Oh my goodness, can you remember when you had to call a company about a little problem and got the ‘Thank you for calling Joe’s Plumbing Supplies, all of our agents are busy at the moment, please hold. You are number 23 in the queue.’ It was like watching Microsoft defragging.
Even the ubiquitous ‘For sales, press 1, for marketing, press 2…..’drove people crazy, especially if someone was using a mobile phone and had to take it away from the ear to punch in a number on the display.
Personally, it made me adamant that I would never contact or buy from that company again. Unfortunately though, it was usually a bank or a home services supplier and so I was locked in. I’d have loved to speak to a computerised voice if it would have helped me.
Fast forward several years….that computerised voice is now sounding more and more natural. And it can answer quite difficult questions and requests, as long as the programming is up to the task.
Welcome to the world of voicebots.
What is a voicebot and how does it work?
There are two types of voicebots. One is primarily designed for general use, such as Google assistant, Alexa and Siri, and the other is specifically designed for business usage. The business model is the one I am going to focus on.
With a combination of artificial intelligence (AI), natural language processing (NLP), natural language understanding (NLU), automatic speech recognition (ASR), and machine learning (ML), the business voicebot is integrated into an application or a website to assist consumers to navigate through the service and find what they are looking for. It has the ability to determine the intent of the user, by understanding the semantics and context of the communication.
When a caller connects to a voicebot, the voicebot will welcome the caller and ask a general question about the service(s) required. When the caller speaks, the voicebot transforms the speech into text, sends the text to an AI analyser, which uses NLU to understand the objective. The analyser then chooses a correct response, converts the text to speech and sends it to the voicebot who delivers it verbally to the caller. All this is done in under a second, in other words, about the same time that two real people communicate. The only difference at present is a possible slight change in stress. A voicebot might say ‘vegetable’ instead of the more common veg’table, but that’s not a big deal, it is still understandable to the caller.
The voicebot can also be ‘taught’ to use different languages, such as one using English and Spanish in the USA. It’s all about how the voicebot is set up and how it is used. Direct speech is much faster than typed, and so the transactions will take less time.
Are voicebots safe?
Some people are uncomfortable giving personal details online, and that is quite understandable, given the large amount of data breaches that are happening, but actually using a voicebot is better, and more secure, than speaking to a human. Let me explain.
If you are talking to a real person, say a sales consultant, there is a very slight potential for the consultant to use your data in nefarious ways. I’m not suggesting that this happens on a daily basis, but if there is a human element involved, it could happen.
Using a voicebot for verification purposes reduces that risk, as all data will be stored on the system. Credit card or social security numbers will not be exposed to typographical errors, and somehow people feel more comfortable giving sensitive information to a machine, rather than a human. The data is not touched by human hands. If a real agent is in the middle of a call with a client, he or she can transfer the call to a voicebot to get the required information, then the voicebot will re-transfer the call to the agent once the data is received and stored.
Having a robust security and compliance system in place will definitely reduce any potential for fraud or theft of data. Rigidly applying PII and GDPR rules will increase the strength of the system in the collection, storage and accessibility of data.
Consumer expectations
Over the past couple of years, online purchasing has become the norm. While plenty of people are reluctant to talk to a ‘robot’, it would seem that over 70% of those that encountered voicebots were quite satisfied with the results. Obviously there were times when a voicebot just couldn’t handle someone’s accent, or slang usage, and there are several collections online of hilarious interactions over the past few years between consumers and voicebots. Additionally, some people spoke too slowly, some too quickly. Past experiences with poor voicebot engagement has left a lot of scepticism. But as the technology has the ability to deal with digressions and changes in topic, with (almost) natural sounding speech patterns, I think that more people will take this onboard and learn to appreciate the speed and quality of a good voicebot.
Voice or chat - which is better for your business?
Ay, there’s the rub. It depends on your customers. There are three avenues of choice;
a traditional human service, with the drawbacks of staff, waiting times, and difficulty in the operation of a 24/7 service,
a chatbot service, which is text-based and great for non-linear ‘decision tree’ journeys where customers choose from a drop down display to get to the department or service they need, and is easier to train,
a voicebot, where connection is faster and the conversations are more engaging, but has a more linear approach.
Having used all three, my preference is the voicebot. Chatbots are fine, but my ability to make mistakes is much greater, and I always feel rushed. The traditional human service is not on the cards for me, unless I have no choice.
If a business can provide a voicebot with accurate communication, and pleasing vocal skills, timbre and articulation, the benefits will outweigh the downsides and costs. There are off-the-shelf solutions available which are pretty amazing, and can be easily adapted to any type of business environment.
As far as costs go, anywhere upwards of $500/month will get a professionally created voicebot service, with customisable services going for $1500/month and up. That’s a fairly hefty bill, but if it improves sales, marketing and customer engagement, and reduces other associated costs (call centre manning, for example), it’s worthwhile.
Alternatively, a business can build its own voicebot, and train it for company-specific functions. Voice actors can be hired to record personalised responses. This takes time and effort, but it is not beyond the capabilities of most companies.
If your business has an active sales and customer service channel, a voicebot will free up agent time and allow them to concentrate on high value tasks, improving the quality of their work. A voicebot can also perform customer satisfaction surveys, and of course you can ‘voice’ your FAQs for that extra touch.
With a robust security system in place, and a friendly, engaging voicebot, customers will eventually get over any reluctance to use such a service, and they might even, as I do, find it an enjoyable and satisfying experience.