Future of Work: Transcript of interview with Gary Angel, CEO and Founder of Semphonic

Recorded on: May 18, 2012

Host:                     Welcome to the Human 1.0 Future of Work series, hosted by Scott K. Wilder, digital strategist and founding partner of Human 1.0. Today’s call is with Gary Angel, founder and CEO of Semphonic, the world’s leading web analytics and measurement company.


Scott K. Wilder:    Hi there, my name is Scott Wilder. Welcome to the Future of Work series where we try and give you practical tips for tomorrow’s practitioners. Today I’m excited to introduce Gary Angel of Semphonic, the world’s leader in web analytics and measurement consulting. And I’ve known Gary for, god, it must be close to 10 to 15 years right now and I’ve watched Semphonic grow into a world leader in this space. And so, Gary, why don’t you say hello and tell us a little bit about Semphonic?


Gary Angel:          Hey, Scott, thanks. So yes, I’ll give you a quick personal journey and then the Semphonic journey as well. You know, I started out as a programmer, actually, which I think is actually a pretty interesting background for a lot of the work we do. I never regret the work I did as a technical person. And I sort of evolved from a programmer into someone who was working with large-scale credit card databases and financial services and got very involved in working with large amounts of data and doing analysis of people’s transaction behaviors and what that meant about them. And that led pretty naturally into Semphonic, because Semphonic really does that, albeit in one particular area which is focused on digital measurement.


And when the web started to grow, I live out here in the San Francisco Bay area and saw a lot of people getting very excited about the web. Got very excited about it myself and thought about the work that I was doing over in measuring people’s behavior with credit cards and thought, you know, there’s a lot of similarities that people over on the websites are showing you their interests by the way they navigate and the way they move through your pages and what they look at and what they buy and how they spend their time. And I thought that a lot of the techniques that we developed could probably be brought over into the digital realm and would drive interesting measurement that would be useful to people as they explored the Internet.


And I think that, in many respects, that was a great vision. When we first started Semphonic we were really ahead of our time in a lot of respects. I think that people were still struggling to figure out what the Internet was about and the demand for measurement was really low. Nevertheless, I have to say despite that we grew up a fair amount as a company in the dotcom bubble. And that wasn’t so much, I think, because we were doing interesting work. In fact, we were consistently asked not to do very interesting work by people. All they really wanted to know what the most inflated view possible of the hits on their websites so they could go sell themselves to venture capitalists.


But, which really isn’t a great measurement function, but what we found was that in fact a lot of the things we thought we could do based on the techniques we developed didn’t actually work that well in digital. Digital data turned out to be pretty different; a little thinner gruel; a little harder to work with. In the wake of the dotcom bubble collapse we shrank right back down, too. I think we barely survived that, you know. All of our clients didn’t just stop giving us business; our clients literally went away. They didn’t exist anymore. And that was tough. We went through a few years where we really just didn’t have clients.


And it really wasn’t until about 2004, 2005 that we really started to see, one, the web really start to get traction again and people start to take a serious interest in measurement. I think a part of that was the development of tools like Omniture SiteCatalyst and WebSideStory had a tool called HBX that really made it a lot easier to start looking at the way people were navigating your website and to use that behavior in your business. And we quickly realized that those tools were the future. We became very heavily involved with them. And we started growing as a company. We’ve always focused very much on the enterprise practice of analytics. We’re a company that, you know, we come out of a background of analysts and that’s always what we focused on.


We found over time that we had to offer a broader range of services to people; that a lot of clients were enamored of the idea of doing real analysis, but came to us and said, “You know what? My web analytics data just isn’t any good. It just isn’t useful. I don’t want you – I’m not going to pay you to do analysis because I know the data’s garbage.” And so we started working on implementations and helping people get their tagging and implementation set up properly. And we found that a lot of organizations struggled to really get the basic foundations of reporting in place so that they could actually ask interesting analysis questions.


So we built up a practice around that. And over time we’ve become a full-service digital measurement consultancy where we work across the whole spectrum of problems around web analytics, and also around new channels like mobile and social. But we’re still in almost every case focused on really doing analysis for people because I think it’s at the analysis level that you actually start to see results from data. It’s where you actually start to take the data and make recommendations and how to think about how to improve your business; how to do better from a marketing or an operations perspective. As a company that’s what we’ve always been focused on and I think all of the other stuff is just getting you to the point where you can really do that.


Scott K. Wilder:    So I want to jump in for a second. You said you guys are mostly analysts. So my question is twofold. One is can you talk a little bit more about what type of analysts? Is it more of a business analyst; a data analyst? And then also I’m really curious about who tends to be your internal client in these companies or who’s your point person; where they are in the organization. Are they always in marketing or are they in other parts of the company?


Gary Angel:          Yes, both those are pretty interesting questions. I think the first one in particular because digital analytics tends to be a little different kind of analysis. As I mentioned a lot of the modeling and predictive techniques that we brought over from database marketing turned out, at least in the early days, not to work so well on the web. The data is not really at the customer level. There’s so much anonymous behavior. You’re working with streams of data that aren’t necessarily that interesting. Someone viewed page X, they viewed page Y, they viewed page Z. It turns out not to be as useful for doing customer segmentation and analysis as, hey, what they bought at a store with a credit card was.


So we found that over time a lot of our analysis ended up being focused on the things we could get at with web analytics tools. And those things tended to be around the way the website worked. So we developed a lot of interesting practice about how to measure the function of pages on a website; which pages are performing well and which aren’t. We developed techniques really specific to digital and the web for doing that. We evolved that practice into something where we actually started measuring use cases on the web and categorizing visit types and looking at the way people were coming into the web and the way they were splitting off and then building tests around each particular use case so that people could do A/B tests and multivariate  tests within that.


So a lot of our analysis is focused on what I would call classic web analytics, which is how do we make the website or mobile apps or whatever it is, the digital channel, better as a tool. But increasingly I find that we’re getting back to a world where people are doing more interesting analytics; the sort of analytics that we actually started out with have begun to fall back within the range of digital data. What I would call customer analytics, where you’re actually looking at the lifetime value of customers; where you’re actually building predictive models that help you understand which customers might attrite or which are the best candidates for retention that really help you understand of your marketing efforts which are giving you incremental lift.


This kinds of analysis, in fact, was relatively rare in web analytics until pretty recently. And I think it’s only with, one, a considerable explosion in the amount of data we have, two, access to that data outside of web analytics tools, and three, just a level of maturity where organizations are ready to consume that that we’ve started to see that customer level analytics flow back onto the table. And I think that’s a great thing because that’s really where the most potent analysis and the biggest ROI is for businesses looking at analytics.



Scott K. Wilder:    So you’re saying that companies are finally looking at the lifetime value of their customers and classifying them in certain segments based on that value itself?


Gary Angel:          They’re starting to get there. You know, it’s such a basic thing. It’s something that 15 years ago in direct marketing we did routinely. Everybody did that. That was just the standard in practice. And I think in the world of digital measurement and web analytics it’s still only the most mature enterprises that are doing that effectively. But at least we are starting to see the leading edge companies really delve down into that level of information at the digital level.


Scott K. Wilder:    I think that was the biggest surprise for me coming to Silicon Valley in ’91 from a credit card company is just how long – or just from ’91 to now just how few companies were really truly looking at lifetime value through multiple, you know, involving multiple channels.


Gary Angel:          It’s true. You know, I often am surprised by how slowly the state of the art in analysis evolves through companies and in many cases even take steps back. As we move into new channels and we create new types of businesses, they often have to relearn lessons that had become standard practice in other disciplines and in other kinds of companies. So it actually isn’t unusual in my experience for in some ways the state of the art to go backwards for a while. That’s despite the fact that the technology we’re working with is getting better and better on a very consistent basis.


Scott K. Wilder:    So what do companies have to do – first, two things. So first of all, what’s changed besides this big data phenomenon? And secondly, you know, if you were saying here are the two or three things a company needs to do to start thinking about lifetime value or customer data, what would you recommend?


Gary Angel:          Well, there’s a couple things. I think right now, to be honest, I think there is a technology barrier that companies have to cross. It’s very difficult to do customer level analytics within the web analytics solutions that are on the market. So I think most of our clients who are actually crossing that barrier are taking the data via data feed, via alternative feed, however they’re doing it, they’re taking the data and moving it into an analytics warehouse. And that’s part of what’s driving the whole big data phenomena. People are all of a suddenly finding they have to handle the data instead of their software as a service vendor. You know, as long as it was Webtrends or Google or Omniture handling all that data, nobody cared all that much about big data. But when it’s you that has to do it, suddenly it matters a lot.


And I think that’s where big data really comes in is we’ve seen a spread in terms of the number of companies that really want to get their hands on the data directly from being a relatively small number of people who are software service technology companies who are deeply experienced in it, had custom and proprietary solutions for it to a lot of large enterprises feeling like, hey, wait a minute; now we’re taking this data. We have big data problems. We need technology solutions that allow us to get at it.


So I think on the one hand there’s a technology barrier, but even more than that I think that what really has to happen is a fundamental change in the way people think about their metrics. I’m afraid, really, that web analytics as it’s commonly practiced has fostered a lot of really bad ideas on people and there’s been a tremendous focus on the wrong kinds of things; you know, looking at traffic; looking at high level conversion rates; things that really don’t matter and aren’t at the right level for the business to really understand their business and the way it works.


And so I think actually a part of what people need to do before they can effectively take advantage of big data and data warehousing is rethink their approach to analytics. Make it much more audience focused. Make it much more segmented. Make it much more granular that what people are doing right now.


Scott K. Wilder:    So I think you just answered the question I was about to ask, but I just want to dig a little bit deeper. So specifically, where should they focus their attention and what would that entail?


Gary Angel:          Well, at the customer level. I think one of the big differences that I see in people who are doing analytics at a sophisticated level versus people who aren’t is that people who aren’t are still talking about site metrics. Well, site metrics just aren’t that meaningful. And if you’re going to look at site performance metrics, you need to look at them in terms of customers. But there’s a whole bunch of new metrics and ways of thinking about your business that open up when you start focusing on the customer.


You mentioned lifetime value. Lifetime value is really fundamental. You know, if you’re going to measure the effectiveness of your acquisition marketing, it’s not enough just to do good attribution. Attribution tells you how you’re crediting the various campaigns, but it doesn’t give you any sense of whether the end result was a profitable customer or a customer that’s costing you money. And in fact, if you’re really going to decide how much you need to spend to acquire new customers, you need to understand how much those customers are worth to you over the lifetime of that customer. And I’ll tell you, people are investing a huge amount of money in the problem of attribution and totally ignoring the problem of lifetime value, which is actually I think for most businesses far more fundamental and far more meaningful.


Scott K. Wilder:    And when you say attribution, just for some folks, can you double-click on that and define it?


Gary Angel:          Sure, so one of the big problems that digital marketers have faced is that all the measurement that most people do has been very siloed. So they invest in pay-per-click at Google. And their pay-per-click agency gives them a report that says, hey, I sourced 10,000 people to you and I got 1,000 conversions. And then they go over to the display guy and he says, hey, I sourced 10,000 people to you and they got 1,000 conversions. The problem is that maybe 500 of those conversions were shared; that they were sourced by both display and pay-per-click. Each channel takes credit for it, and up the organization it looks like each channel is performing well, but when they add up the numbers they say, wait a minute. We should have had 2,000 conversions and we didn’t. We had a lot less than that.


And so in effect, you know, what attribution is about is figuring out the contribution that each of your digital marketing channels contributes to your acquisition of customers. And that is an important problem. You know, when I say people are spending too much time on it relative to lifetime value, I don’t mean to suggest that it’s actually not a fairly interesting discipline in and of itself. I just mean to suggest that it’s only a small part of a larger problem and that in fact you can do a 100% perfect job of attribution but still be fundamentally misunderstanding whether your marketing spend is effective or not.


Scott K. Wilder:    So are there companies out there or organizations really setting the standards in terms of approach or is that the wrong way to think about it? I mean maybe it’s, I mean standards would be great for certain things, but we’re at such an early stages of level that instead of standards maybe the question more is are there organizations really laying out, you know, here’s the how-tos and how to figure it out?


Gary Angel:          You know, I don’t know. I do find that there are different industries, certainly, that seem to be really good at different aspects of things. For instance, I’ve noticed that in many respects people in the ecommerce field seem to be doing a significantly better job around attribution. People in hospitality seem to be more focused on lifetime value. They have a better sense of, you know, bringing in a customer and how much that’s worth to them; particularly in an environment where they have to deal with multi transactions.


You know, our financial services clients also tend to be pretty good with lifetime value, but it’s a little easier for them because so much of the value is trapped right away in the initial transaction with the customer. You know, if you go off to a brokerage and you park $100,000 in the account they have a really good sense of how much that is and they may never expect to get more transactions from you, so it’s a little easier for them to measure lifetime value. In a multi transaction environment that can be more challenging.


And then I find, you know, our media clients are really not that great at things like lifetime value, but they’re really good at measuring things like customer engagement with content and content effectiveness in terms of customer engagement that maybe our retail and hospitality clients aren’t so good at. They tend not to be as effective in measuring content. So there’s a lot of different aspects to digital measurement. And in general what I’ve sort of seen is that different industries tend to focus on different aspects of it. Obviously the ones that are either easiest for them or most important to them, and they get really good at it, but then they’re often doing a really poor job at some of the other things that would actually be pretty helpful for them to understand.


Scott K. Wilder:    And what about, you know, the web versus – I’ll say the Internet versus social versus mobile in terms of analytics? How are they different? How should people think about them and, you know, what’s on the horizon for them; especially in the mobile space?


Gary Angel:          Yes, it’s interesting. You know, in many respects we’ve come to think of fixed web and mobile websites as nearly identical from a measurement perspective. There are differences, but in fact in our client base, most mobile web measurement has become very similar to fixed web measurement. It’s a little different, actually, but with the evolution of smartphones and browsing experiences that you can use the same kinds of tagging and measurement strategies, many of the similar kinds of reporting and analysis that you do on the fixed web are very appropriate to mobile web. There are some differences resulting from the types of websites and screen size and device dependencies. But by and large I think it’s fair to say that much of the practice that’s been developed and has gotten reasonably mature on the fixed web is highly appropriate to mobile websites.


Now that’s not true in mobile apps. I’ve been consistently disappointed with the state of measurement in mobile apps. We see tons of mobile apps that are not measured at all. And when they are measured, it’s basically thoughtlessly. They’ll slap the most basic app measurement into it. They’re basically measuring app opens or app downloads. They’re really not looking into how people are using the app or how it fits into the broader customer journey or whether the app (sounds like: gooey) is really working or not. I think there’s a lot of interesting measurement for mobile apps that’s left on the table.


Scott K. Wilder:    It sounds like email in the early days.


Gary Angel:          Yes, in a way it is. I mean I think it’s surprising because people are definitely spending money on this and mobile apps can be incredibly measurable and I think a really important part of understanding the overall customer journey. They’re a more impactful touch point, often, than mobile websites or even fixed websites. So I think there are a ton of measurement opportunities there that people really aren’t taking advantage of.


Social is its own kind of thing. It’s very distinct. Even more distinct than mobile apps from fixed web and mobile web. And mobile apps do need to be measured differently. They present a different set of measurement problems, but there are some similarities. Social is almost completely different. Social is different kinds of data. It’s unstructured text. It takes different kinds of tools and different kinds of analysis. And it’s actually good for many – for mostly different kinds of measurement and analytics.


So we tend to use social measurement for very different things than we use web analytics measurement for. And I think, again, in social we’re seeing a pretty primitive state of measurement. The tools are often pretty buggy. Some of the things they claim they do they don’t do very well. People don’t really understand the social measurement lexicon. They don’t understand how to take advantage of the measurement they have. And by and large in the organization, the penetration of measurement; people just – people aren’t looking at social media measurement very much. And when they are looking at it, they’re generally looking at it at really shallow levels.


So I think of the various disciplines, social media measurement has the farthest to go, but mobile apps in some ways is more disappointing to me because I feel like social media is evolving really rapidly. The tools are coming along. People are growing. It’s just that it’s very, very new and it’s very complex, so it’s a little harder to get a handle on. I feel like mobile apps, the technology and the practice areas are there. Companies just aren’t taking advantage of it.


Scott K. Wilder:    Say more about what’s there and – yes, I guess what’s there and then how, in terms of the mobile space, how this might evolve in the next year or two.


Gary Angel:          Well, so mobile measurement has become significantly better in the last year. The web analytics vendors have significantly improved their software development kits around tagging mobile. There are some pretty good mobile vendors out there, people like Bango and Localytics, that if you instrument well can provide you with pretty good measurement around your mobile apps. So people have a range of choices about how to instrument mobile apps. And I think that there is a set of practices evolving around thinking about mobile applications from a more sophisticated paradigm; one that isn’t very page based. You know, one of the challenges to mobile apps is they don’t look like websites. So it’s not page by page by page. And that paradigm does not work very well. And if you try to fit your mobile app into a page based paradigm, you’ll find, I think, that your measurement isn’t very interesting.


So increasingly we focus on mobile apps as being measurable in terms of what we call units of work, where we really look at a person trying to do some kind of function on the mobile app and how successful and how long it took them and how they navigated within that function. And we find that that allows us to create different and more interesting metrics, maybe, than we do if we try to stick to a page based paradigm.


Where it’s going to evolve? I don’t know. You know, to be honest, Scott, I’ve been consistently disappointed with how rapidly mobile measurement has evolved. Again, not so much from a technology or practice standpoint; it’s from a take-up standpoint. I mean I think from our business we’ve had many more companies take us up on interesting social measurement than interesting mobile measurement, despite the fact that I think that mobile – interesting mobile app measurement is easier to have, and in some ways right now, more practical. So I just think the take-up has been lacking there. And I’m not sure how that’s going to evolve with people. I’ve got to believe it will. Ultimately I think as people realize how much money they’re spending around applications and as those applications get significantly more use and become a hole in what people understand about what the customer is doing, I think the measurement will grow. But frankly I’ve been burned on that kind of prediction for the last couple years, so I’m not that confident about it.


Scott K. Wilder:    And in the mobile space, is it the marketer you’re dealing with first in terms of analytics or do you – does IT get involved or is it someone else?


Gary Angel:          Yes, in almost all the spaces we work with I think marketing is primary. I think that our stakeholders tend to be marketing folks who are focused on how they can talk to customers; how they can acquire new customers; how they can handle customers from an operational standpoint. Sometimes that goes beyond marketing. Obviously in many enterprises once you’re a customer sometimes you shift hats and sometimes you don’t. Some of these marketers are focused on the retention problem, too. But I would say almost across the board, whether it’s social, whether it’s mobile or whether it’s fixed web we deal primarily with marketing.


The one place where that is changing is, you know, what we sort of touched on earlier. As more organizations are starting to think about moving the data in-house, looking at big data solutions, there’s certainly been increasing involvement of IT. I think it’s not that people are abandoning the software as a service model, but a fair number of companies are supplementing it with internal IT. And at that point when you start talking about those kinds of technologies, now it isn’t just a marketing discussion anymore. It’s clearly an IT discussion and a marketing discussion.


Scott K. Wilder:    Well you mentioned big data and I don’t think any conversation about measurement or analytics these days happens without mentioning big data. So let’s just start at the most basic level. You know, I’m reading about big data multiple times a day. And, you know, if I’m a C level individual how should I think about it? Here’s a good example is I got a call two days ago from somebody who said I’m now the big data czar in my company. I don’t know what to do. I don’t know what that means other than unstructured information that I need to structure.


Gary Angel:          Well, you know, I think the first question to really ask yourself is do you have big data. Because one of the things that I guess I’ve seen is that people tend to think of all, for instance web analytics data, as big data. And there certainly are companies in our client base that have true big data in any – by any definition of the term. You know, when you think about CNN.com or walmart.com, those kinds of companies are spinning off volumes of data from their websites that I think would fit anybody’s definition of big data. So they have to – the thing about data is that as it grows orders of magnitude, it becomes – it starts to be unusable in traditional technologies that people have used. But you have to ask yourself whether you’re really there or not.


And not all big data is unstructured. Some of it’s structured. Obviously web analytics data is structured data. So if you’re CNN.com, you’re throwing off a tremendous amount of structured data. It’s still structured database-like data. There’s just so darn much of it. And of course social data is another place where you tend to spin off lots and lots of it, but it’s unstructured. So, big data really encompasses both those things.


But I think the first questions to ask yourself is am I really outside of what traditional technologies can handle? Do I need to look at a set of new technologies? We’ve found that some organizations that we have have looked at, you know, what I would call new big data technologies when in fact their data volumes are very manageable under what I would call traditional relational technologies; things like SQL Server in Oracle and Teradata. So one question to ask yourself is do you need to invest in new technologies or do you have the technology you need?


I have to say, too, that sometimes people think they don’t have the technology they need because they’ve implemented it so poorly. You know, if you take your web analytics data and you dump it in a database and you don’t model it well, one of the things you’re going to find is that even with relatively small amounts of data you can get miserable performance with traditional technologies. And so there’s first an exercise, I think, to understand what is the scale of your data and then to relate that to the appropriate technologies. And there’s a whole bunch of technology solutions that are potentially appropriate for people ranging from the traditional solutions like SQL Server in Oracle to bigger solutions like Teradata to sort of highly specialized solutions like Netezza where it’s a data warehouse appliance or Astor, which is a massively parallel system, to systems like Hadoop that are really only appropriate for people who have true big data problems. But if you really have those problems, you may need to be up in the Hadoop or Astor worlds.


So I think part of coming to grips with big data is first coming to a realistic understanding of where you fit from a technology perspective. And then that second thing I think, you know, you mentioned social data and unstructured and structured data. It’s really important to understand where your big data problems lie, because there’s a big difference between handling unstructured data and structured data in terms of the types of technologies you might want to deploy and how you can optimize against those technologies. So understanding if you do have big data or what your data is, whether it’s structured, whether it’s unstructured; that’s sort of the next step.


After that, I mean, part of it is you’ve got to figure out what you want to do with it. And I think any time, you know, I guess there’s that assumption that, hey, we have all this data. We need to do something with it. But I can say from personal experience that a lot of times people do build warehouses and then have no idea what to do with them. And that certainly is backwards. You know, if you think about if I’m in charge of big data, you need to figure out, well, first of all what is that data and what can I do with it. And that’s going to have a pretty big impact on how you think about, again, what you need to do with the data and what the technologies that are appropriate for it are.


Scott K. Wilder:    It’s a very simple question. I mean this is fascinating. This is a simple question I have, but since Gartner tells me that 85% of Fortune 500 companies fail to effectively exploit big data for their advantage, what are some questions or things to think about or how should I go about thinking about what the hell am I going to do with big data?


Gary Angel:          Yes. Well let’s talk. And again, it goes back to what the data is. Let’s talk about web analytics data. I’ll start with that and then we’ll talk about some of the others.


Scott K. Wilder:    Okay.


Gary Angel:          I I think with web analytics data a lot of times, you know, once you move the data inside it is big data. You’ve got these large streams of server call events. And I think what you need to look at is how you can take that data and get – use it to get a better understanding of the customer. It’s about that jump to customer analytics. And we find that the biggest opportunities around big data for most enterprises are where they start doing segmentation to drive targeting. That simple. You know, segmentation is a proven technique. People have used it for decades now to actually improve the marketing ROI of their company. People haven’t done a lot of segmentation to drive targeting and messaging within digital. And I think that’s a true, powerful, understandable, immediate, impactful use of the data. And it’s one that often does take big data technology if your website’s spinning off really high volume. So I think segmentation and targeting are one of the top things that you can do.


I think going back to some of the things we also talked about earlier; focusing on understanding the lifetime value of customers and the incremental lift that campaigns actually give you to that lifetime value. That’s a really fundamental analytics challenge. And in fact, to solve it almost always takes a heck of a lot of data. You know, if you’re doing display and you’re doing search, you want to track impressions that people have and you want to combine that with all the digital data on the websites. It often is good attribution and good lifetime value measurement now often involved multiple channels, each channel of which will be spinning off a lot of information.


One area we’ve had a tremendous amount of success with in the times we’ve actually been allowed to do it by clients, and we’re not always allowed to because they just don’t seem to get it, but looking at the relationship between the web and the call center. You know, those are two areas of big data. Call center spins off a lot of data for big enterprises and so does the web. And there’s a really strong relationship between them these days. And understanding by putting those two sources together you can often get a really good understanding of when each channel’s failing, you know. When people are going to the call center when they had a problem that would have been easier and better solved on the web or people are going to the web when they had a problem that actually took the call center. And, you know, some of it’s about call avoidance, but a lot of it for our clients is about getting people to the right place that’s going to solve their problem in the way they want to solve it in the easiest possible fashion. And putting that call center and web data together is a really nice use. It’s often a big data use because you’re basically talking about two channels that each of them tend to spin off a lot of data.


So I think those are some of the kinds of things that I see as big data challenges. I think you’re often going to find that when you start thinking about big data you’re often talking about mixing fairly large data sources. And that creates problems of its own. It’s hard enough when you have one really large data source. When you start joining them together, it creates even more problems; especially for traditional technologies. So you have to think about what kind of data integration you need and which technologies are appropriate to that. Because it’s not all just about how much data you have; it’s about how much you have to integrate and join that data.


Scott K. Wilder:    One thing I try and bring up whenever we’re talking about analytics is the importance of qualitative information and not just quantitative data. You and I have talked a lot about this in the past and I’m just curious, you know, what your thoughts, if you want to share your thoughts on that. Because I really believe that, from my perspective, not enough companies really focus on what are people saying and how are they saying it.


Gary Angel:          Yes, I definitely agree with that. I think we’re both very much onboard with the idea that as powerful as behavioral analysis is, and I’m a big advocate of behavioral analysis; it’s what I do for heaven’s sakes. But we’ve found that we can often do a better job of helping people really use the information when we can supplement that behavioral analysis with attitudinal data and demographic data and psychographic data and all the kinds of data that we used to use in classic marketing. And there really is and I think will always be a powerful role for that kind of stuff. It plays a different sort of role in our understanding of the customer than behavioral data. It gives us the contextual background. It fills in questions that from a behavioral perspective might be extraordinarily difficult to answer. But if you just ask somebody, hey, what were you thinking here or what couldn’t you find on this page, it often solves the problem.


And I do believe, too, I’ll say this, I think maybe the biggest opportunity at the enterprise level right now is to really consistently and aggressively exploit attitudinal, social, textual data that’s collected. Most enterprises we work with collect vast amounts of data at the call center level. They collect vast amounts of social media data. They collect opinion lab data on their sites. They do attitudinal surveys, both on and off-line. But all those efforts tend to be completely siloed. And not only are they siloed, they’re non-standardized and the distribution of the information is poor. And so what we find is that every research effort tends to be a one-off. And that’s really ineffective.



You know, I often compare it to the world of behavioral data warehousing where we’ve found that building a customer data warehouse and bringing all the information about the customer into one place and then using that information consistently, you know, creating a set of standards for how we talk about and think about those customers, making sure that the data quality is consistent, we categorize customers the same way; all the things that have become standard practice around structured behavioral data, none of those things are done in the world of attitudinal and unstructured data. And I think there’s a tremendous opportunity for competitive advantage to organizations who are willing to put the effort in to effectively build a voice of customer warehouse; to consolidate all that information; to put standards around it; to make sure that they’re doing it on a consistent basis and to distribute it out to everyone in the organization in a consistent fashion. I don’t see most of our clients doing any of those things and I think there’s tremendous benefit to that relative to cost; probably more than anything we’re doing on the behavioral side.


Scott K. Wilder:    I think that’s tremendous–I think if people listen to this they’re probably going to go out and try and get the URL vocwarehouse or voice of the customer warehouse.


Gary Angel:          Which unfortunately doesn’t exist. I mean one of the issues to this I think is that the demand is so misunderstood that there aren’t technology vendors that are really fulfilling that function. There are some vendors that I think are working in that direction, but in fact have only very piecemeal solutions for those kinds of things. So, yes, it’s a place where I think most people will find that if they want to do that they’re going to have to bring some technologies to the table and integrate them themselves because there isn’t a one-off solution to that problem.


Scott K. Wilder:    So why have the vendors kind of stayed away from that?


Gary Angel:          Oh, I think that’s purely demand. I think this is the fact that enterprises have not understood that this is something they should be doing. No one’s heard of them because no one’s buying this stuff. And I think that my take is that the market is just not understood and that if there are technology vendors out there filling this niche, they have to be evangelical because people don’t get it yet. People aren’t doing this kind of thing. So I think it’s more a case of lack of demand and lack of understanding about how transformative this might be for companies to really broaden out everybody in the organization’s understanding of who their customers are, what their attitudes are, how those are changing and what they can do about it. I think that’s really valuable information at multiple levels in the organization and it’s just not consistently created, distributed or exploited right now.


Scott K. Wilder:    And correct me if I’m wrong here. We’re not just alking about sentiment analysis or simple text analysis, because I think sometimes when we talk about voice of customer and qualitative information, those two terms come up. We’re talking about something a lot greater than that.


Gary Angel:          Yes, I think we are. I agree with that. I mean sentiment analysis might be a little piece of that. And I think sentiment analysis, you know, as we’ve discussed before, it’s very challenging. A lot of tools don’t do it very well. You have to be very careful about servicing those numbers up to people. But even when you have it, it often doesn’t feel that meaningful unless it’s done within a lot of contextualization of the data. You have to make sure you’re looking at a real sample of legitimate consumers. You know, not your paid competitors who are out there Tweeting. Their sentiment probably doesn’t matter to you that much. So I think you have to be looking at the right sample. You have to be – you have to understand that you’re often misclassifying sentiments; very hard to do; especially in channels like Twitter. And then you have to understand what people are talking about; what’s that sentiment relevant to?


I think overall brand sentiment just isn’t that interesting. It’s much more understandable and actionable when I can understand whether the sentiment is related to issues around customer support or issues around product functionality or issues around my advertising campaign. Those are three fundamentally different things, and frankly, to understand overall that my brand sentiment on Twitter is going up or down doesn’t really tell me very much because Twitter’s not a representative sample. And so I have to be able to contextualize it relative to what that sentiment’s about before I can actually understand what it means and what to do about it.


Scott K. Wilder:    Definitely. So you’re dealing with all these different companies. Is there something you’ve noticed in terms of company culture that is more open to, you know, web analytics or mobile analytics or it’s you’re just getting one perspective on it because you’re coming through the marketing department? And I guess the other side of this is, you know, is there some things that you think companies should think about from a cultural perspective? So first part is if there’s anything in terms of the companies that you notice that’s different when they’re more receptive versus not and then also should they think about how the culture could change to be more receptive to what you guys are trying to do?


Gary Angel:          Yes, there really are differences; no question. I certainly see that some companies are just far more sophisticated and open to using data in their decision making. And those companies tend to do a lot better job of it. I think one thing I’ve seen is that in our world we tend to believe analytics is important. And it is important. But in fact there’s lots of different ways that companies can get competitive advantage. And in some industries analytics tends to be a bigger competitive advantage than others. I think if you look at, say, the retail catalog industry. Analytics was absolutely essential to competitive advantage in that. And so those companies tend to be pretty good at it. And as they moved over to the web, a lot of their analytic practice was able to adapt because people were able to understand, well, these things work but these things don’t, but, dang it, I still want those numbers. I’m going to find ways to get good numbers. Because that’s the way they did their business.


On the other hand there are businesses where their primary differentiation or competitive advantage is that they have a great sales force or that they’re a great manufacturer or that they make a really good product. Those are all legitimate differentiators. It’s probably more important that you have a great product than you have great analytics. As a result, analytics tends to suffer in those businesses. And I think sometimes we hear lip service in companies where people are talking about we’ve got to become a more data driven company. Well, that’s probably true. They probably should become a more data driven company. But nine times out of ten the companies that are saying that, data has never been a competitive advantage for them. And so oftentimes the companies aren’t set up to do it well. They don’t have an existing infrastructure. – they don’t have existing people who are used to using data. And I think that’s really important.


It takes a long time to build a culture and change a culture, and so I think – I guess our experience has been that companies who came to the digital world with a culture that was focused on analytics have almost always done a consistently better job with digital analytics and are much more culturally attuned to it. The other thing I will say is that necessity is definitely – definitely breeds adoption in the sense that I find businesses that are under stress. That can be crippling, too. Sometimes the business is under so much stress that it just can’t spend money on things. And that can fundamentally cripple an analytics program. If you can’t buy an analytics tool; if you can’t afford analytics resources you’re probably not going to be a good analytics shop. But if an enterprise is large enough to invest in areas and the enterprise itself is under stress, oftentimes that’s a pretty good driver of analytics.


I’ve noticed, for instance, in digital media, which is, you know, publishing guys and people who have print and digital; they’ve been under a lot of stress. That’s a very dynamic, changing industry where many of them are getting compressed and pressured and they’re having to figure out how to do their business better. Well, that forces them oftentimes to take analytics seriously. You can’t coast. You can’t get by on your margins and what you’re doing well. And that’s often a driver. So, you know, those may seem like odd answers in a way, but in the real world that’s kind of what I see as the things that actually drive companies to be good at this stuff.


Scott K. Wilder:    It’s funny you talk about the margins. I was telling somebody today that I no longer subscribe to a print publication that’s more than $15 a year. And that’s because they keep soliciting me until, you know, and keep dropping the price and so when they get to $15 then I take it.


Gary Angel:          Yes.


Scott K. Wilder:    So, you know, there’s so much going on in this space. I’m just curious about how you stay up to date; where you get your information. Are there key thought leaders, websites, blogs that you follow or even, you know, old-school books or Kindle books that you read?


Gary Angel:          Well, you know, that’s interesting. I guess I’m not – I’m probably going to shoot myself in the foot here and say I don’t do a lot of that kind of stuff. Despite the fact that I blog religiously and I hope and pray people will read my blog and look at it. But you know what? I find, like a lot of people, I’m very stretched for time. You know, I certainly understand why people don’t read my blog because I don’t read a lot of other people’s either. And I often find it’s not that I wouldn’t be fascinated or they’re not writing good stuff; it’s just that I don’t have a lot of time for it.


I find, you know, I tend to pick up – and I think this is one of the advantages that I have. At Semphonic we work with a lot of different enterprises. We’re very hands-on with a lot of efforts that are pretty cutting edge. And I got to be honest; most of what I pick up from a practice standpoint is involved from seeing what companies are actually doing; seeing the things that they do well; getting a chance to try things out on them and then taking that and applying it to other clients. So I think in many respects I rely on our ecosystem of talking to people and getting that experience and bringing that back.


You know, I do a lot of speaking and conferences as well. And conferences can be good. I especially like the, you know, the sort of conversational stuff with people you know and hearing what they’re doing and just pressing them. “Oh, what are you doing that’s cool? What’s interesting?” You know, that kind of conversation, I think, more than the presentations tends to be really interesting. And so I get a lot of, you know, sort of word of mouth kind of feedback on that basis. And I wish I had a chance to do more reading and do stuff out there. But I’ve got to say, I suspect like a lot of people I feel myself pretty time crunched to get out and do that kind of reading.


Scott K. Wilder:    So I’m going to put you on the spot here. So at these conferences or recently what have you heard that’s cool and exciting that people are doing?


Gary Angel:          Well, I guess as couple things occur to me that really sound kind of cool. You know, I think, I was talking with Tom Betts at the Financial Times. And one of the things we talked a little bit about was mobile. And I think one of the things they’ve done that’s really interesting is really integrate their mobile data with their web data to get that view of, you know, when they publish stuff and it gets consumed in one channel are they cannibalizing other channels or not. And that’s a really fundamental question for a publisher, right. If you’re starting to put out a tablet subscription, if you’re starting to put things out in multiple channels, really understanding whether you’re building readership, attracting new readers or cannibalizing existing channels is really, really critical. And I thought their effort to do that was both interesting and powerful and I think they came up with some really interesting results about it, too.


Another one was I was talking to one of our clients, a hotel chain, and they’d done some really fascinating work around attribution; really deep-diving into the impact of some of their branded buys. And had found that in fact they just weren’t getting much value from it despite the fact that it looked very successful. The incremental lift was much less than they were anticipating. I thought that was a – they had done some really sophisticated research and I think, you know, again, that’s their business, right. It doesn’t necessarily mean it applies to everybody, but I thought that the research itself was really interesting; the kind of research that other people ought to be doing to see how their marketing spend is really effective or not. And I guess in that case it touched off a lot of really fiery discussions within the organization. Hey, that’s an indication that analytics is working. If you’re not bringing stuff back that people are fighting over or arguing about, you’re probably not doing a good job with your analytics.


Scott K. Wilder:    Definitely. Any other companies or case studies that we should, you know, just think about and can learn from?


Gary Angel:          Well, you know I mentioned our call center work. And I think some of the biggest ROI work that we’ve seen is in helping people right-size their call center channel. And some of the most interesting work we’ve done there is in actually looking at people who either weren’t going to the website to solve problems that really were more appropriately solved on the website or were going to the website and not finding things and then going out to the call center, thereby becoming somewhat less satisfied.


And we actually did some predictive modeling to figure out when people were likely to call the call center, to have problems that they might go out to the website for. And actually tried to do sort of anticipatory – not really marketing because this is more operational, but sending out emails to people telling them how to go to the right channel; how to solve their problems. So if we thought they should be calling the call center we steered them in that direction. If we thought they should be on the web we steered them in that direction. And we were really able to shift a lot of people’s behavior into the right kind of channel. And I think that’s a great use for that kind of analysis and one where you can really see immediate results that impact the whole organization, both from a cost and a satisfaction standpoint.


Scott K. Wilder:    I mean that’s really great to hear, because, you know, I think you and I’ve talked about this. Some of the best ideas I’ve ever gotten in terms of product design or even, you know, working on the Intuit community came from listening to calls at the call center. It’s just a, you know, rich resource that I think gets undervalued. So, Gary, as we wrap up here is there any kind of last – anything else you want to share with the folks here?


Gary Angel:          You don’t have any more guidance than that? No, I’m not sure I do. I’m kind of talked out. {


Scott K. Wilder:    No problem, you have given us a lot guidance today. There’s a lot of great information and I loved the whole VOC concept of the warehouse and also the call center attributes. So Gary, once again, works at – he’s the CEO and founder of Semphonic. You can find him at Semphonic.com, and I’ll list that URL on the website. And then his blog is at semphonic.blogs.com/SemAngel and you’ll also be able to find that listed as well. Gary, I really want to thank you. This has been really helpful for me personally in terms of my own business here and also just things to tap into for the future. So Gary Angel, I want to thank you for joining us today.


Gary Angel:          No, my pleasure. Thanks Scott.


Scott K. Wilder:    Okay. Thank you everybody and again, this has been the Future of Work series and you can find this on csuitetwo.com; csuite-T-W-O-dot-com or at reachingsmbs.com. Our next conversation will be probably with Ron Lichty, who is an expert in Agile; Agile product development. And he’s going to really talk about what’s happened and what’s going to be happening in the future. Thank you for joining us.