Disruption Works Chit Chat

Chatbots - Back to basics, explaining what they are and what they are for.

Disruption Works Season 1 Episode 30

Today, we are all business, going back to basics on the conversation of chatbots. As part of our education push to help businesses understand what chatbots are and how they can be used in simple terms and will the basics covered.

  • What is a chatbot?
  • What are the benefits to a business?
  • What are the benefits to a customer
  • What is the difference between live chat and a chatbot?
  • How is the content handled?
  • AI/ML as part of chatbots, what is that about?
  • What other options are there for businesses to take a look at?


Find out more about chatbots on our site.

Our latest series of podcasts, concentrates on voice and how that is going to impact the next few years with tips along the way. Find out more about voicebots here and if you have any subjects that you would like us to discuss then email info@disruptionworks.co.uk with the subject Podcast and we will see what we can do ;-)

Auto transcribed by Teams.

00:00:12.120 --> 00:00:18.010
 Sean Bussell
 Hello everyone and welcome to another exciting edition of disruption works chitchat with Sean and Steve.

00:00:26.740 --> 00:00:57.270
 Sean Bussell
 Today we're going to be talking about chatbots and basically going through the basics of everything that a chat bot is and how it could be helpful for your organization as well as your customers. Now to give everyone a basic explanation of what a chatbot is, and sometimes people refer to these as chat boxes, I get that it's chat and it's within a box, but essentially they're called chat bots and that phrase is combination of two terms. Obviously one is chat, so IE conversational and the other is bot.

00:00:57.360 --> 00:01:05.410
 Sean Bussell
 So basically a robot and the idea is that this automates the conversation between the business and customer or consumer.

00:01:06.160 --> 00:01:36.970
 Sean Bussell
 Now chat bots are designed to provide immediate responses to inbound enquiries, and this can cover a plethora of journeys. From talking about what services you do, answering their FAQs, capturing customer details, etcetera. And they can deplete, be deployed wherever it makes the most sense for you and your customers. So this can be on the website, it could be on your social channels as well, such as WhatsApp, Facebook Messenger, Instagram and so on and so forth. Steve, is there anything you'd like to add in terms of to kind of basic definition of our chatbot is?

00:01:37.350 --> 00:01:50.270
 Steve Tomkinson
 No, I think you pretty much covered it there. It's it's all about automating that conversation really and it's you know the chatbot chat conversation that's coming into into a business. So yeah, that's you're pretty much covered.

00:01:50.690 --> 00:01:56.540
 Sean Bussell
 So we've touched on the basics of what it is, but Steve, what are the benefits to a business of deploying a chat bot?

00:01:57.490 --> 00:02:30.620
 Steve Tomkinson
 Well, the people are very precious at the moment and but every being is business or organization gets in a lot of enquiries into the business that our day-to-day ones or they have very simple process driven journeys and there are can be automated. Basically you're getting a human to be the robot in this particular scenario. So the benefits of having a chat bot are that you can automate quite a lot of those journeys. You know we see about 70 to 80% of those journeys automated.

00:02:31.000 --> 00:03:01.210
 Steve Tomkinson
 And UM and responded to. So any queries that come into the business, if you promote the chatbot and put it there front and center as a point of contact, then what people will now do is they will use that and start asking the chatbot of queries of what needs to happen next. That makes such a difference to the people that then trying to deal with complicated queries because they have more time.

00:03:01.530 --> 00:03:18.120
 Steve Tomkinson
 I see your staff of a lot more space and capacity rather than asking where can I park or you know, how do I cancel my ticket or you know all those kind of data day queries that you have that always come into any business. We all have them.

00:03:17.330 --> 00:03:30.540
 Sean Bussell
 So to keep it basic then Steve, it's it's we're saying it's sort of the 24/7 customer service experience answering simple questions to help save staff time, I guess. And from the customer's perspective.

00:03:23.460 --> 00:03:23.720
 Steve Tomkinson
 Yeah.

00:03:28.400 --> 00:03:28.690
 Steve Tomkinson
 Yeah.

00:03:31.360 --> 00:03:34.550
 Sean Bussell
 One of the one of the benefits to a customer of using a chat bot.

00:03:34.830 --> 00:03:48.300
 Steve Tomkinson
 Well, pretty much the 24/7 that you've just mentioned. So it means that they can contact and have a have a conversation with the business at the time that suits them. So of course that means they're what they're doing is they're actually having a.

00:03:49.680 --> 00:04:18.570
 Steve Tomkinson
 Uh, they're getting the the responses they need the information they need. UM, 70 or 80% of the time and and then even even if the default question can't be asked, they can probably then give all the information that's needed at the time, forget about it and leave it with the team to pick up when they're in and and those. So the customers are really nice experience in that whole process because they either get their question answered.

00:04:18.690 --> 00:04:30.040
 Steve Tomkinson
 Which is fine and or they get they they've started the process, they've raised a ticket, they've put a query into that business with all the information that's needed.

00:04:30.790 --> 00:04:51.970
 Sean Bussell
 Excellent. And let's give a quick example then. So let's assume that I run a restaurant. You're looking to come to my restaurant, but you've got some questions. They're fairly simple where you couldn't find the answer. You weren't interested in looking too deep on the website. So let's talk through a quick example. Maybe you weren't sure what to wear or whether you could just turn up or what was. Let's talk us through a quick scenario, Steve.

00:04:36.040 --> 00:04:36.300
 Steve Tomkinson
 Yeah.

00:04:51.940 --> 00:05:05.040
 Steve Tomkinson
 Yeah, well, that those the those are kind of typical of this. There are simple questions. So if you're going somewhere that's quite nice. You know maybe a quite nice high end brasserie or something like that and you want to know whether you can wear trainers to that then.

00:05:06.100 --> 00:05:37.090
 Steve Tomkinson
 You might have to pick the phone up because that information might not be specifically on the website or a bit. Like you said, it can be very deep in the website, so trying to find the information people just haven't got the patience anymore, so you can ask the chatbot the chatbot come back. That's fine. We're more than happy for you to wear trainers and they can go away. That question is done, finished, answered and they can then turn up kind of comfortable. They can wear their new trainers and they can have a good evening without.

00:05:37.180 --> 00:06:06.960
 Steve Tomkinson
 Fairly self conscious, should I, shouldn't I? They've asked the question already, but that crucially that business has not had to take another phone call no matter how short that phone call is. It still runs into a few minutes of time to pick it up. Go, I can I help? What do you need to know? I've got a book in. OK, all those interactions take time out of the call center or the teams time to say, oh, I need to.

00:06:07.300 --> 00:06:11.370
 Steve Tomkinson
 Deal with this query. The query they didn't know anything about it, it was just dealt with.

00:06:12.070 --> 00:06:19.580
 Sean Bussell
 And all that down downstream benefits of a business deploying a chatbot because maybe they provide an excellent customer service experience.

00:06:20.120 --> 00:06:46.590
 Steve Tomkinson
 Well, of course there is. I mean the if you're very easy to deal with and you're very easy to use from a customer's perspective than a customer's gonna have a much more positive experience if it's hard. So if that person, if we take the same example and that person was trying to find out about the trainers, it's not an important query, but it's a it's a nice experience query.

00:06:47.780 --> 00:07:10.650
 Steve Tomkinson
 But they couldn't get through because they couldn't go on to. They couldn't get three to the the the business because the phones were engaged. So that's to try several times before they could or they want. I'm gonna have to delve through this site. Let's have a look at the terms, you know, dress code, West dress code on here. All those things, they're gonna have to do might not be even being specifically answered, but.

00:07:11.480 --> 00:07:27.170
 Steve Tomkinson
 That's all gonna be a pain, and it's just gonna be an uncomfortable. I wish I could just find out, and they might just give up and then chance it or then not where the where. Some smart issues. If you like. They're just not comfy with, you know.

00:07:26.910 --> 00:07:44.220
 Sean Bussell
 But you, I mean, you're talking about an example where someone's have already made the booking, but they might, they might be holding off making the booking until they've worked out whether they can wear their comfy trainers or not. So again these we're trying to use a chat bot to remove these barriers to sail, so to speak, so.

00:07:30.790 --> 00:07:31.130
 Steve Tomkinson
 Yeah.

00:07:37.810 --> 00:07:38.690
 Steve Tomkinson
 Yeah.

00:07:43.550 --> 00:07:49.720
 Steve Tomkinson
 Well, we do. We do say that there's a lot of complexities to a lead generation journey. You know that actually.

00:07:50.720 --> 00:08:01.950
 Steve Tomkinson
 I think every business understands that is normally a lot of touch points before somebody books or does something with you, a restaurant if we keep it on the restaurant journey then a restaurant does have.

00:08:02.950 --> 00:08:34.790
 Steve Tomkinson
 A few points of call before somebody actually will book a table there. It's there may be a referral from a friend. It may be a review they've seen in the paper. It may be just that they found them on Google and saw the reviews on Google. You know, all those are touch points and this is an additional touch point that you can go right. Do you do halal food or do you have a good vegan menu and all those types of questions? I've actually important questions and they haven't booked yet like it's stead.

00:08:34.980 --> 00:08:56.510
 Steve Tomkinson
 But because they can ask those quickly, they then just continue onto the booking journey and it might not be the chat, but that does that journey. That chat bot will probably do the sign posting of that journey so that then is the lead generation still. But it's actually it's promoted a booking because you've had those additional points of contact.

00:08:57.080 --> 00:09:02.750
 Sean Bussell
 OK. Steve, give us a snappy. Did what's the difference between live chat and the chat bot?

00:09:03.450 --> 00:09:11.790
 Steve Tomkinson
 Well, live chat is a human agent that's managing the chat and a chat bot is that it's a bot that's that's entering the chat. So that simple.

00:09:11.840 --> 00:09:26.960
 Sean Bussell
 So essentially, yeah, you still need a human to resource that the same as you would answering the phone. So you're not really dealing with the queue, you're just you're just moving it somewhere else. And I guess there's some limitations around live chat as well because maybe you're not gonna manage that 24/7.

00:09:17.770 --> 00:09:18.030
 Steve Tomkinson
 Yeah.

00:09:27.610 --> 00:09:37.120
 Steve Tomkinson
 No. Well, it's yeah. It's only going to be resourced as you've got humans to resource it. There is some, you know this if you want to hand over because there's a superbly.

00:09:39.650 --> 00:09:56.740
 Steve Tomkinson
 A complex journey in that then uh, then they might need a human agent to put in there, but what happens usually is that all the information is gathered by the chat bot so that the human agent doesn't have to spend any time doing that gathering. And then if they're then passed through.

00:09:57.870 --> 00:10:11.750
 Steve Tomkinson
 They have all the information and they can go. OK. I can see what you're trying to do. I've had a little look and here is the answer. And or here is the task done because I had enough information from the chat bot and I've made my intuitive.

00:10:12.520 --> 00:10:21.970
 Steve Tomkinson
 Process to do that, so those are really good kind of mixes of journey. So you do have that, but if you're trying to do just live chat on its own.

00:10:22.960 --> 00:10:29.740
 Steve Tomkinson
 You're pretty much might as well get them to phone up. You know there's not. There's not that much advantage, a little bit, but not a lot.

00:10:30.370 --> 00:10:38.200
 Sean Bussell
 And and and it's the same thing as getting an engaged phone. If you go on live chat and the ones there, then it's just a poor experience, OK?

00:10:37.590 --> 00:10:46.590
 Steve Tomkinson
 Yeah, we get more and more of that as well. So we're saying that that very engaged live chat. So it then becomes just a frustration anyway. So it doesn't really help.

00:10:47.160 --> 00:10:55.910
 Sean Bussell
 So we've talked a little bit about the basics of chatbot Steve Howell do, how does it work? What what's, what's delivering the content?

00:10:56.910 --> 00:11:06.680
 Steve Tomkinson
 Well, you've got a a programmed conversation flow. So people like ourselves design the conversation flow to make it.

00:11:07.240 --> 00:11:22.880
 Steve Tomkinson
 Uh entuity have smooth for the customer either through button driven journeys or through free text. We tend to do a little bit of both because usually if button journeys are the ones thereafter, they will follow the buttons people are.

00:11:23.760 --> 00:11:54.270
 Steve Tomkinson
 Pretty much that lazy now that they weren't even type in a question, you know, so you you have, you have both scenarios, but if they it isn't there they can free type and then that brings back the responses that are programmed into the bot to answer questions. So you're FAQs we talked about you know kind of where a pair of trainers that would be understood. So that sentence can I wear a pair of Chinese would be matched up with the response and answer and you know that those conversation flows are designed very specifically so.

00:11:54.440 --> 00:12:19.030
 Steve Tomkinson
 Want the answers to get there and two, that they have a way out of those answers as well, so that isn't the right one. They can ask a question again. Well, they can be given a menu of options, you know, so they've always got a good flow and conversation route, but the answers can be text answers, but they can also be media answers, carousels, you know, links to pages, a lot of the.

00:12:19.930 --> 00:12:44.220
 Steve Tomkinson
 Answers now are deep, deep linking into the site we mentioned earlier that there's a lot of content on sites now and a lot of signposting goes on where the chatbot, because somebody just wants to ask a question that we target to an answer and the answer could be, oh, here's the page you're after on the website because this has got the full information, you know, a bit of a summary or find out more.

00:12:42.140 --> 00:13:02.810
 Sean Bussell
 Yeah. Is it summary? Yeah, here's a summary, if you like. More info, click. Yeah, OK, good stuff. Now, there's quite a few buzzwords that people think they need to be shopping for when buying a chat bar. And that's artificial intelligence and machine learning. Are those things? Do they really exist within chatbot? What's the context in this scenario?

00:12:46.140 --> 00:12:47.360
 Steve Tomkinson
 Yeah, yeah, yeah.

00:13:03.710 --> 00:13:32.490
 Steve Tomkinson
 Where there is some AI is a is a an overused term I think, and AI is broken down into lots of different areas, but machine learning is one of the big portions of AI and basically it's a lot of data learning what the pattern is behind it and that's basically it. And the computer starts to understand the pattern and then it gives answers based on those patterns. So it's a really, really powerful.

00:13:32.560 --> 00:13:37.480
 Steve Tomkinson
 Elements of AI. But AI is a much more intuitive.

00:13:39.800 --> 00:14:09.830
 Steve Tomkinson
 Route that that provides a lot more information on computer vision and file. You know, investigation of sentiments and things like that. Now we do use some of that within the bots, so sentiment recognition, we do machine learning based on phrases, but there's not AI as such. The problem with AI is it does have to learn. So it has to have a tremendous amount of data in the background.

00:14:10.510 --> 00:14:40.080
 Steve Tomkinson
 Now the voice bots that we do are different. Umm, we use, uh, natural language understanding in there to understand what's been said. So speech to text. But because it's a chat bot, we don't need that layer because we're already getting text. So the text then is pretty well driven, to be honest with you, there is a text to speech elements within the chatbot that's used very little, but we don't necessarily worry, worry too much about that.

00:14:40.460 --> 00:14:54.030
 Steve Tomkinson
 But AI chatbots as such, that is very much a buzzword, and it's not really a thing in a lot of regard and not a useful thing for businesses that have to have the right answers on day one. You know, it's a long.

00:14:44.920 --> 00:14:45.170
 Sean Bussell
 Yeah.

00:14:52.890 --> 00:14:57.000
 Sean Bussell
 Yeah, you're gonna lose a lot of that, OK.

00:14:54.140 --> 00:15:12.950
 Steve Tomkinson
 It's Allen. Yeah. Yeah. But you think it's OK for wrong answers for a while until he gets the right answers. Now, that's OK. If you can do it in a research environment, but it's not OK if you're trying to answer questions like kind of where trainers and then it goes. Yes, you can book a table for tonight. You know, because it doesn't know it yet.

00:15:13.700 --> 00:15:14.000
 Steve Tomkinson
 And I.

00:15:13.800 --> 00:15:25.100
 Sean Bussell
 So we we I mean we touched on the fact that they're they're business I guess like ours that will build all this stuff on behalf of their clients. What are the other options deed for people to get started with a chat bot?

00:15:22.000 --> 00:15:22.210
 Steve Tomkinson
 Yeah.

00:15:25.370 --> 00:15:54.630
 Steve Tomkinson
 Well, there are a lot of platforms out there so that people can give it a go themselves. So and they're, you know, they're very cost effective in a lot of regard because they are you know there's a few out there. No no they're cost the cost effective in the way that you know there is a cheap starting point but it's but the problem is is you've got a value your time and whether you have the expertise in here. I was saying earlier about conversation flow and that's an important part of the design.

00:15:35.940 --> 00:15:39.020
 Sean Bussell
 Well, it's supposed to come and say, but I can't see it.

00:15:43.090 --> 00:15:43.580
 Sean Bussell
 Yes.

00:15:55.190 --> 00:16:03.580
 Steve Tomkinson
 And we've made some serious, you know, some mistakes through the through the period of of building chatbots over the last 2-3 years.

00:16:04.160 --> 00:16:29.750
 Steve Tomkinson
 And that we've started to realize what the good patterns are, what works best, what kind of mix of buttons and free free text journeys should be in there, how to identify which ones are the top journeys that should be presented, and how then to use platforms, the best to present a really nice, very natural conversational flow.

00:16:30.930 --> 00:16:53.880
 Steve Tomkinson
 That takes time to get to that point, and we've already had people coming into us to say no. We've had we tried try to use a platform that costs us, you know, 100 pounds a month or something. But so it's it was cheap and cheerful, but we hemorrhage time. We lost so much time in this process that it just didn't work and we didn't get anything in the end of it.

00:16:54.340 --> 00:17:05.900
 Steve Tomkinson
 You know nothing that we could publish because it just didn't work, you know? And that's the point. I think that's the those are the options is that you can do that build yourself. But the problem is is that.

00:16:58.130 --> 00:16:58.570
 Sean Bussell
 Yeah.

00:17:06.550 --> 00:17:19.620
 Steve Tomkinson
 You're very quickly become unstuck it you know, because it's not a website, it's something else. And that difference is people that were there aren't enough people that know that kind of have that knowledge yet.

00:17:20.740 --> 00:17:26.570
 Steve Tomkinson
 Uh, we are, we, we we have doing this and this is what we do. So from a specialist perspective.

00:17:27.870 --> 00:17:35.190
 Steve Tomkinson
 It we built stuff from day one that works and I think that's the that's the the crucial option to to consider.

00:17:34.450 --> 00:17:50.740
 Sean Bussell
 And you said that this is not a website which is true, but are there chatbots out there that just purely scrape your websites are IE they do a Google search, let's say within your own website go I, you know, can I wear trainers? I think this is the answer and it's always gonna be I think this is the answer and then the link to the page at all.

00:17:51.030 --> 00:18:14.340
 Steve Tomkinson
 Yeah, there's a certain amount, though. It's a certain amount of knowledge based ones. They're scraping ones that they're kind of, they're a little bit Google does a better option and Microsoft do a better option of adding knowledge base information in there. So you can actually have documentation and article based content and that starts using some of that machine learning we're talking about earlier to pick out the right answers.

00:18:15.740 --> 00:18:25.050
 Steve Tomkinson
 But the hardest setup, they're not the the you know, they're not simple things to to go and deploy. And also it's not very conversational. It's a very.

00:18:25.860 --> 00:18:49.160
 Steve Tomkinson
 1 dimensional uh process. You've gotta remember that the FAQ part that we've kind of talked about just for simplicity sake, is only part of what we would normally design. There's also a lot of processes that we start to automate when we get the conversation with our clients or we go is. But what part of that conversation do you actually need to do? You know, there is a.

00:18:27.310 --> 00:18:27.980
 Sean Bussell
 Yeah.

00:18:50.180 --> 00:19:14.940
 Steve Tomkinson
 There's a process that people are coming into you for. Might be a booking or it might be a, uh, a query of a certain type. What do you want? We can automate that. Is that useful to you? Would that take a lot of pressure away from the team so that those are the additional bits? So the scraping thing is a great idea from the service, but it's very one dimensional and it doesn't do the big enough job.

00:19:13.180 --> 00:19:13.670
 Sean Bussell
 Yeah.

00:19:15.790 --> 00:19:18.620
 Sean Bussell
 So who can we go? Can we buy chatbots from?

00:19:19.460 --> 00:19:32.750
 Steve Tomkinson
 Well, I mentioned there's, uh, there's a some of the big vendors, so Google, Microsoft is probably a complicated one. There are three other vendors like chat, fuel and people like that out there. So there are a few about.

00:19:33.270 --> 00:19:46.110
 Steve Tomkinson
 Umm, but it again. Do you usually give it a a blank platform? You know, Google dialogue flow is a is a big platform and it's very, very powerful tool. But I feel that.

00:19:46.180 --> 00:20:02.840
 Steve Tomkinson
 It's, you know, you could open it up and have a look at it and go ohh. Don't even know where to start here. You know. And that would be then, a complication in its own right, you know, just like so many layers to these platforms now that it's it can cause you a problem.

00:20:03.520 --> 00:20:22.790
 Sean Bussell
 Get stuck. Alright. Well, I hope everyone found that useful as a as a nice starting point in around what? What chatbots are and some of the basics. If you have any questions, do you feel free to get in touch and if you'd like a more in depth chat with us around what opportunities are for your organization for having a chat bot then again just get in touch and we'll be happy to help. Thanks very much.

00:20:23.080 --> 00:20:24.230
 Steve Tomkinson
 Thanks a lot. Cheers.

People on this episode