It has happened to most of us. You just got home after a hard day in the office and are just settling down with a nice glass of wine. The phone rings.
You pick it up and there is total silence at the other end of the line... Is it a homicidal maniac trying to intimidate you and your family or... is your partner having a torrid affair... or perhaps... it could be a gang of burglars trying to find out who in the neighbourhood is not at home... You quickly dial 1471 only to find that "the caller has withheld their number".
An hour later, you have just sat down for your supper, when the phone rings again. You rush to it, pick it up, and guess what... nobody there. 1471 confirms it: they don't want you to know who they are. A tense silence descends over the dinner table. Your mind is now racing. Is your partner trying to avoid your gaze? Has he/she been behaving a bit odd lately? Are your children in danger?
The silence is broken by the phone ringing yet again. You stare at each other in horror for a moment and then hesitantly lift the receiver. To your relief, a chirpy voice at the other end is trying to sell you some insurance, or perhaps a new credit card or an unsecured loan... You tell them precisely where to go and life returns to normal.
The reason this is happening to more people more frequently is directly related to the proliferation of predictive dialing systems deployed by contact centres up and down the land. The temptation to use such systems is great because they create huge operational savings by making the agents' time more productive.
The idea is simple. In a typical outbound dialing operation, where the agents dial out from a list, only about 20-30% of their time is spent actually talking to customers or prospects. The rest is spent dialing and waiting. The main problem is the fact that depending on the time of day only 20-40% of calls get connected to the right person. The rest are either not answered or busy or an answering machine.
A predictive dialing system would do all the dialing automatically and only present the agent with live, connected calls. This typically increases the talk time to 70-90% as a proportion of the agent total working time. The savings are enormous and therefore more and more organisations are spending on this type of technology.
The term "predictive dialing" is derived from the fact that in order for such a system to do its work it needs to predict how many agents will become available within the next few seconds. Also, it needs to predict what the likelihood is that any call made would actually be connected to a human. It then takes all these predictions and decides how many calls to place. It's an informed gamble.
All these predictions are usually based on an analysis of historical information. The way this works, is that the system will calculate an average of call durations for each individual campaign and then assume that future calls will not deviate too much from that average. The same applies to success rates, in the sense that if 20% of calls were connected on average for the last hour, chances are that in the future one call in five will also be connected.
Different predictive dialer vendors compete with each other on the basis of the sophistication of the mathematics used in the delivery of their predictions and the accuracy that these complex formulae can deliver. The truth is that, whatever the level of mathematical wizardry, in the end a prediction is just a prediction and since most of them are based on probabilistic methods they only work for large numbers. So the more agents the call centre has the more accurate the dialer becomes in its predictions.
The so-called "silent calls" or "nuisance calls" that we described earlier are the result of all these predictions going slightly wrong. And they always do, because the main feature of averages is that on average they are wrong. The fact that the average call duration for a campaign is three minutes does not mean that all calls are three-minute long. Indeed, the truth may be that about half of the calls are 10 seconds long and the other half six minutes long, and so on.
So, when these complex probabilistic calculations go wrong (and they invariably do) two things can happen. In the first case, agents can become available before there is a call for them to get on with. So they wait for a while until the dialer manages to deliver a connected call. This means that the call centre productivity is reduced and the company doesn't get the entire potential benefit from the dialer.
In the second case, and this is what causes "silent calls", a call is connected but there is no free agent to take it. This is when the dialer has no choice but to "drop" the call, i.e. leave the person who picked up the phone at the other end with a period of silence followed by a dialing tone.
Let us not forget that it is actually in the interest of call centres to minimise the number of silent calls made. There are two main reasons for this. The first is that any person that is finally contacted following any number of silent calls is likely to be less receptive to whatever message the call centre is trying to convey to them. This means that even though the call centre is very efficient the conversion rate of the campaign is degraded. In other words it's better to be inefficient but effective than efficient and ineffective.
The second reason is that the public has now a choice. The Telephone Preference Service (TPS) and Silentcall-Gard are two ways in which people can remove themselves from the radar of most outbound call centres for good. This means that unless the more blatant abuses are stopped, the number of people that are still available for calling will decline. There are already 1.8 million people registered with TPS.
Is there anything that call centres can do in order to avoid making silent calls? The only absolute guarantee that no silent calls are made is not to use a predictive dialer at all. This is not likely to happen unless such dialers are banned. In the current global market for call centre services this is not only unlikely but also highly unfeasible. Self-regulation in the industry remains the only reasonable option.
The next best thing is to adjust the technology so that the minimum possible amount of silent calls is made. Most dialers will allow call centre managers to adjust their dialing parameters so that the "nuisance rate" or the percentage of silent calls out of all calls dialed is kept under a certain limit. The current DMA guidelines are set at 3%.
The problem with this is that it has a detrimental effect on the performance of most dialers. This means that call centres will see a significant degradation in productivity and as a result a negative impact on the bottom line.
Recent developments in predictive dialing technologies are going some way towards addressing this problem. The idea is to reduce the reliance of the dialer on probabilistic prediction and base more of its decisions on hard facts. One way of doing this is by integrating the dialer closely with the software that agents are using on their desktops.
Most call centres use some form of call scripting. These are software applications that guide the agent through the conversation and assist them in capturing data. The great advantage that these types of applications have is that the position of an agent within the call script is a strong indication of the time left before the call is likely to end.
Integrating the call scripting software with the predictive dialer in such a way that the dialer is constantly aware of the agent's current position within the call script can eliminate a major element of guesswork from the internal workings of such dialers. The net result is that the dialer can base its prediction on the number of agents likely to become available on facts rather than averages and other complex mathematical artifices.
As a consequence of this new idea the "nuisance call rates" can be kept low without having a dramatic impact on the performance of the dialer. This means that call centres can begin to realise the savings derived from the use of dialers without having to alienate their potential customers in the process.
It is not a perfect solution, but it goes some way towards a situation where outbound telemarketing can begin to restore some of its tattered image as nothing more than a nuisance. And who knows, we may even be able to enjoy our dinners in peace once again.
About Danny Singer:
Published: Tuesday, June 8, 2004
VoiSentry is a voice biometric engine for speaker verification / identification. It is delivered as a virtual instance to be controlled by an application via APIs and can be deployed in whatever is the most appropriate environment i.e. public cloud, private cloud or on-premise. It is generally deployed as a white-label or OEM product, whereby Aculab’s partner integrates it into their own solution/platform and offers it as an integral part of their service. Unless they go public with it, none of their clients would even know that it was Aculab’s VoiSentry engine under the hood. As with any biometric speaker verification system, VoiSentry is language and dialect independent.
Agara is an autonomous virtual voice agent powered by Real-time Voice AI. It is designed to have intelligent conversations with your customers, vendors, and partners without any assistance from human agents. It can handle a wide variety of calls including inbound customer care calls, outbound lead generation calls, appointment scheduling calls, and overdue payment recovery calls.
Agara is available for several industries including banking, insurance, retail, e-commerce, airlines, and telecom. Powered by advanced Real-time Voice AI that understands speech in real-time, automatically determines the right process to follow and guides the caller along in the process with natural conversation.
|4.)||Call Tracking Metrics|
Execute automated actions based on words spoken by your customer.
Our intelligent word detection and transcription tools make it easier for you to decipher the content of your customer conversations and automate action based on those insights.
Transform the audio of your calls into speaker-organized scripts so you can easily search your call log for specific words from a conversation, or automatically detect when words of intent are spoken by your customer.
Listen in on and highlight common patterns in order to execute triggers to initiate a new action or workflow, such as emailing a manager if the word “cancel” is spoken on a call. You can also set up to automatically redact content from a transcript, such as sensitive health information transmitted on a call.
Daisee builds technology that empowers people to solve problems by making interactions simple and smart so they can have a more significant impact, be more productive and be better at what they do. We believe incremental improvements carry huge potency and provide exponentially greater change for the better.
Genesys customers can now automate risk and quality management using Daisee’s speech and sentiment analytics and remediation workflow software. Daisee helps improve customer experience, ensure regulatory compliance, identify missed commercial opportunities & training requirements as well as provide valuable insights back to the business directly from the frontline – the true voice-of-...
Advanced speech recognition and natural language expertise that routes callers by simply asking ‘how can I help you?’. It allows customers to use their own words to ask for what they want and steers them to the right place.
Menus are eliminated altogether and contact centres can use a single inbound telephone number.
SmartSpeech – The Best In Class Speech Analytics Software
An all-inclusive speech analytics application that enables you to visualize your audio using state-of-the-art speech recognition, transcription, and text analytics technologies.
V-Blaze Speech to Text
Our GPU-accelerated, AI-based technology enables you to deliver greater insights to the contact center by transcribing audio into analyzable text.
- Transcribe large volumes of recorded audio quickly
- Integrate transcripts with your analytics or business intelligence platform
- Reduce your hardware footprint and minimize total cost of ownership
InTone Suite - Emotion Detection via Voice Analysis
InTone™ is a voice analysis tool geared to detect genuine emotional and cognitive reactions regardless of the speaker’s language or chosen intonation.
It integrates into existing contact center environments and provides realtime indications of both client and agent emotional reactions throughout all ongoing calls. This unique set of indications assists managers and call center representatives to truly understand the client, improve call center performance, easily detect angry calls, elevate training efficiency and provide a tailored customer experience in every single call.
Nemesysco's contact center technology is also available in API and SDK forms.
Please visit our website for information about our product portfolio.
Voxjar is your one-stop-shop for driving better phone calls with your customers. We're a modern Conversation Analytics and Call Coaching Platform.
- Custom Charts and Dashboards give you the freedom to analyze calls from the highest level down to the details of each conversation.
- AI-Powered Speech Analytics transcribes, searches, and analyzes your reps' phone conversations, at scale.
- Quality Assurance and Call Coaching tools let you evaluate and improve agent performance with a human touch.
- Long term recording storage for historical analysis
All of these together provide the holistic, easy-to-use software that you need to manage a modern call team and create better con...