Site Search

How Smart Customer Service Drives Loyalty

Want to avoid interrogating your customers?

Smart customer service can drive customer loyalty and shareholder value. So find out how to take advantage of new technologies that will help your company increase its competitive advantage by helping customers help themselves with less friction and more protection.

  • Optimize self-service for better customer experiences
  • Enhance security by replacing knowledge-based authentication with a pre-answer authentication approach
  • Increase IVR self-service rates and reduce agents' average handle time

Video Transcript

Hello, everyone. And welcome to today's CRM Magazine Web Event brought to you by TRUSTID-Neustar and Attivio. My name is Bob Fernekees. And I'm the publisher of CRM Magazine. I'm going to be the moderator for today's broadcast.

Now our presentation is titled Smart Customer Service. But before we start, I just want to explain how you can participate in this live broadcast. At the end of the event, we'll have a question and answer session. So if you have a question during the broadcast, either one of the presentations, just type it into the question box provided, click on the Submit button, and we'll get to your question during the broadcast in the last 10 minutes or so of the presentation.

So if for some reason we can't answer your particular question, don't worry, we'll follow up within a few days by email. Plus, if you'd like a copy of the presentations, you can download them as a PDF from the resource icon on the Console once the event is archived.

Now to introduce our speakers for today. We've got Lance Hood-- he's the senior director of product marketing at TRUSTID-Neustar-- and Brian Flynn, VP Solutions Architect at Attivio. If you want to review the speakers' bios, you can just click on the arrows under their photos and read all about them.

So now I'm going to pass the event over, again, to Lance Hood from TRUSTID-Neustar. Welcome to the broadcast, Lance.

TRUSTID Inbound Authentication Solution

Thank you, very much, Bob. It's a pleasure to be here and a pleasure to speak to our audience today. Just to kick things off, I'll give you a brief overview of Neustar and then talk a little bit about the TRUSTID solution.

Neustar is an identity services company. We have a tremendous amount of identity information about people, the places they live and work, and the various ways that they communicate. And we use that information to deliver communication services to businesses, everything from how to help you know what time of day to call someone at and what number to use to make that communication, whether that's in support of your selling or in support of marketing activities or customer service or collections efforts.

We also provide a service for your inbound calls. And specifically, it is an inbound call authentication service. And that service was acquired by Neustar-acquired TRUSTID-- which is an independent company that became a part of Neustar in January. And that'll be the focus. This inbound authentication service will be the focus of my discussion today.

Importance of Call Centers

And let's start by just highlighting, number one, why call centers are so important. This is a table that shows the preferred communication channels by generational group. And over on the right hand side, highlighted, you can see that the preferred communication channel for three of the four major generations that exist in the marketplace today is a telephone, the call center.

And overlaid against that on the right hand side is the percentage of wealth within the country and how that is distributed. And what you can see is that for the people that have the most purchasing power and the most investment potential, their preferred communication channel with businesses, whether it's in financial services or health care or travel or retail, is the telephone. In fact, the only generation that doesn't yet show up is Generation Y, which you might also think of as the Millennials, who don't yet prefer the telephone channel as much. But we will see over the next few decades, as their percentage of wealth, their purchasing power increases, the complexity of products they're buying and the complexity of investments they're making increases, they probably, too, will begin to favor the phone channel.

Preventing Call Center Fraud Remains a Challenge

So call centers are really important. But for many organizations that need to identify who's calling in order to deliver services, this is the common experience. We don't answer the phone saying, thank you, for calling, how can I help you? We generally answer the phone call by saying, who are you? Actually, I don't believe who you just told me you are. I'm going to interrogate you now and ask you four or five questions, so I can have confidence I'm really talking to the person you claim to be.

Now the irony of that is that that process really doesn't work well at all. Forrester conducted a survey last year that was sponsored by Neustar and found that 73% of fraud instances involve personally identifiable information. So all of this process that we're taking people through where we believe or try to believe that only the real customer can have the answers to these questions doesn't really work. It's actually a great opportunity for fraudsters to take over accounts and gain access to sensitive information or even financial assets.

And the problem isn't getting any better. The reason this doesn't work is there have been so many data hacks over the past few years. In 2018 alone, there are over 1,200 published data attacks. That doesn't include, obviously, the hacks that organizations aren't even aware of that have occurred. And from those hacks, almost half a billion customer records were exposed, more than double the amount of exposure that occurred in 2017.

So more and more personal information is floating around out there for the bad guys to get. And this is just the information that's available through data hacks. The citizens have been exposing their information through social media sites, as well, for many years. That information is never going back in the can.

So through both data hacks and the proclivity of people to just share their own information, there's just a tremendous amount of answers to all of the knowledge-based questions that we all face when we call our bank or other institution. So it's not a surprise when Neustar did a survey in January of this year-- and we did the same survey in January of the year before-- that call center operators, the people who were in operational roles, customer service roles, fraud detection roles, are progressively getting less and less happy with their current authentication approaches.

Last year, about 49% were very or somewhat dissatisfied with their method of authenticating callers. This year, that jumped to 63%. So we have a system in place in which no one is really happy with it. Customers have a lot of friction. And call center operators are frustrated with it as well.

So you might ask the question, how did we end up in a situation like this, where we take our one of our most important business assets, our customers, and we treat them almost as if they were criminals? So let's look back in time a little bit. If you ran a call center, let's say, in the 1960s or 70s, or even into the 1990s, you could be really confident that when a call connected, that the phone number that that call was claimed to come from-- often called an ANI. That ANI, that phone number, was accurate 100% of the time. And based on that, you could quickly identify who was calling and immediately go into delivering service.

Unexpected Consequences of Easy Access to Data

That changed in around 2003 when the global phone system connected to the internet. And that brought a lot of great things to the phone system-- better quality, greater bandwidth, new features, data services, all sorts of great stuff-- that we almost, today, take for granted.

But one of the unforeseen consequences was-- of that connection-- is that a bad guy, a fraudster, could download some software from the internet, connect with his computer into the telephone system, and spoof-- which is basically means make phone calls and change the claim phone number.

So all of a sudden, all of this infrastructure had been developed. And processes that allowed for easy identification of callers based on their phone number went away. And that's where the rise of this new form of knowledge-based authentication came from to start asking people questions. And then there were data hacks. And now we need to ask more complex questions. And we need to ask more of them. All of which has resulted in a terrible customer experience, increased call center costs-- because we're paying people to ask those questions instead of deliver service-- and in the end, it's poor security, anyway. It doesn't work. Because the bad guys have the answers to those questions.

And spoofing was really the problem in the beginning. But fraudsters are clever. They are now also using virtualized call services to call call centers. This is a great tool for the bad guys. Because they can be anywhere in the world, sit down at a computer, dial in and use services like Skype or Google Project Fi, and make phone calls in the call centers that are completely legitimate calls. But they're untraceable. They can be anywhere in the world.

And what fraudsters are really good at is social engineering. So they love to do both spoofing and use virtual call services to reach your call center agents and, essentially, start that process of convincing them they are one of your customers when, in fact, they are not.

So how do we solve this problem? Well, ultimately, the problem is, how do we better authenticate people?

Three Primary Forms of Authentication

There are only three ways to authenticate someone, whether you're talking about a web experience, a mobile experience, or in the call center.


And the three categories are Knowledge. It's something that, in theory, someone should uniquely know. We already know that doesn't work in the call center. Because the answers to the questions that we can ask are often known by the bad guys.


The second category is called Inherence. And that's something that is uniquely associated with someone, from a physical standpoint or a behavioral standpoint.

So an example of that is your fingerprint. In the context of a call center, it's a voice biometric voice print and the ability to have a customer store a voice print and then, subsequently, when they call in, compare their voice to that print and use that to authenticate them. Very similar to if you stored a fingerprint and then, as you use your cell phone, you authenticate that way.


The last category is Ownership. Ownership is probably one of the oldest forms of authentication. A physical key is an ownership factor in authentication. If it goes in the lock and it correctly turns the lock, you're granted access. Probably the most successful example of ownership factor authentication has been credit cards. Credit cards and that physical card has really transformed retail commerce over the last 40 and 50 years. Well, it turns out that phones, which is a physical device, can also be a very powerful way to authenticate someone.

And that's what we're going to be talking about for the rest of my section today is how to use phones as an ownership factor of authentication, comparable to how credit cards are used in retail commerce. And using this very powerful, very accurate way to authenticate someone can reduce or eliminate your use of knowledge-based authentication. And by the way, it can be paired very nicely with a voice bio authentication system to create a multi-factor authentication approach with two strong ways to authenticate that don't rely on knowledge at all.

Authentication: The Credit Card Analogy

So to explain this a little bit, I want to use the credit card model as an example so you really can grasp what is going on when I say, using phones for ownership authentication. So if you use a credit card-- I went to Home Depot over the weekend and had to buy a piece of wood. And when I go into Home Depot, and when it's ready to check out, I insert my credit card into a reader. And that initiates a process by which the credit card network interrogates that card, assures it is a valid physical card, that it is the correct card to be tied to the account number associated with that card, and finally, that it's actually in a transaction at the appropriate physical location. And if all those things check out, then an authentication token is sent to Home Depot. And I get to complete my transaction and leave with my goods.

The same process takes place within the system that Neustar delivers for phone-based ownership factor authentication with a call center. In this case, what we're doing is we sit inside the phone network. And we will physically inspect that phone to ensure that it is the right phone that is associated with the claimed account number, which is, in this case, the phone number. And we can do that for both a mobile phone, and we can do that for a residential phone.

In the case of a mobile phone, we're looking at the phone as well as the SIM card that's in the phone. For a residential call, we're looking at the copper connection point for where a phone is, actually, out on the street connects into a household. And further, just as the credit card network is looking at the physical location of a transaction, we're going to look at the routing of that call from the phone all the way through the phone network into a call center.

Power of Device Authentication

And when all of those boxes are checked, you have an incredibly accurate way of saying this is a real call. It's coming from the phone it's supposed to be coming from. And if they match that to your customer records, you have an incredibly accurate way to authenticate someone.

And the beauty of all this is it actually takes place before the call is even answered. So when a call center agent picks up the phone, or when someone goes into their IVR system, they're already authenticated and identified.

And essentially, in a nutshell, what we're doing is making sure that that phone is from a device that is unique, it's an authentic device, that it's physical, just like a credit card is, and it's not been virtualized or spoofed or hacked or manipulated in any way. It's also important to note that we do this for 100% of the carriers.

In North America alone, for example, there are 4,200 different phone companies that can originate calls into a call center. And because we sit inside the phone network, our technology is able to inspect all of these calls. There is no dependence or any variation in quality or accuracy between the carriers that either originate the call or that deliver the call into a call center. And we can also inspect calls of all line types, whether it's a mobile phone, or a phone for a voice over IP call, from, like, Comcast, for example, or other types of residential phones.

Value of Low Friction Authentication

So ultimately, this puts us in position to deliver great value to our customers. And the value equation really is, I have a high quality trust that I have the right caller on the phone. I can match that caller to my CRM database. And that allows me to deliver a lot of value within an organization.

So what kind of value? Well, again, what we're doing is we're looking at the call as it comes in, pre-answer, typically within three or four seconds, half of a ring tone, we will complete that process. The customers become really part of what we'll call a trusted caller flow. And that trusted caller flow can have a different routing and different customer experience.

Typically for our customers, they're going to reduce the amount of knowledge-based authentication. If they're asking five questions now, maybe in the future, they'll ask just one question. And the results that they typically see is about a 20% reduction in your call center operating costs. Because you have agents helping people, now, instead of spending 30 seconds, a minute, a minute and a half, identity-interrogating someone.

A number of customers have done customer satisfaction studies. And we typically see about a 10% increase in a CSAT rating, based on just the fact that they authenticated more seamlessly through the TRUSTID-Neustar system. We also see increases in IVR containment and more self-service, which, of course, is a benefit to the customer that can self service. And it also reduces the amount of agent time spent in delivering services. Because they don't have to take those calls.

And that's driven by two factors. One is that the actual time spent authenticating is simplified, so people can stay in the IVR. And our customers will also add new functions to the IVR-- riskier features, for example-- that they wouldn't be able to offer to their customers unless they're highly trusted.

Benefits of Segmenting High Risk Calls

And the other benefit of this is that now you can focus-- if you are concerned about fraud fighting or other issues with potential criminals, you can focus your fraud fighting efforts on the subset of callers that were not authenticated. So it improves the efficiency of fraud fighting as well.

The other thing I wanted to note is in that value equation, accuracy of identifying trusted callers is the first step. The second step is being able to, now, link that to my CRM records. And therein lies the challenge. I may have a trusted caller. But they're calling from their cell phone. I only have the residential number on file. So how do we deliver value in that case?

And that's where Neustar has an integrated product called AccountLink that allows for matching to customer records even if you don't have the calling phone number on file. So let me just quickly cover how that works.

The first step in that, again, is that the caller comes in, the ownership factor authentication process completes, and a credential is delivered by Neustar, there isn't a match in the database. But what Neustar has the ability to do is, for example, say, well, this phone number in the phone network is associated with this person, Joe Smith, who lives at this address. And guess what? That is a match to your database. And I also know that he has this cell phone number. So based on that logic, we can then say, that is, in fact, Joe Smith calling, and still deliver a superior experience to that by reducing the additional authentication effort that's required.

Now the question, of course, is, when in all of this should we be authenticating? In the survey that we did earlier this year, and we did the prior year, as I already mentioned, we asked, when is the preferred time to authenticate a caller? And a couple of things jump out.

When you look at the results comparing year to year, both in the 2018 version and 2019 version of the survey, we found that very few people preferred to have agents doing authentication. What we've seen is a real shift in the preference in the last year.

Pre-answer Authentication is Growing

In the 2018 version of the survey, respondents-- who, again, are customer call center managers, customer experience people, IT people that support call centers-- found that about 38%, 39%-- so almost an equal balance between-- authenticate before the call is over and authenticate during use of the IVR. This year, a big swing toward pre-answer authentication.

So that's one of the things that I think makes the solution that Neustar offers so valuable is it completes this process before the phone is even picked up. Another advantage of this service is that it's fully functional on day one. The day you turn it on, we will be processing 100% of the calls that come in.

Some other technologies that are used, for example, voice bio, require an enrollment process that's often fairly onerous. Because before you store someone's voice print, you want to make absolutely sure you're not storing a criminal's voice print. And that leads to a very intense enrollment process with a lot of knowledge-based authentication questions, which is onerous and results in a lot of people dropping out of that process even if they'd like to use that technology. That's not a problem for our particular solution.

So let me just kind of wrap this up and, again, focus on what are the key benefits of improving your authentication process with a pre-answer solution. First of all, there is a 20% cost reduction that most of our customers see by reducing the time that agents are spending authenticating. And that's typically going to save you $0.50 per call and 30 to 120 seconds.

Earlier Identification Improves Customer Experience

The second is it's going to be a great customer experience. Instead of interrogating your best customers, you're now answering the phone saying, what's your name? Oh, great. How can I help you? You're getting right to service. And also, we tend to focus a lot on the improvement in the agent experience. But the reality is it's a great experience for agents as well.

And finally, as I mentioned, it increases your fraud fighting efforts. You can get much more focused on the non TRUSTID callers. So thank you, for your time today. And I'm going to hand it back to Bob to continue today's webinar.

Hey, great. Thanks, so much, Lance. That was really interesting. I didn't realize 2003 was such a pivotal year in fraud detection or prevalence of fraud. So anyway, now I'm going to hand things over, again, to Brian Flynn. He's the vice president of solutions architecture at Attivio. Welcome to the broadcast, Brian.

Great. Thanks, Bob. Thanks for having me. And thanks, everyone, for joining today. Some really interesting conversation, so far, around how we can really help improve the customer service experience.

I think, with Attivio, one of the things that we have a pedigree in, which, we've done over the last 12 years is how do we really leverage artificial intelligence for the purposes of answering the tough questions, right? Maybe things as simple as, how do I reset my router, to some things that are more complex, like, is your particular modem or brand of particular modem compatible with the type of band that we have for your cable environment in your particular house? Things that I'm sure we've all had to ask ourselves. Or maybe, customer support in the past that might take an hour, it might take a day, it might take a week to get responses back.

Natural Language and Artificial Intelligence

And so, really, what we're trying to talk about today and showcase is what are we doing with artificial intelligence to help improve the self-service experience, enabling customers to answer their own questions? What are we doing to help improve the call center experience, making sure that agents, whether they're level 1 support, the first line of defense, or maybe even kind of the escalation engineers, so to speak, have the information at their fingertips, have the answers that they need to answer those very complex questions, in a timely manner, in a way that helps to improve all of the key KPI that I'm sure we all care about?

Meantime, the resolution, customer satisfaction rates. So that's really we're going to focus on today from Attivio's perspective is leveraging Attivio's artificial intelligence to provide the most relevant answer to any end user, whoever they may be, customer, agent, or otherwise. And why are we doing this?

So from our perspective-- and it's not just us-- but, you know, there's a lot of reports that exist in the wild today, this just being one of them, where customer experience becomes the key differentiator, right? I'm sure we all have the sense of urgency around making sure that our customer feels like they're the only customer that we care about, that we all want to scale, we all want to grow with our customer base.

But in treating every individual customer like they're the only one that mattered and they're always the one that's correct is very critical. And more and more, we're seeing across the board here, that ensuring customer experience is the most high priority requirement for our customers is going to be very important for not just any differentiation we want to provide, but the key differentiator for all of our goals for improving our customer experience.

Improving Customer Experience Helps Lower Churn

And there's a cost associated with this, too, right? Poor customer service causes churn, obviously. But one of the key trends that we're seeing in the marketplace based on some of the analyst reports that we sign up for is, what amount of money are we seeing customer churn, poor customer service costing? And you can put a number on that, $75 billion in defections each year, whether that's in high tech manufacturing, whether that is in customer service around telecommunications.

We have seen a lot of customers that have come through Attivio with these types of problems that they're trying to stem. How do I ensure that my customer, again, regardless of how they want to ask the question, can ultimately get the answer that they're looking for?

And certainly, that kind of speaks to a variety of different pain points that you're ultimately trying to solve in order to help enable those customers to be better satisfied with the answers to these complex questions. First and foremost, how do I ensure that the customer even knows how to get an answer to their question?

Maybe as a mature enterprise, I have a website for customer support, I have a website for forms of communications where customers talk amongst themselves about my various products. Maybe I have a marketing based website. Maybe I have a ticketing based website.

There's four channels, right there, all internet based, all browser based, from your mobile device or laptop, that you might be influencing a customer to try and go and answer their question. And they don't even know where to begin. So those are four different places that you might be sending them to. And right off the bat, they might be confused as to what to try and look for and how to ask the right question.

The next piece is once they've even found the place to ask the question, how do you ensure you're getting them the right answer? Again, we've just rattled off four different places where you might be sending your customer today, to try and find the correct answer. But how can we even be confident of the fact that the question they're asking is going to have the correct answer associated with them? That can be very complex types of questions that we've talked through already in terms of, how do I fix this particular issue with my semiconductor device that I'm currently trying to install at this new power plant, as an example?

Those are very complex questions that, even if you know where you should be asking that question, getting the most relevant, the most appropriate next best action, the most appropriate manual that's going to provide the relevant answer to your question can be very complex. And then the last piece here is, how do we ensure that what is relevant today also improves to be what's relevant tomorrow or next week or next month?

As we talk about these various ways that end users, customers are trying to find the answers to their questions. Certainly, your enterprises are growing as well. The different types of products, the different types of services, the different types of solutions that you might be providing to your end users today are going to change drastically over the next year or the next five years, even.

And with Attivio, we want to ensure that-- you know, ensuring that Attivio can provide the most relevant answer every single time is going to be based on the considerations around what are those end users searching for, how are they clicking on the most relevant answer to their question, and taking that information into account to improve what is the most relevant answer next week, next month, and next year.

So these are the types of pain points that, for our customers and for the folks that we talk to on a regular basis, they have these similar types of issues-- siloed answers to questions providing the most relevant answer to that question, regardless of where the answer might live, and then ensuring that you have a system in place that is constantly learning from all of those different end user interactions, regardless if they're end users who are customers or end users who are support agents in a call center.

Can Artificial Intelligence Help with Customer Calls?

So really what we want to talk about is, you know, those are kind of the key problems Attivio was meant to solve, right? We do that today with many customers, which we'll talk about in a moment. But we want to spend a little bit of time of really unpacking, what do we mean when we say Artificial Intelligence? It's a very high level buzz word, as I'm sure we've all kind of experienced in the past, today. And with Attivio, we really try and get down to the nuts and bolts of what is explainable AI, how do we actually leverage various artificial intelligence concepts in the manner of ensuring that we can meet these key KPIs around reducing the mean time to resolution, improving customer satisfaction by providing the most relevant answer to any end user, regardless of who they might be?

From our perspective, there's really Eight Critical Requirements that kind of speak to how you can start to solve these types of problems with the out-of-the-box capabilities of the Attivio product. I think I like to show this slide mostly just to show the breadth of the different types of conversations we tend to have-- where can you run the Attivio software? What are the types of data sources you can connect to? Websites, again, customer forums, maybe ticketing systems, as well.

Unfortunately, today, we don't have time to cover all of this. Certainly happy to have any follow up questions as part of today's conversation on these particular aspects. But I think for today's conversation, the real key pieces we wanted to focus on are, how do we really leverage artificial intelligence to deliver smart customer service? What are the key components that Attivio does provide that really speak to these types of tough problems that our customers and our prospective customers are trying to solve? And it's really going to boil down to these remaining requirements that we've highlighted. And we'll kind of unpack these as we start to walk through today's conversation.

Smarter Knowledge Base of Answers to Questions

So one of the key capabilities of the platform is, again, we've talked about data silos, we've talked about finding the most relevant answer to a question. But I think, from our customer's perspective, they're really demanding the Attivio platform ensures that we can enrich their data sets as we start to provide a kind of knowledge base of answers to questions.

When we say Enrich, that means a lot of different things. When we're connecting to things like complex manuals that have a lot of different troubleshooting tips, a lot of different key considerations around setups, around installation procedures, there is a lot of different key entities that might be mentioned in those documentations, entities of things like company names of things like error codes, products, and services. Those are all types of enrichment metadata that we want to automatically identify and extract so that you can start to provide the most relevant end user experience to say, you know, show me the manual that is actually referencing this particular error code that I have a question about.

With Attivio, you can point us to your repository of user manuals, regardless if they're one page installation procedures or if they're 300 page full binders that have been scanned through in a non digital-friendly format. We can take access to that content and unpack the information that exists within it, and then enrich that data to understand what are those key pieces of information that are described-- the key error codes, the key pieces of software, doing that across 45 different languages-- so not just English, but also other languages, like Spanish, French, Chinese, Japanese, Korean-- so you really truly have a global platform for understanding, what are the total sum of answers that can be provided to these end users, what are the key concepts that I mentioned within them, as well, and what other types of end user experiences we can provide to ensure, again, that end user, that customer, that customer support, or that call center agent can quickly find the most relevant answer to their question? And all of this is using a Attivio's out-of-the-box machine learning capabilities.

Leverage Structured and Unstructured Data Sets

In addition to that, with Attivio, you have the ability to start to build up very, very tailored end user experiences that can start to join together not just some of these unstructured pieces of information that might exist within your enterprise, but maybe, also, structured information as well, maybe information that lives in a database, or maybe information that lives within a CRM, like a Salesforce, for example, with the idea being that, wouldn't it be great if we can not only understand what are the key concepts that are mentioned in these unstructured documents, but also, how do they relate to this very high value structured information that we start to maintain internally?

I like to show this slide. Because it's a good example of how you can start to unify and really provide a 360 view across structured and unstructured data sets to answer these types of complex questions. What are customers in my region saying about the products that they've purchased? The things that those customers are saying-- those are, again, going to come from unstructured documents, whether they're manuals, or maybe, whether they're interactions that are happening on those customer forums. Are you having a good or bad experience with this product that you've purchased?

The nice thing about the Attivio product is that you can, again, access those unstructured information, start to unpack the key pieces of information that exist within it, and then, associate that back over to the structured information that might live in the CRM, the things that those particular customers have purchased over the last 90 days, to provide these types of very relevant end user experiences that you see along the right hand side, ultimately answering the question, is a positive or negative experience with this particular product impacting our bottom line positively or negatively?

So again, it really starts to speak to the ability to enrich these unstructured data sets, whether they're communications, whether they are manuals of information and then joining them together to the other high value structured data set you might already have internally. Another key capability of the platform is leveraging the security that's inherently embedded in every single thing that the product does do of Attivio. So ensuring that when end users are searching for content, when they are looking for content as a customer of your enterprise, or maybe as a call center agent, or maybe as a manager of that call center agent, those three very distinct personas only have access to see the answers they should see in the first place, right?

It sounds like a very simple thing to say. But it becomes very complex when you start to think about these different places where end users might have to ask questions and get answers. They might have very different types of security or requirements based on where those answers might live in the first place. And with Attivio, you not only have the ability to ensure that we can honor those permissions-- again, making sure that when Brian Flynn is doing a search for this particular manual or this particular problem we have with this particular customer, I should only see the answer if I should have access to see that document in the first place.

And the same type of requirements should be applied to me, my co-workers, and any of my employees or employers up and down the chain. Those are all critical requirements that we ensure we can maintain as part of any implementation of Attivio.

Customer Service Can Improve with Machine Learning

And then, the last piece here, around some of these critical requirements that you should consider when you're trying to understand, how do we improve the experience of finding the right answer for those different types of end users, is leveraging machine learning for providing the most relevant answers to those questions. Every time an end user is asking a question, every time an end user is being provided an answer that they're interacting with, whether they're clicking on that result, that answer you're providing to them-- maybe they're commenting on it, maybe they're rating it, they're giving it a thumbs up or a thumbs down-- those are all key signals that Attivio is inherently equipped to understand. Meaning when end users are searching for answers and they're actually interacting with those answers, we can feed those signals back into the product experience to ensure that whatever the key concepts that exist around those answers they're interacting with are going to be things that influence the next time that end user or a type of end user that looks very similar is asking a similar question.

This speaks to the problem we were trying to relay earlier. You can have a great solution for providing answers to questions that really do break down those silos, that can search across customer forums, that can search across file systems, maybe, or even public facing documentation. But if you don't have a system in place that's constantly learning about how those end users are searching for content, or what types of new products and services you might be providing, it's only going to be relevant for a certain period of time without a significant amount of heavy lifting.

With Attivio, we inject machine learning directly into that process so that whatever is relevant today may not be what's relevant tomorrow. But we want to understand that and provide the most relevant answer every time.

Just a couple of case studies about customers that are leveraging these types of experiences today for customer support experiences. We have a major aircraft manufacturer who had the challenge of understanding, what is the number of global Aircraft On Ground events that are occurring? And then, what are the correlated call notes, case notes, engineering notes that are associated with these types of events that we want to understand the patterns that exist across those particular AOG events, as they like to call them.

So the idea here is that they have all these call notes, they have all these engineering notes. They know when aircrafts are on ground. But there was no real way to analyze and understand what are the detection of patterns that we're seeing across these particular hotspots? Maybe it's this particular type of product that they provide is causing this particular manufacturing error over time, these particular types of error codes.

With Attivio, we actually had the ability to access and, essentially, index all of that information, enrich those data sets, again, identify the key facts that exist within those particular call notes, engineering notes, to ensure that there is a repository of knowledge across both structured and the unstructured information-- again, those call notes, those engineering notes-- to ensure that, over time, they, as an enterprise, start to learn, what are the things we need to be paying attention to, and really start to get proactive about around supporting these particular types of very complex pieces of machinery to the tune of actually implementing Attivio and moving from the global customer service rankings of number four in the globe to number two within a year of implementation. Very powerful stuff there.

Leveraging Potential Customer Insights Across CRMs

Another case study that we like to talk about is with one of our close partners, ThermoFisher, where they actually use Attivio to unify a lot of different information that existed across five different customer relationship management systems, not just Salesforce, but also, Siebel, and a lot of other internal custom bespoke database-driven CRM systems that they built.

In a world without Attivio, they had to actually, again, go to those different places to try and find, you know, how do I upsell this particular customer who just wants to get a new line of beakers in place, because they are 30 years out of warranty or 30 years out of date? With Attivio, you actually have the ability to access all of those CRMs at their point of data, pull that into a single instance of Attivio, and then, again, enable customer support, enable sales representatives to ask a very simple question-- are there any types of warranties that are out of band or upsell opportunities that I should be identifying and getting the most relevant answers so that the customer support experience goes up?

These particular customers of ThermoFisher understand, oh, that's a good point. I actually didn't realize I was under warranty already. Let's go ahead for that free re-up and then, maybe, expand into other business units as well. So that's the idea with ThermoFisher is connecting to these data sources and ensuring that they have the most relevant answer, no matter who the end user ends up being.

And here's the value impact that we provide-- again, really ensuring that you're empowering those end users to solve those issues on their own, whether it is those call center agents, whether it is those end users that are trying to answer their own questions without having to pick up the phone at all in the first place, we've been able to improve deflection rates by the tune of 5 to 10X. Really powerful data points there, ensuring that customers can really quickly find the answers to their questions.

Optimizing the agent workflow. So in those scenarios where maybe there is no knowledge or no answer that's most relevant to this particular customer, because it hasn't been encountered before, let me try and find the agent that can actually help answer my question on my behalf. We can provide these same experiences to customers as we can provide to internal incident managers or incident fulfillers or call center agents. Whoever they might be, internally, they want to have the same type of experience. Let me quickly do a keyword search or a natural language query that's going to provide the most relevant answer to my question.

And then, lastly, the ability to actually unify these interactions and understand the intent of them at any particular point in time, now and in the future, to the tune of improving customer satisfaction by a score of 2 to 5 X. We've seen a lot of customers that have deployed these types of end user experiences across, again, the customer-facing interactions, across internal agents interactions, and then, be able to measure these key customer benefits very quickly, very easily, and very effectively to help justify not just their ROI with Attivio, but help justify the benefits of what customer satisfaction can do to improve their bottom line and keep that churn rate down and keep their customer base growing.

And the last piece-- I'm going to skip ahead here. The last piece I want to cover off on is we are a recognized leader in this space, particular on cognitive search. Very recently, Forrester has actually published out their most recent cognitive search wave where we are recognized as one of the leaders for these types of requirements that you might have-- custom and complex cognitive search applications really focused on whether it's customer support, whether it's risk avoidance, maybe it's unified digital workplace. With Attivio, we have a focus on many different solution areas. And today, we're just talking about smart customer service. But there's a lot of different capabilities that Attivio does provide. And I'm happy to kind of help answer any questions as a follow up to this conversation today.

That's it for my presentation. I just want to thank everyone for listening and joining. And I'll pass it back over to Bob, at this point.

Hey, great. Thanks so much, Brian. Lots and lots of great information there. I just want to remind everybody who's got a question for Brian or Lance, now's a really good time to type it into the question box, hit Submit, and we'll get to them.

Smart Customer Service Q&A

Hey, Lance, just to kind of jump back to you. A couple of things jumped out at me. One is, you know, it sounds like a great solution for just kick-starting the conversation and getting rid of all of that identity interrogation that goes on at the start of a typical call, all for great reasons. But you know, it's still not a great way to start things off.

If you do run into somebody who is not on a recognizable device, you'd just go down path B and ask all the typical questions, I would imagine. And then, if you're still running into questions, then it gets kicked into some sort of fraud protocol. Or how does that work?

How does a typical fraud protocol work?

Yeah. Generally, that's the case. And most of our customers will start off and really focus on those authenticated calls, which we kind of call a Green Result, right? It's a good result. And the first thing is, how do I change my call treatment? How do I change my routing to take advantage of the fact that I now have this highly trusted customer on the phone? And so that's kind of the first task.

And then the next step really is, OK, let me start thinking about how I can perform more of a risk-based authentication strategy on my non-authenticated callers? Because still, many of those callers are going to be perfectly fine. They're not, by any stretch, all fraudsters.

For example, we don't authenticate a call that comes from a business line. Because we don't know who within a business is actually making that call. And so we provide an indicator of how trusted the non-authenticated calls are. And so you can look at, perhaps, a slightly improved call treatment for a moderately trusted caller. And then when you have a real no-trust caller, those are the ones you're going to want to look at more intently, just as you said.

OK, great. Hey, Brian, you know, we obviously live in a world of lots and lots of acronyms. I don't know if you could basically tell us the difference between NLU, Natural Language Understanding, and NLP, which is something that I think has been around a lot more time, and how they improve customer service. Can you kind of walk us through those acronyms?

Can you kind of walk us through NLU and NLP acronyms?

Sure. This is for Brian? Yeah. I can definitely take that one. It's an interesting question, right? I think a lot of different people have a lot of different opinions about what natural language processing and natural language understanding the differences truly are. You could even do a Google search on Wikipedia-- a Google search or look on Wikipedia to try and get the different answers there.

From Attivio's perspective, it's pretty simple. We like to think about NLP. From the question and answer perspective, from the types of solutions we're trying to provide, NLP is strictly speaking around how do we enrich these answers that we want to provide to end users? How do we start to extract some of those key facts about the information that we're looking at?

Again, if it's customer communications, maybe there's sentiment analysis we want to perform, and that's something we can do through our NLP capabilities. If it's documentation that has a lot of different key facts of information, error codes, product and services that are being mentioned or referenced, those are key entities we can automatically extract as well. Versus NLU, that's really more on the actual question and answer side of the house. Meaning you have all of these answers in one place of Attivio, and now you want to ensure that those end users can ask their questions in a way that's very natural to them, in a natural language manner.

The difference between who understands how to fix this problem, versus what is this problem, and how do I fix it? Those are two very different questions that you can ask an automated artificial intelligence platform and get two very different answers. The first one, you might get an expert. You might get a person and a phone number to call. The second one you would get documentation related to that particular question.

So when we talk about NLP and NLU, those are two very distinct terminologies that we like to describe in our realm, so to speak, but we've been doing a lot of heavy focus on the NLU side of the house. Really trying to ensure that when your customers, when you're end users, are asking questions, they can start to really get to the experience that they already expect today with devices like Siri and Google Home and Amazon Alexa. Right? They ask those questions very similarly.

Great. Great. Lance, does the trusted solution improve fraud detection? Can you expand on it? I know you covered this in your presentation, but can you expand a little bit on that?

Yeah, absolutely, and I think you can regard strong authentication is really the foundation for fraud fighting. An analogy I'll draw is if you watch a crime show or something, or you interface with like, I guess, real detectives. If there's a crime, one of the first things that happens is that the investigators of that crime are going to try to clear anyone that isn't really a suspect.

They can't take every possible person, and take him down to the police station, and run him through a big interrogation. So they've got to get really focused. There's only so many detectives to go around, and they don't want to inconvenience people who really couldn't have been a criminal.

The same exact philosophy applies in a prospective manner. If we're trying to run a call center, we're trying to identify and screen out all of the people that are calling in who are not possibly going to be a fraudster, and that's really the role of our technology. We're guarding those accounts, identifying the good callers, and what that does is it lets the fraud teams get a lot more focused on these non-authenticated calls.

And that's really of value, because any organization and any fraud team only has so many people. They have so many systems, and if they can get those focused on a small subset of calls, instead of trying to look for fraudsters in all calls, that's much more efficient. You can almost think of it as, instead of looking for a fraudster, it's a needle in an entire haystack. With strong authentication, you take a lot of that hay out, and you're left really just looking for a fraudster in a hay bale, instead of an entire haystack. And that's really the fit between strong authentication, making fraud fighting a lot more efficient and effective.

Right. Right. OK, kind of like standing in an airport and making everybody take off their shoes when you're only looking for one guy that did it 15 years ago, but all right. Great.


How many customers have self-service portals as opposed to interacting with live agents?

Yeah. No, it's a good question. Typically, what we see is with Attivio we can handle both. Right? The idea is that the same place where customers, where your customers, want to ask their questions and get answers, that getting those answers from Attivio is just as critical to provide to those customers as it would be to provide to the agents. And maybe-- which is often the case-- maybe those agents would have access to more sensitive information that wouldn't necessarily be public-facing. That necessarily you wouldn't want to publish out in the wild, so that the customers could find the answers to those questions, and that speaks to the security aspect of Attivio.

But I think the bottom line is, usually what we see in the marketplace, there's a lot of customers that are looking at how do I help the customer first? How do I help enforce a much better self-service portal experience? But they always have their eye on the prize, ensuring that it's not just about the customer, but it's also about the agent too. It's also ensuring that, when those calls do come through-- because I think we all know that they are going to come through in some way, shape, or form-- that those agents can be much better and faster and better equipped at actually answering those questions very quickly.

So I think at least the general pattern that I've seen, and that we've seen in the marketplaces, folks tend to start by thinking about how do I improve the actual self-service experience? But they very quickly understand that, oh, wow, this same capability that Attivio is describing to me can not only help improve my customers' ability to answer their questions and really help to deflect cases and thereby decreasing the case load for our agents, but also improve the agents' response time as well, that mean time to resolution. So that's kind of a progression that we see, but I think where Attivio wins is we handle both very elegantly and very securely.

OK, great. Lance, I'm going to jump back to you. The question is have you implemented the TRUSTID solution with a voice bio authentication product, and just while you're answering that, can you just clarify what voice bio is? Is that voice recognition and biographical data, or does it mean something else?

What is TRUSTID with voice bio authentication?

Happy to do that, and voice bio certainly has all the sizzle within, so many people tend to look at it as one of the first alternatives when they're looking to improve authentication. And voice bio's basically a process by which a customer can enroll in a service to store a profile of their voice, and when they subsequently call in, and they're talking to an agent, or they can talk into an interactive voice response system, and match their live voice to their stored print. And in that process, complete an authentication very similar to, I got a fingerprint, it matches my stored fingerprint.

We believe at Neustar that the end destination for organizations that do need to identify callers is a multi-factor approach. It is a layered approach of both an ownership factor approach using the phones themselves combined with a voice bio approach. And we think that that's going to happen sooner rather than later. I just spoke at a conference about a month ago with a representative from USAA, which is one of our customers, and also they use voice bio as well, and the two technologies together work wonderfully well.

We can't authenticate every caller. We can't authenticate, for example, a caller from the PBX in a business, because we don't know, as I mentioned, who's on that call. But you could still authenticate them using a voice print. And there are situations where there's background noise and things that prevents the voice bio authentication from working. But that person could be calling from a cell phone that's a solid phone, and the TRUSTID-Neustar service can authenticate them.

So we believe that they work very nicely together, and that ultimately, that's going to be the destination for the market within the next, I'd say, three years or so. And I think, as a consumer, wouldn't we all like that? Wouldn't we like to call our bank and just have somebody say, hello, who are you? How can I help you? And know at the same time that your account is very solidly protected against a takeover attempt.

Absolutely. Absolutely.

Now, Brian, I know that you've got to be working with huge data sets. How many different types of data can Attivio connect to? Can fill us in on a little bit like that?

Sure. Yeah, so with Attivio, as you say, we kind of hit broad swaths here in terms of the places where answers might be found. It might be internal-facing data sources. It might be external-facing data sources.

How many different types of data can Attivio connect to?

The beauty of Attivio is that we have connectors, as we call them, that can connect through all these different places, all the places where the knowledge might live. That might be databases that live internally. That might be file shares, your shared drive, your Z drive, or your G drive, or whatever you might call it. The ability to connect to you and crawl the unstructured content that lives in those places.

We also have web crawlers that can crawl public-facing or even private, internal-facing internet sites. We also have application connectors, things that can connect to places like SharePoint Online or Atlassian complements or Jira or Zendesk or Salesforce, as the case may be. So all told, there's about 60 different types of application connectors that exist. Many different more connectors are supported through our database and file system and web crawlers.

In addition to that, we also have support for over 600 different file types, not just PowerPoints and PDFs but also word documents, Excel files, rich media files. The list goes on and on. The nice thing about that approach is it allows us to be very flexible in the solutions we provide.

For example, we have a lot of support around deployments within financial crimes analysis and fraud analysis, understanding unstructured communications as they exist within the financial services institutes. And making sure that the things that those traders are saying, those people on the floor are doing, they're not necessarily saying and doing the thing that might get them in hot water with the regulatory bodies. So that's a very good example of how different types of data source connectivity that Attivio provides out of the box can enable very high-volume, very high-value, risk-avoidance use cases. Going above and beyond customer support and enabling those customers and moving into fraud detection, fraud analysis, based on the spoken word of communications and things of that nature. So it's a pretty long topic that we can get in on a little bit more, and I'd be happy to answer any follow-up questions that come out of today's conversation.

OK, great, and last question, a similar question for you, Lance. Where does the TRUSTID solution integrate? Can you give us an overview?

Yeah. There are two common points. It gets, obviously, a little bit techie here, but within every business for a call center, there's a device called a Session Border Controller or an SBC, which is essentially the telephone equivalent of a router and firewall. That's one of the integration points. And then secondly, we can also integrate within the IVR system, the Interactive Voice Response system. So either one of those are common integration points for us.

And again, what we're really doing is making sure that the call is held for typically around three to four seconds, so that we can get the information that came in with the call. We can do all of our inspection that I described earlier, and then we return a response with the result of our authentication activity. And that includes also a number of additional data parameters that are returned to the client that they can use for more sophisticated call treatment and call routing.

Great. Great. I see that were up at the top of the hour, so we've run out of time for questions. But please contact either Lance or Brian on your own if you'd like, have any more questions for them, and thanks to everybody that joined us today. Thanks to everybody that submitted questions. I'd also really like to thank our speakers for today, again, Lance Hood, Senior Director of Product Marketing at TRUSTID-Neustar, and Brian Flynn, VP of Solutions Architecture at Attivio.

If you'd like a copy of these presentations, you can download it once the event is archived. Also, if you'd like to review the event or send it to a colleague, you can use the same web address that you used for today's event. It'll be archived for 90 days. Plus, don't worry, you'll get an email tomorrow with links back to the presentations and the actual presentations. So that will also tell you when the event has been archived and you can share that with anyone you'd like.

And just for participating in today's event, you could win this $100 American Express gift card. The winner will be announced on July 31st, a couple weeks, few weeks, and don't worry, we'll reach out via email if you're selected as this month's winner. And that concludes the broadcast for today, so thanks for joining us.

Thank you, ladies and gentlemen. This concludes today's web conference. We appreciate your participation. You may now disconnect. Presenters, please stand by.

View Full Transcript
Related Resources