All these technologies converged in the 1990s. Several companies started to develop PC boards that could connect with the telephone network, using analog or time-division multiplexing (TDM) adaptors to talk with a switch. These boards communicated with other boards were packed with digital signal processors (DSP), programmable chips that could independently detect signals like DTMFs on the telephone line and provide the relative events to the software running on the host. At the time, I was working with Natural Microsystems, a Massachusetts company that was a pioneer in computer telephony.

Since this architecture was geared towards PCs, whose cost was falling while their power was raising fast (following Moore’s law), companies were able to set up contact centers at a reasonable cost. It was the dawn of industrial customer service: people started to be able to use their phone to try and resolve issues or set up services that previously needed an in-person visit to an office, or sometimes writing letters with no guarantee of a response.

However, while technology certainly enabled the birth of call centers, the humans work to make them function was also completely new. Agents had to be recruited and trained, a new organization geared to make answering calls as efficient as possible had to be developed. As with other sectors of industrial work, there was a need for specialization: agents could not be experts of everything and so a need arose to qualify calls for routing to the groups of agents in charge of specific topics.

The Automatic Call Distributor (ACD) was born, fronted by an Interactive Voice Response system (IVR). This was the beginning of the new century, the golden age for companies like Avaya and Genesys, innovators in the call center software field: every airline, every bank, every telephone carrier suddenly wanted a call center and business was booming.

(Call center is now an old term. Successive generations of marketing lingo have changed the name of the thing, first to contact center – with the addition of messaging and chat to the voice channel – then to customer experience (CX). We’ll change terminology as appropriate…)

So, companies wanted a call center, but to do what? This was a new capability and initially it was driven by technology (because we can) and by competition (because our competitor is doing it). Consumers were probably pleased but the ability to call a company was not seen by the public as a reason to buy its products or services more. Since calls were mostly to complain about a troubled offer, it was rather a double negative: mitigating problems, try not to lose customers. This does not bring in new money, so for companies, the call center was a cost, not an investment.

But they couldn’t go back: the genie was out of the bottle and the public was used to it. And so, in order to reduce costs, contact centers were operated on a shoestring budget, with the bare minimum of agents. At the same time, callers needed to be filtered: only the most motivated, and persistent, could be allowed in and IVRs were the perfect tool to provide this filter. And so, the menus got longer and more confusing, music on hold was invented (it could be seen as an advance, but for the low bandwidth of telephone lines that are geared to human voice and not music. This is why it sounds horrible.) and the wait and bad experience necessary to reach call center agents became a widespread meme.

Occasionally, there were attempt at improving the situation. In the late 2000s, texting and messaging app started to be used more and more as people switched from fixed phones and mobile “feature phones” to the first smartphones. Call center software vendors incorporated messaging into their product, which became “contact centers”. Also, the first limited voice recognition software started to appear and suddenly people were asked to “press 1 or say “sales”, press 2 or say “support” – not very user-friendly actually since the time it took to read a menu grew substantially. But something was also changing in the companies’ attitude towards their customer support function.

Slowly, customer support was seen as more of a competitive advantage. The first step was the recognition that, with products and services largely equivalent between different players, consumers could switch allegiance quickly and easily. Customer service was an area of possible improvement, and one in direct contact with the public. I was working at Genesys then, and I remember the spiel we adopted in customer conferences: how customers were a lot more likely to churn after a bad experience, so it was imperative to avoid one.

But this was still a negative approach: the real turning point was when companies started to factor in the customer experience not only as a cost, but also as a way to increase revenue, switching the contact center from a pure cost to be a revenue-generating activity. The switch is all in the accounting, but it’s fundamental: now there are reasons to invest more in the contact center and measure the overall revenue brought in by a better customer experience.

Which brings us to the latest development: the rise of conversational AI for customer service. Conversational AI was made possible by advances in natural language processing, understanding, and rendering, and the advent of Cloud architectures and the abundance of data available to “train” the Artificial Intelligence algorithms. Conversational AI changes the paradigm of how calls (or text-based interactions) are qualified, and in many cases allows customers to self-serve. The key is to understand the natural language that people use. This allows the computer program (Virtual Agent) to find out the intent of the call very quickly because the service domain is treated as a flat field and not a tree. Language is much more expressive than tones or single words and so an interaction goes through the qualification phase in 1, 2, rarely 3 question-answer exchanges. At this point, the Virtual Agent has enough information to either engage a computer application to satisfy the customer or forward the call to a human agent.

The customer experience is much better, and costs are also reduced for companies. This is because the biggest cost in contact center operation is the agents. Conversational AI has been proven to help boost agents’ productivity, and increase their job satisfaction: routine, boring interactions are best served by the Virtual Agents, which also collects information for the human agents to use. Agents are left with more interesting and engaging conversations.

But having more productive agents means that fewer agents are needed; happier agents means that turnover is reduced, with less need to train new agents since old hands are more able to solve customers’ problems faster. At the same time, fast, frictionless experience is what customers want when they contact companies: the best customer experience is when problems are services are provided quickly and easily.

Conversational AI is the present of call qualification, and its near future too. In the next few years it is easy to predict that contact center will rely more and more on a mix of humans and bots – the bots dong the grunt work, the humans taking advantage of it to deliver better and better experiences to customers. Beyond that, who knows? Making predictions is hard, especially about the future (Neils Bohr, various attributions).

The importance of customer experience in Telecommunications

Telecom companies, and especially mobile carriers, are completely dependent on their relationship with customers. Competition is intense and regulations make it easy to switch from one carrier to another. Since most carriers are equivalent for their device offering and network coverage, growing the business comes down to price, and customer experience.

Pricing is an internal strategic matter, but for customer experience, it needs to be agile, effective and, whenever possible, delightful, to keep the existing customers and attract new ones.

Customer experience means every (and any) contact between the company and the user, regardless of the channel. To compete effectively in this area, building customer loyalty, it is important to invest in strategies that personalize messages, streamline processes, and build an experience that will surprise and delight users.

Unfortunately, the telecom sector has a long history of poor performance in managing complaints. Precisely for this reason, carriers need to meet the growing consumers demand for better service and adopt robust and updated tools.

We at Interactive Media have seen this first-hand, due to our long experience with fixed and mobile carriers. For example Davide Arici, pre-sales manager, notes that “technology in itself is not enough: you need to have the support of vendors experienced not only in technology, but also in how to apply it intelligently to remodel the service processes“.

With an experienced partner that shares the common goal of a better customer service process, telecommunications companies can use disruptive technologies – always leveraging in full the supplier’s know-how – to launch a continuous improvement process in the short, medium and long term.

The power of artificial intelligence in customer experience

In telecommunications companies, one of the main areas where AI can make a difference is in customer service. Simplified and lean interactions, without wasting time, is what makes the consumer experience positive and Conversational AI is key to provide it.

“From this perspective, artificial intelligence enables fast and pleasant service based on features such as an open dialog for self-service, approaching human conversation“, points out the Arici. This way, it is possible to interpret the customer’s request and address it quickly in an automated way, without requiring the branching of static menus.

The key is simplification. The services offered by virtual agents powered by artificial intelligence need to be formulated and adapted on an individual scale, meeting the demands of each type of operation. Interactive Media, which has focused on customer service technological evolution for more than twenty years, has the necessary expertise to support companies in strategic choices – including the application of AI.

In addition to engineering, the deployment of conversational AI is like a recipe: the final system mixes diverse and complementary ingredients, technological, linguistic and psychological. “When it comes to customer service, the focus is on cutting the domain over the application’s target audience, shaping the dialogue to make it efficient in a specific relationship ecosystem”, emphasizes the pre-sales manager at Interactive Media.

So, the added value of the company that provides the artificial intelligence devices is not only in the robustness of the technology, but also in the ability to understand the operational details, reviewing the service flow and customizing the model according to the usage environment. To guarantee the effectiveness of the solution and the quality of the results obtained throughout the process – during and after the deployment of virtual agents in the telecommunications environment -, make sure you know the products and services of potential partners, and choose wisely what to deploy.

Voicebots in customer service

Voicebots are conversational AI applications, that conduct a voice dialog with humans. The application we describe in this article is to automate customer service – using their natural language capability to substitute the old IVR systems and saving customers from having to press keys to generate tones and slowly reach the desired service (if they are lucky).

With older IVR systems, still quite common today, users waste substantial time listening to menus and waiting for the option that seems the best fit for their need. Unfortunately, besides customer frustration, this originates many errors that causes calls to be routed to the wrong queue. The best outcome in this case is that the agent taking the call will transfer it to the right department, but often the call simply does not complete, and the user must call again – maybe a different number.

A well-designed Voicebot is instead an excellent tool to solve the problem of understanding the users’ intents. “Voicebots are a fusion of several technologies, some very new and other more traditional, like AI, computational linguistics on the one hand, and old-fashioned Information & Communication Technology on the other.” says Roberto Valente, Interactive Media’s CEO.

Users can interact with Voicebots in a natural way, answering open questions like “What is the reason for your call” Not necessarily the Voicebot will understand what the user wants immediately to the point that the service can be provided, but in this case it can ask disambiguation questions that restrict the field of possibilities, to make sure the intent is understood correctly: for example “would you like to fly in the morning or the afternoon?” It all comes down to the Voicebot design, and with 10 years of experience and many successful deployments, Interactive Media guarantees excellent customer experience results and a high percentage of interactions handled by Voicebots.

How Voicebots work

Voicebots are extraordinarily complex software applications. They must be able to listen, understand, talk, solve problems, look for information within the company’s knowledge base. They must do this with a high degree of security and for many users concurrently – hundreds, or like in some of the services deployed by Interactive Media, thousands.

One of the most important of the technologies involved is the Speech-to-text engine. Its function is to transcribe the voice from the user into the corresponding text. According to Mr. Valente, for better performance the Speech-to-text engine should be primed with information regarding the spoken language to be expected in the specific application domain. This could be for example financial services, or air travel, or telephony services provided by a mobile carrier. Each sector has its own specific set of terms that appear more frequently than in normal speech. There is then specific customer information that is very hard to understand, like names, alphanumeric codes (license plates for instance), addresses. In this case the system cannot use the context to narrow down the number of terms to recognize. As an example, if the conversation is about animals (for instance, at a veterinary clinic), the system may initially understand a word as “vow” but it can quickly correct it to “cow” – a far more likely interpretation. But for surnames, that are all unique, no such correction is possible.

The semantic engine is another fundamental technology for Voicebots. It is not enough to isolate words; the system must also understand their meaning – this is where the semantic engine comes in. As an example, “a ticket” could be many things, but a traffic ticket is different from a customer service ticket or a concert ticket. The role of the semantic engine is to distinguish these different tickets, and other more complex constructs.

Mr. Valente also highlights how important it is for Voicebots to be able to integrate with the telephone networks, and the numerous Contact Center software suites. This is an aspect that is often underestimated while implementing a Voicebot project but is often essential.

In addition to the ASR engine, the semantic engine and the telephony and Contact Center integration platform, there are other important parts to a Voicebot environment, such as a graphical environment for designing and managing the application flow, the management of alerts, reporting and service provisioning. All this is needed for a reasonable ease of implementation and operation of the Voicebot application.

Why voice?

People like to text and type short messages from their devices. It is a very convenient way to communicate on most occasions, but as a habit it’s fairly new, sparked by the availability of devices that let people text and type to communicate. If we focus on customer service, until 10 years ago it was only over voice. There many more channels on which to contact companies now, but telephone calls are still the most common way to do that, and certainly the most satisfying.

After all, if you get upset about a service or a product, it’s difficult to yell while typing: you could use ALL CAPS but, somehow, it’s not the same. Putting jokes aside, voice is what people use when they need to have real-time feedback, and voice enables people to convey information much faster than chatting: if you type fast and well you can put in 40 words per minute. But an average talker will speak 150 words in the same time (even excluding from the stats the end of pharmaceutical commercials). Finally, when all else fails people pick up the phone and call, so one could say that while voice calls may be going down as a percentage of the total interactions, their importance is actually going up.

Also, there are occasions when it’s OK to talk, but not to type: don’t text and drive! Although there are also occasions when texting is the only way to communicate, like when you are at a rock concert…

So, voice has a big role, and especially voice over the telephone network: dialing 1–800-SUPPORT is still the easiest way to get you there. Adding voice support to bots is consequently a great way to expand the reach of conversational technology in the customer service domain to the 50% or so of communications that are currently out of reach.

The challenges of voice

For bots, voice is harder that text. While voice can be transcribed into text rather easily by an ASR (automatic speech recognition) and the transcription can be fed to the bot’s AI, this is still an additional step that needs to be integrated into the system. There are also several TTS (text-to-speech) services that can be used to convert the bot’s answers back to voice — still another step.

What’s more, the knowledge base and AI training for text and voice is not completely overlapping: we say things and use turn of phrases while speaking that we wouldn’t use while typing; on the other hand, the ASR will not make typographical mistakes that are common in chat and must be accounted for by chatbot engines. But these are issues that can be overcome with better AI training — we at Interactive Media know this since we support both voice and chat in our conversational Virtual Agents.

More challenging, the system needs to be very responsive for voice: while no-one would object to a 10-seconds pause between typing a message and receiving a response, try that with voice! And so, the integration needs to be architecturally sound and fast. And not all ASR systems are created equal: while recognition performance of the latest ASRs is uniformly quite good, some systems have an advantage for specific tasks: for instance, Google Speech APIs excel in recognizing addresses due to their integration with Google Maps. It makes sense to use different ASR vendors for different parts of an application.

And then, there is the telephone network to deal with. There are certainly RESTful APIs that are easily integrated into a conversational system, but at volume they can be expensive. Also, usually the companies deploying the bot already have their own telephony infrastructure, and it doesn’t make sense to overhaul it for the use of the bot. Be it implemented through a local switch (PBX) or a SIP trunk from a carrier, telephony is more challenging to integrate with than a purely HTTP based interface.

Finally, if the interaction does not complete within the self-service conversational domain it will need to be forwarded to a human agent. This implies not only forwarding the call to a Contact Center suite (usually over SIP), but also passing over the context gathered so far, and for this an integration with the CTI interface of the Contact Center is needed.

So, there are several factors that contribute in making voice and telephony for bots a complex proposition.

An offer to help

Interactive Media knows a lot about voice and integration with other voice platforms. We started with voice applications, telephony and customer experience in 1996, and so we have both a long experience in what it takes to integrate successfully with the telephone network, and a super-solid platform that has evolved to incorporate the latest architectures and protocols into a proven foundation for all voice communication.

We also have a platform for conversational application with several sizable deployments, both for voice and chat. This has helped us understand the most impactful features of the telephony platform and optimize them as they relate to bots.

So the idea is simple: Interactive Media is on a mission to help chatbots add voice to their repertoire. This starts with telephony integration, of course, but continues with speech transcription and generation, and integration with Contact Center platforms — we integrate natively with several of the most common ones. Our software is already in the cloud, and you can try it at https://www.phonemybot.com.

We are looking forward to giving all deserving chatbots the gift of voice.

Cost of customer service

The cost of customer service is a financial metric that can be divided into two types:

• hourly cost: which must also include overtime, benefits and charges when calculated for teams of employees.
• total cost per service: the total cost of the period (day, week, month or year) divided by the total service at that specific period of time.

These KPIs are widely used as a criterion for comparison, considering that there are several solutions to deliver the service, from outsourcing to the implementation of virtual agents, such as chat and voicebots.

Abandonment rate

The abandonment rate indicates the percentage of people who drop out of the contact center calls while waiting for customer service. To calculate it, just divide the total number of dropouts by the total number of calls and then multiply the result by 100.

This is one of the most important KPIs, as it reveals points that need immediate improvement. If your calls are not being completed, certainly your customers are not only having trouble solving problems with your company, but they are also irritated by this situation.

The dropout rate, however, can be linked to several different factors and, therefore, other metrics need to be analysed to understand what can be done to lower it as much as possible.

FCR (FIRST Call RESOLUTION)

The FCR rate assesses the percentage of calls that are resolved in the first contact. It is, therefore, an efficiency KPI directly related to your company’s ability to identify the most recurring problems of your audience or your customers and offer quick solutions.

Its result is obtained from the difference between the total number of contacts resolved in the first call by the total number of contacts made, times 100.

Naturally, the faster the demands are resolved, the greater the customers satisfaction. Time is just as valuable as money, so any metric aimed at gaining agility in the different stages of the purchase process is highly appreciated by consumers in general.

TME (AVERAGE WAITING TIME)

If a long service time is stressful, a long waiting time is even worse. This is one of the main complaints of consumers of large companies that depend on call centers. The famous electronic songs played during the waiting calls have already gained space even in comedic scenes from successful series and films.

We have the TME by dividing the total (sum) of all waiting times by the number of completed calls. Its result allows us to estimate the efficiency of the team and the digital support solution, including pointing out the need to invest in training or expansion.

As you can see, service indicators are closely related. A small problem at the beginning or in conducting a contact can increase the waiting time for others and affect the quality of everyone’s experience.

TMA (AVERAGE SERVICE TIME)

Following the same direction as FCR and TME, TMA is used to identify the average time that has been required for your service to be performed. The TMA is the result of dividing the total and uninterrupted time of all calls by the number of contacts received.

However, this KPI needs to be used with caution. While it is always desirable to provide agile service, your strategy needs to be based not only on the customer’s solution, but also on delivering the right experience. This means that there is no point in solving the problem quickly if the impression generated by this call is impatience, neglect, or any other negative feeling. Whether performed by a human agent or carried out by a chat- or voice-bot, it is essential that your service is adapted to the rhythm and demands of your audience.

NPS (NET PROMOTER SCORE)

NPS is a methodology that aims to identify the contentment and loyalty of a company’s customers. It is one of the most used service and Customer Success KPIs in the world and its application is relatively simple.

In practice, it is a kind of a satisfaction survey in which a single question is asked: on a scale of 0 to 10, how much are you willing to recommend our company to a family member or friend?

When presented at the right time, the simple and incisive character of the question favors the honesty of the respondents, and their answers are classified into three groups:

• 0 to 6 (detractors): it does not necessarily mean that these people do not like your product or service, but it certainly indicates that they expect more from them.
• 7 and 8 (neutral): these are those who probably had their problem solved but did not have such a satisfactory experience.
• 9 and 10 (promoters): these are the ones who felt very well looked after, either by the product or by the experience provided, and are willing to promote their brand.

This type of analysis, as well as all others presented throughout the article, is usually performed with the help of software that is also able to organize this data and indicate the best actions to be taken by managers.

Strict monitoring of these metrics is one of the safest ways to evolve your service strategy and, consequently, increase customer retention and satisfaction. However, this is just the beginning. It is very important to optimize your internal processes, and technology can also play a key role in this regard.

Interactive Media is an expert company in the development, implementation, and improvement of virtual conversational agents. In addition to improving the experience of your customers and improving its service KPIs, the Interactive Media virtual conversational agents also benefits your employees by freeing them from exhausting and repetitive tasks and concentrate on really improving the customer experience for the benefit of your customers, and your company.