly preferred method for outbound contact is the telephone call, due, in part, to cultural reasons. Alternative outbound contact attempts such as email or instant messaging are very likely to be considered spam and quickly rejected. Therefore the outbound campaigns are usually performed by call centers exclusively.

The goals of an outbound campaign can be varied. Typical campaigns focus on sales, lead generation, personalized services, appointment setting, surveys and market research.

Sales campaigns typically focus on promoting the company’s products and services. Often they target new customers, trying to stimulate their interest in the features of whatever they want to sell. This practice is usually called lead generation, and is a marketing activity. Other than going for new customers through this method, companies frequently attempt cross selling (suggesting complementary products to a customer that has already decided to buy something) and upselling (persuading the customers to buy a complementary product or an upgrade that they did not originally wish to purchase). These techniques have proved very effective in practice, drastically increasing revenues.

Personalized services are another marketing tool that can be efficiently implemented via outbound call centers. Companies typically reach out to their most important customers making them new offers or even setting appointments if needed. Important customers are not only the ones generating most revenue, but often dissatisfied customers that have had problems in the past and have expressed some kind of complaints (though many companies tend to ignore that segment of their customer base, eventually losing them out to competitors).

Outbound call centers are also used for conducting surveys and market research. The results of these campaigns can measure public opinion on products and services and record other types of relevant data about customers. This information is critical for creating new business plans and designing new products and services, as well as improving existing ones.

There is also a whole breed of outbound call centers that handle collection campaigns from people that owe money to a company. These call centers have rapidly increased in terms of both amount and activity since the outbreak of the global crisis in 2008, as companies are starving for cash.

The heart of outbound contact centers: the predictive dialer.

Outbound call centers are powered up by a computerized system that automatically initiates outbound calls, the dialer. Originally, an auto dialer was used, which simply initiated calls for idle agents. Today, most call centers performing outbound campaigns use a predictive dialer. The predictive dialer uses a variety of algorithms, taking into account historical and real-time data from reporting solutions, to predict availability of resources and aggressively push for maximum efficiency.

The typical predictive dialer supports/performs the following functions: it monitors how the calls it makes are answered, discards unanswered calls, busy signals, fax machines and other automated services and only connects calls that are answered by people. The predictive algorithms try to match the expected number of calls answered by people with the number of available agents at any point. The calls attempted by the predictive dialer are thus a lot more than the available agents. This automation saves significant agent time and results in substantially more productive agents (agent talk time can increase by more than 100%).

Aggressive usage of the predictive dialer (the intensity of outbound calls performed compared to available agents is usually controlled by real-time configurable parameters) can lead to call abandonment. This happens when the predictive algorithm fails to match correctly the successful number of calls it attempts to the available agents, resulting in silence from the moment the customer picks up the phone until (s)he is connected to a representative. This is extremely annoying to the customer (some call centers address this by initiating the call using an IVR-automated message in case an agent is not readily available) and in many countries it is addressed by legislation (allowed abandonment rates are usually 3% or less).

About well designed contact centers

Leaving the technology basics aside for a moment, let's take a look at a few key things a company must keep in mind when designing its contact center.

There are two major components that define a successful contact center. The first is company-centric and it includes efficient business processes and operations. The second is customer-centric and it revolves around customer satisfaction. These two components are tightly interconnected, to the point that customer satisfaction cannot be really achieved without an efficient contact center. And vice-versa, extremely low customer satisfaction can often result in contact center over-burdening, rendering it unable to accomplish its purpose.

Efficiency is the keyword that one has to keep in mind when designing a contact center. It applies to virtually everything: business processes, cost structures, employee performance, technological components and so on.

In this context, the contact center should be able to generate direct revenue through successful sales (of both new products and upgrades), customer retention, collections or whatever else the enterprise deals with. It should also be the foundation of long-term revenue generation through sustaining and expanding the company's customer base (which tightly relates to the customer satisfaction issue that we will discuss about later).

Another important strategic consideration is to keep operations simple and cost-effective. Efficient business processes are critical, since they avoid unnecessary overheads. A correct mix of well-designed automated systems and employees contributes to cost effectiveness. The technological component of the contact center has to be complex enough to automate most routine tasks, always within the company's budget. A frequent reason of failure when designing and implementing contact centers is overextending in terms of costs, with the company ending up with incomplete or severely flawed automations (this topic is rich enough to require a separate post to explain in some detail).

The last thing to discuss regarding efficiency is how to measure it. Every contact center should have dedicated resources to collect statistics about its operations and consolidate them into relevant information. Parameters such as cost per customer interaction, revenue per customer interaction and employee performance ratings are very important in order to improve the contact center. All major contact center equipment and software vendors offer such solutions in many flavors, suitable for every situation and pocket. These solutions typically collect and report both historical and real-time data, facilitating employee performance.

There are various other considerations that one has to take into account to design and improve a contact center. However, the most important thing is to be focused on customer satisfaction. A lot of companies today have contact centers focused on dodging complaining customers.

Keeping customers satisfied has become a full scientific field in itself. In the consumer-centric society, customers become increasingly more demanding and are quick to switch to a competitor if they are unhappy with the customer support. A well designed contact center can be a strong source of competitive advantage in this regard.

One of the key factors that the customers usually require is contact center availability. This includes sufficient operation hours depending on the service (for some services, office hours support is enough while for others contact centers should work 24hours/day, 7days/week). Apart from the operating schedule however, what is usually more important is resource availability to actually serve the customer. Personally, I am extremely displeased when I need to wait several minutes listening to music while waiting in the phone queue to ask a simple question, or when I send an email inquiring about the status of a delayed purchase order and I don't get a reply ever. This problem can be usually alleviated by a correct utilization of self-service technologies like a well-designed website with an intuitive interface or a well-designed IVR.

Another very important issue is first contact resolution (FCR). This term means that when a customer reaches a contact center agent, (s)he should be able to answer the question or solve the problem without any call transfers, unless some very special conditions apply. FCR tremendously increases the efficiency of the contact center (saving time and effort) and also usually results in very happy customers. To increase FCR, the contact center must employ a combination of well-designed interconnected systems (typically glued together using Computer Telephony Integration technologies) and well-trained and skilled human resources.

Again, there are several other issues to consider regarding customer satisfaction but the two aforementioned issues (availability and quick issue resolution) are probably the most vital.

PBX: The core of the Call Center

What is the PBX?

A PBX (standing for Private Branch Exchange) is a telephony system responsible for connecting calls among a group of telephones and also for providing connectivity between this group of telephones and the public switched telephone network (PSTN). In the case of the call center, the group of interconnected telephones includes agents and supervisors, as well as other company employees. The end points within the enterprise that are connected to the PBX are usually called extensions.

The PBX was initially introduced to reduce the amount of circuits required to connect everyone in a company to the public network. It also eliminates costs of internal calls, since these are handled without involving the telephony providers. Another benefit of using PBXs is that it provides extra functionalities in addition to those provided by the PSTN.

The PBX entity is usually a computer running software that manages internal calls and calls to the PSTN. A number of telephone lines that connect the company with the PSTN terminate at the PBX. Within the PBX there is a network of actual lines and also a console is usually provided for managing the PBX.

PBX functionalities.

The main functionalities of the PBX are call establishment between telephone sets of two users, maintenance of the connection while communication lasts, and disconnection when the users wish to terminate the call. In addition to these key functionalities, there is a large amount of additional enhancements that depend on the manufacturer. Some well-known examples of such additional capabilities are automatic call distribution (ACD) based on a set of predefined rules, call transfer to other extensions, call waiting, conference call (a call that involves 3 or more parties simultaneously), voice mail and so on.

Types of PBX: circuit and packet switched.

Traditional PBXs are circuit-switched. When a call is established in a circuit switched network, a dedicated communications channel is created for the duration of the call. Each circuit cannot be used for other calls until the previous call is terminated and the circuit is released. A new connection can be then set up.

A newer generation of PBXs came along with the Internet. These are the IP PBXs, which use packet-switching. In packet switched networks, voice is converted to data and then transmitted over data networks, using the IP protocol. The IP PBX is responsible for performing this conversion from voice to data and vice-versa, as well as connecting to the PSTN. The conversions are made using a group of standards known as VoIP (Voice over Internet Protocol). The IP PBXs led to the development of CTI (Computer Telephony Integration) Technologies.

Week in Review: 03 Jan 2011 - 09 Jan 2011

Avaya announces acquisition of Konftel, 'a cutting-edge innovator in audio collaboration technology'.  Read the press release.

Interesting article discussing the benefits of hosted IVR and the strategic drivers behind decisions to follow that path. 

Insightful Q&A session with Broadsoft on TMCnet about Unified Communications and how they affect contact centers. Read the full article.

Short history of VoIP and how we came about to an era that paying for phone calls is close to becoming old news. Read the full article.

Excellent post discussing the benefits of proactive customer service. Read the full article.

IVR and how it fits in the contact center

Definition of IVR.

The term IVR (Interactive Voice Response) is used to describe a computer system which interacts with people using a combination of telephone keypad inputs (DTMF) and voice inputs. The system uses either prerecorded or dynamically generated prompts to communicate with the user. The interface created using IVR systems is often called a VUI (Voice User Interface).

IVR flavors and Speech - enabled IVR.

Traditional IVR supported user input only by DTMF. Thus, the user would be presented with some options through a prerecorded prompt, and would select one of them by pushing the appropriate key in his phone-pad. New technological breakthroughs in Speech Recognition have altered this to include speech as an input method. In the beginning, using predefined vocabularies was the only way to provide accurate inputs. Today natural language recognition technologies have been developed to allow for more intuitive inputs.

In a similar manner, the IVR system would traditionally use a combination of prerecorded prompts to "talk" to the user. Today this can also be accomplished by speech synthesis which is achieved through a technology called TTS (Text-to-Speech).

At this point it should be noted that the quality of the voice - oriented technologies such as Speech Recognition and TTS is heavily dependent upon the language used. For languages such as English, which apply to a very large segment of the market, voice technologies are extremely well refined, to the point that a synthesized speech segment may be indistinguishable to a prerecorded message. Naturally, for more "exotic" languages with smaller market significance, things develop somewhat slower though not very far behind.

The role of IVR in the contact center and the benefits.

IVR systems are typically used as the reception point in any inbound call center. Their typical role is to engage with the user and extract some information about what he wants to do. If the user/customer simply wants some information, then the system is capable of retrieving it from appropriate databases and proceed to inform the caller. IVR systems can also perform predefined procedures, without any human intervention. For this reason, they are often called as self-service systems. If what the user requests cannot be provided by the automatic system, s(he) is typically transferred to a human agent.

The benefits an IVR provides to the contact center are twofold. First, they are capable of handling large volumes of inbound calls 24 hours a day, 7 days a week. And second, in a large amount of cases some actions are performed without human intervention (for example when the user/customer requires simple information or wants to perform simple/repetitive tasks). Furthermore, using the initial information gathered within the IVR, the call can be routed to an appropriate agent with the skills to quickly server the caller (this is called caller segmentation). The information gathered within the system is typically propagated to the live agent using CTI (Computer Telephony Integration) technology, so there is no need for repetition. Thus, IVR contributes significantly to the reduction of the amount of costly human resources needed to operate the contact center.

With the introduction of the SIP protocol as the backbone of most contact centers today, the IVR functionality has been altered further to include complementary methods of communication in addition to DTMF and voice. Video and Instant Messaging are two of the most hot additions.As bandwidth availability increases and smartphones grow, a lot of industry experts expect video to become predominant in the self-service part of the contact center.

Contact Center Overview

Contact Center Overview

The schematic above displays the general architecture of a contact center.

Contact centers operate as either inbound (the customer initiates the communication with the enterprise) or outbound (the enterprise initiates the communication with the customer).

When a customer wishes to contact the enterprise, he is offered a selection of communication methods, such as telephone, web interface, video, e-mail, SMS or fax. Depending on the complexity of his current need and his personal preferences he may choose any of the offered methods.

As soon as the customer initiates the procedure, his request is channeled through a variety of systems, depending on his contact method choice. In most cases, the customer first encounters an automated interface, which is designed to gather some initial information about the customer and his request. Typical information that can be gathered in this stage are the identity of the customer and the general category of his request. Depending on the contact method again, this initial information gathering can be either a predefined procedure (for example, when using SMS, the customer should already be informed about specific keywords that indicate specific intentions/needs) or an interactive procedure (such as a phone call Q&A session with an automated IVR system).

After the initial information gathering procedure, the data acquired are forwarded to appropriate application servers within the enterprise, which contain the business logic that determines how the customer's request will be handled. The customer may then be provided with more questions/options and the procedure is repeated until the customer is satisfied.

A similar procedure is followed in the case of outbound communication. The only difference is that the communication is initiated by the company. The method of communication chosen in this case is determined by the enterprise business logic as well as preferences that the customer may have previously stated (which may be recorded in the CRM).

The schematic shows the most important systems typically encountered in a contact center. We will briefly explain  the functionality of each one of these, and how it fits in the procedures described above.

The Ethernet bus denotes the enterprise LAN. In IP contact centers (which are the foundation of unified communications) all the information is exchanged using IP communications. In case a communication channel uses different protocols (like SMS for example), appropriate gateways convert the traffic to IP based, as seen in the schematic.

The PBX (Private Automatic Branch Exchange) and the ACD (Automatic Call Distributor) are the systems responsible for controlling the telephony part of the contact center, and distributing the calls either to other automated systems and/or to appropriate persons that will serve the customer. These can be either classic telephony based systems or IP - based.

The IVR (Interactive Voice Response) is an automatic system that gathers information from the user (either using DTMF input or speech recognition) and also fills in the self-service part of the system.

The Outbound Dialer is an automated system that calls the customers of the company, in the case of outbound communication.

The Call Logger is an automated system responsible for recording calls.

The Application Servers is a cluster of various servers that process the data exchanged through the contact center. One of their key purposes, is to provide the business logic behind every automated action undertaken by the rest of the systems. Application servers can also be used for data consolidation, providing statistics about the contact center's operation and various other functions.

The CRM (Customer Relationship Management) system is a complex database - based system that stores information about the customers of the enterprise and their interactions with the enterprise.

The agents and the supervisors comprise the human part of the contact center. Their role is to serve the customer whenever the automated systems are not sufficient. They are responsible for handling phone calls, web chat, e-mail etc. The integration of the humans working at the contact center and the automated systems is typically achieved using CTI (Computer Telephony Integration) technology, which is responsible for retrieving customer data from the automatic systems and presenting it to the agents.



What is a contact center?

A contact center is an enterprise business unit which acts as an interface between a company and its customers. The term is an extension of the classic call center, which used to be the major channel through which customer interaction was performed. Thus, the traditional contact center used almost exclussively telephony. During the past few years, with the rapid technological advances in the field of unified communications, the contact center gradually encompassed alternative methods of customer interaction, such as e-mail, web chat/instant messaging, SMS and video. The inbound and outbound information that flows through the contact center is typically managed by a Customer Relationship Management (CRM) system.

A typical contact center consists of two major entities: a complex ecosystem of interconnected communication systems which automate most tasks, as well as a group of people responsible for managing and refining the customer interaction process when technological limitations apply. The general trend of the past years is an increase in the size of the automated ecosystem and a decrease in the number of people required to fill in the gaps.