Latest 15 Jun 2021 | Bots and Automation | 11 min read

Automated but not Robotic: 5 Bot Building Tips to Increase Efficiency and Customer Happiness

I recently spoke at The Customer Experience Conference and was able to share with the audience my thoughts on how the past 15 months has completely changed the way consumers interact with brands.  

Habits have shifted irreversibly since March 2020, with customers reinventing how, when and where they choose to engage and it's up to brands to adapt to meet these changing needs. 

To effectively deliver exceptional CX it's necessary to adapt automation across a customer lifecycle and in this presentation I offer the 5 top tips for building a bot to increase your efficiency and keep customers happy. 

  1. Make the bot contextual and specific 
  2. Keep the human touch
  3. Don't force the consumer down the wrong path
  4. Integrate to resolve in-channel
  5. Iterate and use errors as your intelligence

Scale your CX efforts with Conversocial’s Bot Platform

You can watch my presentation in full below or read the transcript.



Full Transcript below:

I'm Ido Bornstein-Hacohen, I'm the CEO of a Conversocial we're a customer engagement platform. We allow brands to connect with their consumers over messaging and social channels. And today, and we blend together automation and humans, so agents in the in the contact center today, I'm going to be focusing on a, how to build automation, and why to build automation. 

So when you think about what has happened in the last year and a half already, COVID has really changed the way that we interact and COVID has really changed the way that we and that we buy online from an e-commerce perspective, anyone, you look at it, across the board, we've seen that e-commerce as an example, that has really propelled itself six years into the into the future.

 Some even say 10 years in, into the future. And that has really changed the way that we interact, the things that we're expecting from brands. I think we all know today that when we buying something from somebody online, we're definitely going to have a conversation with them, or most are going to have some sort conversation with them at one point.

And we want that conversation to be seamless and simple. And that's one of the things that we really focused on. We run a survey every year for the last five years with 2000 consumers this year, we also added a hundred brand professionals, so executive within the brands. And we really look at the customer experience, set trends.

We've seen some really interesting things in the last in the last year. First of all, there's a real big shift in communications. 69% of consumers have said that they've used  a messaging channel or a social messaging channel such as WhatsApp, Facebook Messenger, Twitter DM, iMessage or ABC by by Apple or Google business messaging to engage with the brand, 69% did that last year. 

And 81% actually expect brand to provide them with this level of support over messaging channels. So there's still gap between the number of people who are actually expecting to receive a customer engagement over messaging channels, to be able to simply text a brand, whether it'd be on SMS or any of the other channels that I mentioned, and the people who have actually been able to do so. 

in the last year and we've seen that as well. We've seen an incredible growth from a private messaging, communication  standpoint. And it's pretty obvious that consumers are interesting and interested in having that way of communicating with the brand. Not only that, but, consumers actually expect brands to provide them with better digital customer engagement in the in the next year.

So 94% of consumers have higher expectations for the brand, for next year. But when it comes to automation, and we know that, automation can be very useful in having customer conversations from a social service perspective, just allowing the consumers to do things on their own.

50% of consumers say that they've had a bad bot experience in the past, but 69% say that if they had a good experience, they would actually gravitate towards it, if their answers were able to be supported faster, if they were able to get what they needed to get, they don't need to speak to anybody.

They're all fine. We're doing it through automation and when we look at where the brands are at today – 66% of brands saying that they're getting there only 24% say that they've actually, they're actually advanced in terms of the way that they're managing automation.

So today I want to take the next 10 minutes to focus on some things that we've learned over the last few years while working with brands on deploying  automation. There's five tips, five main things that we've learned: one is to make your bot more contextual and specific. Two is to keep the human touch.

Three, not to force the user down a certain path that you think is the right path and allow them to drive down their own path and support them while doing so. Four integrating and enable resolution in-channel. And lastly, iterate, make sure that you use errors, you use your intelligence to keep on improving the automation and the automation flow.

The first thing, and probably one of the most important things is making the bot contextual and specific. One of the things that we’re we've seen from a customer engagement standpoint is the simpler you make it, the  more you save people time. They're happier they're going to be, and the more efficient you're going to be as well.

And how do you do that? Do you do that by providing context? Think about having a conversation with somebody that, the more you can provide context into that conversation, the deeper you can go and the faster you can resolve things, right? You need to treat consumers as individuals go beyond what you're providing on the website.

Make it personalized. This is one of the things that conversational experiences allow you to do. To really turn it into a conversation, taking that extra step. If somebody is in a certain location and they're asking for store hours, give them the store hours of the store and the location that they are, that goes a really long way.

The second thing is really about avoiding deflecting to the general web. People usually get to have a conversation with you before, because they've already been on it general, but they couldn't find that the, what they needed. And that's why they're approaching you. Don't frustrate them by sending them back into a general place, giving information that is contextual and is, relevant to them.

Use your memory to do the the automation memory. The second thing is really about keeping the human touch. The fact that it's a bot doesn't mean that you should lose the human touch. Think of the bot as an extension of your of your team, right? Like we have conversational designers here internally in Conversocial and we really think about, we put a lot of effort into

thinking about how to design a book that really aligns with the brand narrative. There's an example of, as you can see at the, you can see here to the right, giving it a personality, giving it, the name really goes a long way in terms of having that conversation. Don't hide the fact that it's a bot,

Let the consumers know that it is right, but, give it that personality that really goes a long way. But allow the consumers to escalate into an agent, right? Don't block them from escalating into an agent. They'll find a way, if you try to block them, they'll find a way, if you allow them to actually transition to an agent seamlessly, then there's a good chance that they'll go down the path of automation and try to exhaust all options before they actually get into, ask for an agent.

But if they do make sure that you give them that and that option, so make sure to keep the , the human touch. We as consumers, when we go down a certain path, each of us has a bit of a different way of of engaging. Don't try to force the consumers down a certain path, try to avoid tree flows, where they're going from one step to the other.

If they fail, they go back to the to the beginning. There's one thing that we all need to realize is that there isn't really a back option within messaging. It's a conversation. And in the conversation we always move forward and that's a mind shift that you need to make when you're designing conversationaI experiences. It doesn't matter if they're automated or if they’re supported by human agents in the contact center. Realize that you always have to keep forward. And when it comes to a conversation and remember, Utilize the memory, you utilize the conversation, utilize information that you already have.

If you've already collected pieces of the information, there's only one thing that is, that needs to be changed. As an example, let's say that I'm in the process of booking a flight and then I just want to change the flight hour. Don't force me to go back to the beginning and give you my destination again, and the number of people and so on, allow me to just change that specific item while I'm conversing with the  bot lean on that memory, just in the same way that we're doing it, where it was linking.

To other people. Integrate, that's a very important piece. A lot of what i've  spoken about  so far really relies on the ability of the brand to provide self service in-channel. And what I mean by that is allowing people to actually  transact, allowing people to resolve things in-channel, then how do you do that? By integrating, connecting your front end, the conversational interface, into your backend system, to allow people to transact to allow people, to change things, to allow people to source, to find information, whether it be finding where their packages at, booking a new flight, changing ancillaries within their flight, or just getting general information, make sure that you connect your system into your CRM system, that drive all of the customer related information, but also into the operational system. Whether it be first part, the API system, payment systems, industry platform systems, and so on, and allow your consumers to actually engage with those systems directly from within and the messaging channel.

And the last piece, which is probably one of the most important the pieces, is to iterate and to use errors as your intelligence, right? What do I mean by that? And look at what your consumers are saying, where they're getting stuck, what are the phrases that they're using.

Utilize that data to really get started with your first bot. So the data about what consumers are actually talking about right now, whether it be on, another channel such as phone and email and live chat and so on. So utilize that to get started with that conversation.

But once you build the initial conversation, make sure that you collect all of those fail points. In all of the success bonds, so you can constantly improve and iterate on it. So when you think about it, bring it all together. From understanding the user, using a natural language understanding to connecting to the backend system through the API connection, to utilizing memory and making things, contacting things like previous orders, name and locations and so on. 

That’s bringing all of those things together actually creates for that exceptional conversation experience. And how to  get started? One of the things that we recommended before you start building is to start with a design, understand the conversational flows. 

What are the experiences that you're trying to a drive for? How do you blend together the agents and the chatbots from there? Start building. Start with the first bot flow. And then, once you build launch and start iterating and improving the actual bot flow.

We recommend a three phased approach. And for those of you who have not yet really started with automation, you don't necessarily have to start from the first day when you know, a full robust automation flow you can start small with simple, transactional processes. It could be a simple FAQ. Doesn't necessarily even have to have NLP in the first  stages.

And it doesn't have to be connected to all of the backend systems that we've talked about before. It could just be something that, will help you test the water and help get your customers into a place where they feel more comfortable utilizing, automation to get the response for their requests.

From there, going to a level two. Right level two will allow you to actually go deeper when it comes to the transactional processes. So select one or two processes, go deep, integrate into a backend systems, move to a more rich, conversational experience. 

Add NLU on top of it, but really focus on one or two experiences and one or two channels to make sure that you're going down the right path. Once you've done that, then you see that this is all working well, start running. Round that the customer experience integrate into more business systems, add more channels proactively promoted.

This is where it becomes really interesting because when you can practically promote this is when you're going to see a lot of the efficiencies as a brand, but also your consumers are going to be happier by being able to self serve and transact on their own without having to wait online, within an asynchronous environment. 

And this is an example of one of our customers Volaris an airline out of Mexico. When they started with us, it was mostly on public social media. They launched a private connection to public social media. So they launched FAQ on Messenger.

And then on WhatsApp, that was the first stage. They then enabled one specific API information gathering, to find a flight status. They then started providing things like ticketing and price quotes via dynamic claims. So they actually went deeper and deeper when they started feeling comfortable with that

they've opened things up even further, and they allowed people to check for existing bookings. And now they're actually allowing them to almost do everything that they need within  messaging that allows them on specifically on WhatsApp that actually allows them to actually take down.

The the call center and focus on on providing customer engagement, mainly through WhatsApp. 

The last piece that I wanted to share is really a, about the company called Freshly. It's a meal company here in the United States. And they started with a customer service use case, and then they actually expanded into acquisition and information gathering use cases as well.

They saw a containment rate of 50%. So 50% of people actually resolve their issues without having to talk to an agent. 

Just to recap; make your bot contextual and specific, keep the human touch, don't force the user down your path, integrate and enable resolution in channel and iterate and use errors in your intelligence. These are the things that we've learned throughout the years.

And we really recommend that everybody to take those under consideration. And the report that I presented at the beginning of the presentation is available through this QR code or through our website. We have a one minute to go before we have to conclude, I can see that there is one question, eh, by Dominic who's asking would crawling not imply a bad customer experience, doesn't help me.

We'll not use it, eh, we'll not use it again. and that’s exactly why we feel when you get started, doesn't matter, by the way, if you start with FAQ's or if you start with real transactional processes, you need to provide value to your consumers.

So that value, even if it's in many cases, it just makes it simpler.It makes it simpler for you to find that information. So focus on those things, even if you're not allowed to transact from the first get-go focused on things that are actually providing them with value rather than.

Sending them back to the website, sending them back to the phone, provide some level of value. Then make sure that you are actually collecting and providing that information to your team to keep on enhancing the And the bot, right? Eh, one of the things that we've that we've seen is that although there is some concerns sometimes if I've had one bad experience, I'm never going to go back and use the the bot while we have realized that what we've seen is that consumers do continue testing it.

They do go back to it, even if they've had, one challenging experience. They're not going to jump right outside of it. As long as you are able to provide continuous improvement, it's almost when people go to your website, and they go in and they look at it every once in a while, this has the opportunity to actually change that.

Be part of The Conversation

Get no-nonsense best practices & insight, to help shape your digital customer care journey, delivered straight to your inbox.