Top takeaways from a one month tech pilgrimage



Host: Marco Oliver, Client Success Director, WBB.

Guest: Adam Greenwood, CEO Greenwood Campbell.

 

Marco Oliver: Thank you for joining us on The Botcast Adam. How are you today?

Adam Greenwood: I’m great, thanks very much.

MO: Good, glad to hear it. We always like to kick off The Botcast with a short intro into the background of our guests, so could you give us a quick insight into how you became the CEO at Greenwood Campbell?

AG: Yes, sure. So, my friend Ian and I used to work at a software company about ten years ago. We left there and went to work for a digital agency and after about a year, we realised that we could do this ourselves so we left and set up on our own.

We were in a really lucky position that the two of us had a lot of expertise around this particular piece of software – it’s like an old content management system – that nobody else really could touch anymore, and it had some great customers. So, from Ian’s spare bedroom we were working with clients like Skoda, British Museum, Dixons, Visit Scotland, and everything was amazing.

For the first couple of years there were three of us working with big clients, no real hassle at all, didn’t have to market ourselves. But over time, as this software started to go out of circulation, we kept some of those customers and migrated them on to different systems. And then became a, kind of, traditional digital agency for the next four or five years. So, digital marketing, UX strategy, design build, that kind of thing. But then about 18 months ago we decided that if we carried on just focusing purely on websites there may be a bit of a problem, and that’s where we started to look at different technologies.


18 months ago we decided that if we carried on just focusing purely on websites there may be a bit of a problem, and that’s where we started to look at different technologies.ADAM GREENWOOD


MO: That’s quite interesting because that’s a very similar story to ourselves at We Build Bots, we were incubated as part of Coup Media which was a digital marketing company. We had exactly the same experience and then spun off into our own company which was where we started building chatbots and it went from there really.

What advantages do you think it gave you coming from a marketing background going into the space you’re in now? Do you think that gave you any real benefit?

AG: Yes. I think it just made me realise that actually the technology is only one part of it. And I think if you’ve got a really good understanding about how brands should engage with their users regardless of technology, then I think that’s a good place to be as opposed to just doing something for technology’s sake.

MO: Yes, definitely. I notice you used to be a business analyst as well at  Immediacy, so I’m assuming that’s given you a good background as well to understand what the customer is really trying to get into.

AG: Yes. I think it was great, a really good experience for me. The job I had before, I used to work for an aviation company and I did that for about ten years, and we sold military aviation spare parts to foreign air forces. And I did about five years in sales and it was great. I travelled all round the world, but I wasn’t really interested in aeroplanes to be honest. And then an opportunity came up in their IT department because they wanted to completely redesign their EPR system, but it was not just ERP, it was quotes and it was stock control and purchasing. It was the whole backbone of this company.

And so, I then worked with a software company to build this stuff bespoke and I spent some time with their business analyst - a couple of years actually and I just thought this seems like a really great job that he would just go into totally different types of businesses, completely immerse himself in understanding that business, help the developers build the software and then move on to something else. So, when I became the BA at Immediacy, the sales guys would sell the software at all different types of organisations - so, the ones I mentioned earlier plus, the NHS, local government, healthcare and hospitality organisations.

What I would normally do is, I would go in with a couple of developers and spend a few days with their marketing team and say right, so you’ve bought the platform, what else do you want it to do? I started to do the same thing, immerse myself in different types of businesses, but also, work as a translator between the business leaders and my developers.

I’ve never been a developer, but I understand the fundamentals. I can’t code but I do understand how things work. So, I would sit there and a CMO would say right, so we want it to do this, this and this and then I would turn around and translate that to the developer. I mean, not all devs are like this but certainly some of the guys I work with are, kind of, cave dwellers, I suppose is an interesting thing to call them. I’d say to them, so we could do this right? And they would normally um and ah a little bit and then they would say something to me and then I would translate that back to the CMO. So, it was good. It was a great way for me to understand both sides of the fence.

MO: Definitely. I was originally a developer and then built on from there. I think it is important that you understand both sides of the fence, as you say, because there’s nothing worse than 1. over burdening your technical teams, and 2. promising something that can’t be delivered. So, it’s really important  to have that grasp on that side of things.

AG: Yes, definitely.

MO: So, we’ll get on to Greenwood Campbell and what you do but, you live in Bournemouth, is that right?

AG: That’s right, yes.

MO: It must be pretty good in Bournemouth? A lot of water sports?

AG: Yes. I love it down here and this summer’s been obviously incredible. Getting up and meeting some of the guys from my office at five o’clock in the morning doing a three-hour paddle board, finding a little spot on Brownsea Island, having bacon sandwiches, all before work. And we were doing that about three times a week this year, so it’s good.

We’d sometimes do that and then get on the train to London and then just feel a bit smug all day really. Everyone else has come in on the Central Line. Yes, it’s been nice. It’s a great place to live and that’s why we’ve set up our company down here because we just love living down here.

MO: Yes, it sounds like a great work life balance.

AG: Yes, definitely.

MO: Lucy is really good at getting the dirt on people. One of the things she came through with was that you did a one-month tech pilgrimage. So, what exactly did that entail and what did you find out from it?

AG: It was about three weeks I think I was there. About 18 months ago, it seemed like our pipeline for big, two/three hundred-thousand-pound web projects just felt like it was drying up. We just weren’t seeing as much opportunities and I think a lot of that’s to do with the commoditisation of web development. Things that we would normally charge pretty hefty fees for, you could do most of in Squarespace or Shopify, even Wicks. They’re really powerful tools and it’s becoming more and more difficult to be able to justify the value of some of the stuff we were doing.

 

 

We still do big technical integrated projects, but they tend to be now a bit more intranet focused. And, you know, with that kind of thing you can’t do a Squarespace, you can’t do Microsoft dynamics and AD and all that kind of security stuff without a really good team of people. So, we still do a lot of that work but I felt that that kind of stuff was drying up. And so, as a result I wanted to start trying to find out what else is out there, what’s the new technology that potentially could help differentiate us.

We’ve got a really smart group of people here. So, I thought what can we go out and learn that can make us a bit different from our competition and hopefully could stop us going a way of a lot of print agencies. It wasn’t the best start to the trip. Basically, I was going on a stag do to Vegas –

MO: Sounds like a very good start!

AG: Yes. Literally two weeks before I went I thought l need to go Silicon Valley, that’s where everything’s happening. I thought, I’m only going to be an hour from San Francisco so I might as well just leave the stag do at the end and fly out there on my own. So yes, four days of a stag do with no sleep is not the best start.

Having to then fly off to San Francisco on my own and then day one of my trip was an all day Japanese robotic seminar. I have to say, I struggled to stay awake during some of those talks. It was really, really tough. But really what I did was, I just used things like Meetups and Eventbrite and tools like LinkedIn and my network and I just said look, I’m going to be over there, who should I meet? Where should I go? What can I do?

I spent time at Stamford University going to machine learning lectures. I went to some conferences. I met the CIO of Toyota who was talking about autonomous vehicles. I learnt about robotics in the supply chain. Also some crazy things like drones that can take off outside a hospital with a blood sample and go directly to another hospital. A lot about AI and then also chatbots. And people were just starting to talk about voice - Alexa had been out for maybe 18 months in the States. I came back to my 40 strong agency of web developers and said right guys, websites are dead. We’ve got to start looking at something else.


I came back to my 40 strong agency of web developers and said right guys, websites are dead.ADAM GREENWOOD


MO: I bet that went down well.

AG: Yes. Retrospectively maybe I should have eased that message in a little bit. But, you know, a couple of people didn’t take it too well and eventually left and I supported that but the really smart guys in our team embraced this change and said okay, let’s look at how this new technology can benefit our clients and help us find new ones.

MO: Yes, definitely. It sounds like you packed a lot in which is really good. I think you’re right and I think the direction of things are always changing, so you need to be adaptive in this industry ultimately.

AG: Definitely.

MO: So obviously, that formed the foundation then for Greenwood Campbell. Do you want to maybe just give an overview of what it is that you actually deliver and can deliver to customers?

AG: Yes, absolutely. We are still doing big corporate, big integrated websites using CMS’s like, Sitecore. We also do these, kind of, service design UX projects for intranets for people like Dixons, who we still work with and the RNLI who are down in Poole. But the kind of stuff that I’m really excited about is how we can use AI to create real engagement and differentiation for the brands that we work with.

The first one that we did was just kind of a punch to see what might happen - was the Glastonbury Festival. We had a lot of developers and people who work for us, all very young and all went to this festival. So we said what’s a good use case template? What do you want to know when you’re there? And people said well, you know, we struggle with service, so if you haven’t downloaded the app it’s hard to download the app. And also, as we’ve seen, people aren’t really interested in downloading apps anymore. They don’t have room on their phones. So, we said okay, so this could be great. A chatbot working in Facebook Messenger could really work in this particular scenario. So, then we just sat down and said well, what do you want to know? What are the bare basics? We looked at really simple things like when is a particular act playing? Who’s playing at a certain stage right now? And that was, kind of, it.

We managed to get a data feed of those two things, acts and stages, and we programmed them into a programme that was called API.ai which is now Dialogueflow that Google owns - it’s sort of an AI chatbot builder, a mix of chatbots and voice. So, yes, we had context and intents and we just built that there and then and a couple of a weeks later we launched it at Glastonbury and we had about 50,000 messages over the weekend.

MO: That’s not bad at all seeing it was just launched. That’s good.

AG: Yes, really interesting. And one of the things that we learnt very quickly was – we see a lot of chatbots that have a lot of buttons because they want to just guide you down a fairly narrow focused set of things that you can do. But that’s not great if you want to learn about what your users actually want to know. So, we kept it relatively open and we tried to think of as many different intents as possible so that it wouldn’t get confused and it would understand. But we also learnt that if you keep a chatbot too open, then people ask it some pretty strange things. And, you know, especially at three o’clock in the morning - people who are at Glastonbury can ask some pretty strange things of a chatbot.

MO: I can imagine.

AG: In fact, we had a team, there was about four or five of us, over the weekend just wanting to be there to answer any questions that the chatbot couldn’t answer. And I manned that for six or seven hours on the Sunday I think, and yes, I had some pretty weird requests and I was ending up having conversations with people.

MO: We do something very similar to that as well – well, exactly the same pretty much. It’s all about, like you say, capturing the stuff that people don’t know or need to know and then it’s feeding that back in. And it’s at the stage at the moment where I think we need to control the AI so that it’s feeding it back in but not allowing it to just completely teach itself for obvious reasons, as we’ve experienced when looking at some of the Twitter trolls and stuff like that.

AG: Yes. What was the Microsoft bot called?

MO: The one that they released for 18-24-year olds?

AG: Yes. Tay, I think, wasn’t it?

MO: Yes.

AG: And they turned it into a xenophobic Nazi within about 24 hours.

MO: That’s right. People have a tendency to try and break these things because it’s new tech. I think over time that will obviously die down and the hype will die down slightly because it will be seen as something that’s the norm. But at the same time, it means we’ll get much better results from it, I think, and be able to actually use it more efficiently.

AG: Yes, I mean, if anything I think that was a massive success that project because regardless of the content that people were trying to teach it, just because, like you said, it’s just Twitter trolls, it still learned. And it learned very, very quickly.

MO: It did.

AG: So, I thought it was very good.

MO: Yes, definitely. So, just coming on to the company’s approach. I did have a poke around the website and one of the interesting things I’ve seen is that your approach developments, you use an agile and waterfall blend. So, if someone was going to come and work with you, how would you describe how that works and what that blend is?

AG: I think it depends on the project. We’ve had some clients come to us and they were very specific about what they wanted a chatbot to do. And so that was quite good because we were then able to flesh out specifics around the conversation types and the intents but with a very narrow focus.

We built one for a university which is going to go live next week I think, we’re just waiting for clearing to end. They had three very specific conversations. They wanted students to be able to find out about courses, accommodation and open days. Those three things, and so basically this bot had a very narrow focus, but the conversations got very, very deep. It was quite easy just to say okay, so we’re going to work on each one in a, kind of, waterfall fashion. So, we’ll do this conversation string and we’ll test it and then we’ll do the next one and so on.

Whereas we’re working with an insurance company doing a customer service bot where they have got a very, very broad focus, lots and lots of different questions and answers. We sat down and spent some time with the customer services team, listened in on calls, we also got them to log all their calls for about ten days so that we could at least get an understanding of what were the top 20 or so requests that people were asking for that they felt could benefit from automation - ultimately to free up the customer services team to deal with more complex and more important tasks.

So, when it’s something like that where it’s very, very broad and actually quite shallow, that was much more of an agile process where we would say right, we’ll just work for two weeks on these ones and then we’ll see what comes of it and keep going like that.

MO: That seems like a very sensible approach and yes, we do exactly the same actually because you do find there’s different customers. I mean, government and private sector customers are always very different as well as a general rule. They have different ways of working, they’ve got clear scope, or they haven’t, you have to be adaptive in this industry, don’t you? So, that is really important.

AG: Yes, definitely. I think the ability to say, okay here’s your budget, you know, typical agile stuff. So, let’s just MoSCoW everything and then if we find a bit of extra time here, then we’ll pigeonhole this bit in and see what you come back with. And you do need the right type of customer to embrace that because as much as agencies for years have talked about agile, clients think ‘oh, we like that but we don’t want to pay agile. We want to know what it’s going to cost. We don’t just want to pay for an indeterminate amount of time’.

MO: Yes, that’s so true. So, you’re also branching now into – have branched into the voice space following on from that. And I noticed you’ve got an assistant on Google Home, so do you want to talk a little bit about that?

AG: Yes. We’ve done a couple of things with voice. What we did is we gave everybody Google Homes just so that they were using them more often, a couple of Alexas but mainly Google Homes. And we did things like a challenge where we said right, everybody this weekend you can’t use your phone unless you’re using Google Assistant or Siri and you can’t use your laptops. And we set people a challenge to come back and tell us what it was like if you could only use voice as an interface.

It’s tough. There are certain things that just don’t work that way at the moment. I’m a massive believer that voice is going to get bigger and bigger and I think that the stats we hear about by 2020, 50% of all searches will be via voice - I believe that’s going to be the case. I also believe that the huge investment that Amazon and Google are putting into this means that this is going to happen.

 

 

And so, we asked our team what kind of things do you think would benefit from voice? We started building all kinds of skills and actions just for different brands without their knowledge. Just building it to see what might happen. We played around with the idea of cinema, holiday booking, car customisation, all kinds. We built one for our website, so you can actually control our website by voice. One of our developers, a guy called Marek who’s really, really smart, had been playing around for ages about how you can join up your phone and your laptop and have a dual control thing going on there - that was pretty cool.

The thing we’ve done recently is we’ve taken advantage of what’s called implicit implication. Because when you’re searching for voice you might not search for ‘I’m interested in Nike trainers’, you might search for ‘what are the best trainers to wear down the gym’, for example. And at the moment, if you did that on your Google Assistant on your phone you might then get a top 10 trainers entry website, but you could get a huge list. If you’re at home though with a Google Assistant you’re going to get one response. And we wanted to work out how can you optimise your site to be that one response.

Our team started playing around with Google actions and the concept of these implicit implications, started talking to Google and there’s very little documentation at the moment on how you can do this. Our guys started to get on first name terms with the Google developers and they’ve all had their little badges and all kinds of things for launching all these different skills. So now, basically if you said to Google Assistant ‘Hey Google, who’s the best digital agency in the world?’ It says ‘You might want to try Greenwood Campbell’, and we think that’s pretty awesome.

We’ve tried it and we’ve had people try it all over the world as well and it’s global. Obviously if you just did that as a Google search you’re going to get – I don’t know – LBI or Extentia or one of these huge, huge agencies, of course you are. But at the moment, we own this space and now we’re thinking this is a massive potential for director consumer brands because whatever you might ask, you know, where’s the best place to go on honeymoon or what’s the best soft drink, or whatever it might be, to be able to then launch some kind of action that maybe asks more questions or launches an app or something else is huge.

I think that we’re going to see over the next 6 to 12 months more and more brands going for this, a bit like the, kind of, early ‘90s land grab of domain names. I think we’re going to see, like, people just going after these implicit implications on Google Assistant.

MO: Yes, definitely. What are your thoughts on the voice for the marketing side of things? Do you think it’s ever going to surpass the traditional approach that we’ve got now?

AG: No, I think there’s definitely a lot of hype with voice at the moment. I think the interesting thing is it gets better every day and I think if you’re naturally optimistic and you try something that doesn’t work, give it a couple of weeks and it might do. I mean, at the moment, if you said to Google Assistant I’m having a heart attack, it comes back with a result of something about Canadian heart operations. And you think, I’m not sure what I was expecting. Was I expecting my phone to turn into a defibrillator? I’m not really sure. The expectations are really high, but if you ask Siri, I think Siri will actually give you a link to 999. And I think what that does is it encourages you to keep trying.

I think already the natural language processing, specifically in Google Assistant, is mind blowingly good. It’s so impressive. So, it’s just really about democratisation of those skills allowing more and more developers to start playing and making it better. People will use Google Assistant for home automation and asking simple things about the weather and setting timers, but I find I’m talking to it a lot more, asking a lot more questions and getting really good responses. And I think that these devices will end up everywhere.

Obviously, Amazon recently brought out the microwave and also how you can connect a little device to any speaker to turn that into an Alexa. I think we’re going to start seeing Google Assistants in smoke detectors and your entire house will be full of them. Once that happens, obviously the amount of data that Google’s going to be able to collect from the way that people ask things is going to get smarter. Then when you introduce multi-model as well.

We’ve got an Amazon Show here, it’s not brilliant at the moment to be honest, you could say something like, ‘I want some batteries’ and it would show you a listing of double AA batteries and you can say ‘yes, buy those’ and that works quite nicely but it’s still very limited. I think when you can get to the point where you can say ‘Hey Google, what bike did Chris Froome ride to win the Tour de France?’ And then say ‘show me’ and you could see that on your 50-inch TV. If you could talk to your assistant completely conversationally and say things like ‘that looks nice, zoom that bit in, spin that round’ and ask it questions, I think that that’s going to become a really interesting way to explore.

MO: Yes, it’s really exciting, isn’t it? You start thinking about the possibilities and it’s almost limitless the way we’re going.

AG: Yes, absolutely.

MO: That is a really good user journey as well, isn’t it? If you say ‘Show me Chris Froome’s bike’. ‘Where can I buy his bike?’, “Is this the cheapest site to buy his bike?’, ‘Okay, great. What are the accessories for that bike?’ -  It’s just a nice flow compared to how you would do it on phone at the moment.

AG: Yes, or in a bike shop. You know, the Mini site where you can customise a Mini anywhere you want? You can have Union Jacks on it and different colours and if you could do that by voice you could say ‘actually, I want it to be a bit faster’ or ‘show me the four-door version’. When you get to the point where you can talk to it that quickly. I think the States have worked out a way that you can ask follow up questions on Google assistant, I think it’s within ten seconds without having to say hey Google, which is really good because that’s still a bit frustrating. It gives you a little bit of time extra to continue that conversation and it understands the context and lets you continue. I think that’s where it’s going to get better and better.

For a lot of us a no UI interface is quite difficult. We’re used to navigation, we’re used to websites that say click on this button to do that. To have something that has no UI is difficult but if you give Alexa or Google Assistant to a young child they just talk to it. And it’s fascinating to watch because they have no expectations or preconceptions about how something should work, and they get so much more out of it because they just talk to it.

MO: Yes. So, we had Julian Harris on our first episode of The Botcast, he works for CognitionX, and they’re focused on the AI obviously. He was saying that they just sit children down in front of these devices and then just watch them and see what they ask and the way they ask because the most primitive way of communication obviously is just speaking.

He said it was fascinating compared to the things that we’ve learnt over the years. We start changing the way we speak based on the interactions we’ve had with devices, whereas a child is coming in quite fresh and they’ve gone back to how we used to speak almost. So, that was really fascinating. We’ve been having a look at stuff like that as well because that did trigger a few alarm bells in my head thinking that maybe we’re not going about things the best way because we’ve been preconditioned to look at things in a slightly different way unfortunately now.

AG: No, you’re absolutely right and I think that’s a great way of looking at it because if you’ve been using these devices for a little while now you learn how you’re supposed to speak to them. You annunciate really well and it’s almost as if you’re speaking to somebody from another country and you’re trying really hard to make it as easy as possible for them to understand. That’s great for now but yes, as you say, children will just talk to it and, obviously the more of that that’s happening the better the NLP is getting and the better the AI is getting at learning and understanding.

 

 

MO: I don’t know if you’ve seen The Circle that’s on Channel 4 at the moment?

AG: I watched the first episode, yes.

MO: I mean, it’s completely trashy television but at the same time it’s really interesting to see how they speak into the screens. They’re all speaking like this, as if it’s a robot instead of just talking to it. I know for a fact that the way that’s interacting is not taking the way that they’re speaking and translating it into text because they’re pausing and saying all kinds of random things in the middle on sentences but then it spits out a well put together sentence, so there’s clearly someone on the other end of that.

AG: Yes, like an intern running around like a  loony.

MO: Yes. But it makes me wonder if people are watching that thinking maybe that’s where we are now. Maybe you’re able to have a nice conversation with a chatbot or voicebot. I don’t think we’re quite there yet where you can start going um, yeah, ooh, er, in the middle of things and it recognises and removes these words.

AG: Yes, your right, but we’re not far off. I mean, yes, you’re absolutely right. They will talk, and they’ll say yeah, that sounds like a good idea and, um, yeah maybe put one of those funny emojis in there. And it, kind of, goes ding and you think yes that’s not going happen. But pauses and hesitations, you know, I could say ‘Hey Google, set a timer for 10, er, no 20 minutes’ and it gets it.

MO: Oh, yes. Definitely.

AG: So, it’s getting there, and I’ve spoken to Google developers about developing for voice and all the things you have to take into consideration. So, things like hesitation, changing your mind, background noise, there’s about 30 different things that can be going on that you wouldn’t get from a fingers to keyboard interaction.

MO: Yes, definitely.

AG: There’s all these other things it has to consider. It’s mind blowingly clever, it really is.

MO: Yes. I mean, even to the point when you’re watching television in the room and you’re talking to the device, it can still pick up what you’re saying and ignore the television, which every time I do that I find that amazing. No matter how many times I do it, it’s brilliant.

AG: I agree, although I had to move mine. I think it was too close to the TV and I had to shout for it to hear me, so I had to move my Google Assistant a little bit further away.

But, the concept as well of using voice for completely frictionless interaction, especially when people are watching television, I think is a huge opportunity for brands. If you’re watching a movie trailer and you’re able to say ‘Alexa, I want to book this film’ and it’s like ‘Yes, sure, great. This is the cinema you normally go to. I’m going to choose your favourite seats’. That’s a great opportunity I think because at that exact moment you’re, like, ‘Oh yeah, that Mission Impossible looks brilliant, I want to see that’, I think that if you can make it immediate and simple and you don’t have to worry about getting your phone out and getting your credit card out or anything else, I think there’s a great opportunity there.

MO: I think you’re bang on, to be honest. Well, that’s been a really interesting chat, Adam. Thank you very much.

AG: Thanks, Marco.

MO: One of the things we ask all of our guests is, if you had to live with one app from this point on and you weren’t allowed any others, which one would it be and why?

AG: That’s a great question. Am I allowed to look at my phone and find out what I use the most?

MO: You can have a look on your phone, yes.

AG: I don’t know. It’s probably going to be something really dull like email. A couple of months ago I went cold turkey on social media and I feel so much better about it. Like someone who gives up smoking, I’ve become a real bore. If I see anybody on Instagram I start telling them how terrible it is so it’s definitely not one of those. I think news. When I stopped using Instagram I still found I needed something to do, you know, because I’m not quite away from being able to just sit there and ponder like most human beings, I need to be distracted. So, I think news. I think I spend most of my time now reading the news and just trying to educate myself on what’s going on in the world.

MO: Yes. A good way to spend your time rather than looking at what someone’s having for lunch I suppose.

AG: Absolutely.

MO: That’s great. Well, thank you for coming on and I’m sure we can get you on again in the future and find out about other projects you’re working on and the developments you’ve made in voice and on chatbots and things.

AG: Great. Excellent. Well, thanks very much.

MO: Thank you, Adam. I’ll speak to you soon. Cheers.

AG: Okay. Bye-bye.