Dennis Mortensen
Dennis R. Mortensen is a pioneer and expert in the Web Analytics industry, having worked in the field of internet analysis and statistics since 1996. He is an accredited Associate Web Analytics Instructor – the University of British Columbia and a frequent speaker on the subject of analytics and online marketing. He was the COO of IndexTools, a top tier enterprise analytics vendor, until it was acquired by Yahoo! Inc. in May 2008. Today he is the Director of Data Insights at Yahoo! You can follow his insights on web analytics and online marketing on his blog or in his upcoming Yahoo Web Analytics book.
Interview Transcript
Eric Enge: You were elected recently to be on the board of directors for the WAA. Can you talk about that a bit?
Dennis Mortensen: I wish I could tell you that it was the result of a long, hard campaign on my behalf. The fact is that they asked me to run, which I was very honored to do. But, I was in the midst of closing the things off with Yahoo! and moving my family to the U.S. So I didn’t have too much time to spend on the campaign, but ended up being elected anyhow.
That itself was kind of a kick-start for me to start thinking about what and how I could contribute to the organization. I think I have a task first of all of making sure that we all end up speaking the same language. I think we have spent ten years now speaking something that’s very similar, and sometimes even using the same words, but not always meaning the same thing.
Eric Enge: Such as how unique visitors are defined and that sort of terminology?
Dennis Mortensen: Exactly. If you want to communicate to an advertiser what your page view volume is, you can speak the same language, and you can sell a product that he is very comfortable buying from you because he knows exactly what you are selling him. Unique visitors is probably one of the harder ones, because even if you go down and try to define something as a page view, the term is very different from site to site. We might even be in need of new terms such as a content view.
Conversions may, as well, not be as clean as they used to be yesterday, and we might need to start defining metrics such as events. Conversions and goals are polluted with other types of actions. It’s time for us to clean things up and have some very clear-cut standards set.
Yahoo! Web Analytics or Omniture are not going to get better or worse by aligning to one standard. They’ll actually just speak to the same language. So, that’s one task for me. The other task is, of course, to serve the organization and its members and make sure that the members are getting value out of the organization.
Eric Enge: Right. So you talked about the standardization of terms, but then there are also other aspects, such as how does an analytics package react when JavaScript is not enabled? Do they fall back on IP and user agent tracking using a tracking pixel? What do they do, or how do they handle a situation where there is a cookied user, but the IP address changes mid-session or data coming in with the referral information is stripped.
There are all these different scenarios because the web isn’t actually a very homogenous place and all the vendors treat these scenarios differently.
Dennis Mortensen: I think for me that is part of the ultimate goal. When you put in a tracking script as you and I have done so many times, and you have these ten lines that look pretty much similar but are definitely not similar it can get complicated. So, you have ten lines from Google, ten lines from Yahoo, and ten lines from Omniture. And they essentially do the same thing, as in they collect some clickstream data, and send it back to a server somewhere, and you then get access to that data later on.
But that script itself should be standardized as well. So, the only thing you change in that script is the data destination source where you want the data stored. If I want to change from one vendor to another, it shouldn’t be dependent on the pain or re-instrumenting the web site, it should be dependent on a number of other things. So, I would like to take it that far, but that’s not going to happen for many years.
In a scenario where there is no JavaScript enabled, some vendors, like Google, do not track people who have don’t have JavaScript enabled.
Eric Enge: Right. They ignore that visitor.
Dennis Mortensen: They simply just ignore that visitor. I am not saying that’s good or bad, that’s a choice and they’ve made that choice for a number of reasons. But, then you have other vendors, like Yahoo!, we have a fallback methodology in the pixel.
Eric Enge: Right, a user tracking pixel.
Dennis Mortensen: Exactly, a tracking pixel. And again, nothing is better or worse, it’s just a different choice. But, it also presents two different numbers. First of all the numbers are hard to get aligned to begin with, but they are definitely going to be even harder to align when you collect more data than me (or vice versa).
There should be differences in the solutions, but we need to get them aligned on key points. So, If I look at your site and see that you have 24,000 visits a day, what does that mean? How many visits is that; is that a sample of 98%, is that a sample of 99%, what is that number? If you look at the visits and ask how many people come from servers that are stripping out referrer information, how do we try to figure out if they are a unique visitor? How do I do that calculation?
And that’s just me looking at my Google Analytics, my Microsoft, and my Yahoo!. Those numbers are no different to the extent where you really have to know what you are reporting on before you hand this over to an advertiser, for example.
Eric Enge: Right. Let’s take this further, let’s make it even messier still. In the mobile space there are also fundamentally different methods for implementing these things, right? It’s not just different JavaScript, but there is actually a different place in the clickstream where the data is being picked up.
Dennis Mortensen: Yes. I think what we will see in mobile is that it is messier because initially, you haven’t settled on the default data collection methodology. We have seen the web space default to page tagging. Not that many people are doing packet sniffing or a log file analysis or something else. Most people do page tagging today.
I think over the next couple of years you’ll see people try out different things in mobile. And then, they’ll end settling on some standard, whether that is on the server side, an image pixel or something else. I wouldn’t really know, but I think we’ll end up on one standard.
When you track mobile, it tends to be a limited set of matrices that you can collect, this being a simpler platform. It is easy to see the difference between my Samsung phone, and a desktop PC at work. But, then I can take my laptop and move around town. And, I can take my iPhone and move around town.
The only difference in those two is essentially the display size and some of the browsing technology such as how it renders the page. So, is that really mobile? My laptop gets smaller, and my phone gets extended with capabilities, what does mobile really mean?
Eric Enge: So, just to change directions a little bit, the long-awaited re-branding of IndexTools has now occurred.
Dennis Mortensen: Indeed. And I am extremely happy that we’ve gotten so far this quick. I think that people don’t really care too much about the fact that we did a legal transaction, and handed the company over to Yahoo!. That didn’t really make any direct market impact. But the integration into Yahoo has progressed far enough to have it re-branded.
I am extremely happy that we did this exactly five months after the acquisition. We of course took a different route than you’ve seen elsewhere in the market, which is that we took a not so aggressive idea of what to launch, because we wanted to make sure that we got to market and didn’t just end up spending eighteen months inside Yahoo! perfecting everything, making it scalable for every potential use, and then launching something. We wanted to simply get it out there very quickly; that was point #1.
Point #2 is that we tried not just to do a, me-too strategy. Thinking that just because we compete with Google in search, we must compete in web analytics as well. It’s probably not that simple, whereas I think we might not even compete in web analytics and we might not even want to compete in analytics. So, the strategy that we chose was one where we tried to optimize not just on the feature set of the tool, but also on the fact that most people do understand that the insights are never really better than the data that you collect. So, if you collect crappy data, you are probably going to end up getting crappy insights or no insights at all.
So, what we’ve seen is that even though people like you and me can instrument a site with the right tracking code and get the conversion on the right page and so on, but not everybody has this skill set. So when you try to implement the right SKU code, upload the merchandising information and create funnels, some people may require an engineer to do that.
If you look at the 60,000 merchants on Yahoo stores, I don’t think that many of them a) are engineers or b) have engineers available to them. So, what they do is they take a plain vanilla Google analytics tracking code, put it in the footer, and then it gives them some limited insights, and that’s cool. But, their implementation is quite limited. They put the plain vanilla code in the footer, and then they believe they are done, which is unfortunately not the case.
Then they start looking at the reports because they think they must provide them some sort of magic. But, those reports can’t provide magic, because you haven’t collected magic yet. What we are doing that is a little different is that we are taking advantage of the fact, that we are a major content owner and publisher, this including Yahoo stores and 1,850 other properties.
Because we own the content publishing platform and the analytics platform, we can combine the two, and we can do something where with one click of a button you could get an enterprise analytics instrumentation without doing anything. Because, I know exactly what your Yahoo store is about, what the name of the page is, what the specific SKU code is for that page.
I know what the product name is and what the category is, and the value of that product. Since I have all this information, I can serve the tracking code up at run time. And by clicking a button, it is essentially pasting the code into the footer. By doing this small click I get a perfect implementation, and you get a tool where you have the perfect data set available.
Now you can try to create insights based out of that. I think that’s a pretty cool approach. Whether or not it’s a winning approach, time will tell, but I think that’s pretty cool because we are taking out a very difficult part of the equation here.
Eric Enge: So, in the case of an eCommerce site, you can populate things like a user identifying code, the product that’s being purchased, the price, the quantity, the total shipping method. You can put all that kind of data automatically in the code.
Dennis Mortensen: Spot on. This is important because most people will fail to implement this type of thing correctly if they have to do it themselves, even if they understand the concept.
You could forget to put it on a page, or you didn’t put in that variable, or you are tired, or you are busy, or there was a launch date. There are so many elements that can go wrong. Here you just click one button and we do it for you.
Eric Enge: Right. So, you’ve built that right into the Yahoo! Store functionality.
Dennis Mortensen: Yes!.
Eric Enge: Now, for a publisher who wants to do it himself, they need to implement not a particularly complicated web application, but a little web application to populate the variables, right. That does require a developer, but if you do have access to a developer, it’s pretty straightforward.
Dennis Mortensen: Yes it is.
Dennis Mortensen: And, what we did here as well, is that for those who are at an expert level or don’t want us to instrument it, you can take it off on a page level. So, let’s assume that you like everything we do on the 520 pages, but there are 3 pages where you have a unique Store Finder. That Store Finder is so important to you because you make most of your revenue offline that you really want to instrument that in detail beyond what we’ve done for you.
So, you click a button saying, I don’t want to instrument this page, let me do that by hand. So, we give people the power of us doing it fully automatically, or by hand.
Eric Enge: Let’s talk a little bit about the progress in provisioning servers and the general release to the world at large.
Dennis Mortensen: We’ve had the task as part of this launch of moving away from the two existing data centers that we as IndexTools had in Budapest. Those two data centers will be closing down, and all existing clients will be moved to new Yahoo! Colos. We are setting up two Colocations as we speak, one in Europe and one in the US within Yahoo! of course. We haven’t moved over any existing clients simply because this is Christmas, and we didn’t want to disturb or even add the risk of disturbing any of the accounts over Christmas.
We are ready and that means on the very first days of the New Year we will be moving over those 3,000 previous clients to new Yahoo! Colos. The new clients signing up for Yahoo! Stores will be in the US Colo to begin with. Part of that task was, of course, to take out technology and make it compliant with Yahoo’s stack, meaning that there were certain web server software programs and other specific items that we needed to be compliant with. Plus, from a security point of view, there were certain elements that we needed to go through to make sure that we were compliant. To do that, we handed everything over to a global Yahoo! operations team.
Eric Enge: Right. What about availability for new customers that aren’t coming in through Yahoo! Stores. What options do they have at this point?
Dennis Mortensen: Beyond being friends with me? So far, we have launched this for the Yahoo! Stores, and for the advertisers of Yahoo! who have what we call microsites. That means if you actually have a 7-page microsite on Yahoo! they will have this automatically instrumented because there is a platform for us to build this.
We’ve also built this into the Yahoo! Developers platform. So, if you are one of those developers who does widget development on the Yahoo! platform, you will be automatically instrumented with Yahoo Web Analytics as well. These are the first three, and over the course of time we’ll be rolling out a number of new engagements as well. And then, we are going to open up a bit more to the general advertiser at the beginning of 2009.
Eric Enge: Right. After you have the new data centers fully operating the existing legacy customer base.
Dennis Mortensen: Yes. We want to make sure that all the legacy customers are moved over as well, yes.
Eric Enge: So, what about new releases; what’s coming out in the near future for you guys? What’s the general theme of things that you plan to work on?
Dennis Mortensen: I think we have some opportunities. First of all, I like to say that we have not been dumbing down the tool. I know that there has been a concern for some that for us to be able to scale this we had to dumb it down so we could start aggregating data beforehand. People thought that we might move away from our real-time approach, but we haven’t been dumbing down anything. So, that means if you are a new client or one of the existing ones, you have all the capabilities that we provided before plus more.
What we are adding to the tool is some power beyond just the general features. We are trying to incorporate the fact that we now have some opportunities that other analytics solutions simply don’t have because they don’t own a network the size of Yahoo!. So, we are going to incorporate some demographic information and psychographic information into the tool that other people don’t have access to.
Imagine if you take one of your sites and you collect the information with our tool, and you have 10,000 visits. With those 10,000 visits, you can then do some behavioral segmentation on, such as seeing that 5,000 people looked at certain pages. Plus all that good stuff you can do with behavioral segmentation.
But, imagine if I could tell you that 30% are female, 70% are male; that’s one dimension. Imagine, I could say that this many are interested in autos; this many are interested in this particularly luxury segment of autos. And, I know all that information, because these visitors are part of the group of the 550,000,000 unique visitors a month that visit a Yahoo! property, so I have that information.
We have over a hundred different dimensions that we could look into and incorporate into this tool. And, you will see some of those being incorporated quite soon actually.
Eric Enge: So, does that mean that you are going to be somehow sharing information cross cookies or something like that?
Dennis Mortensen: Yes.
Eric Enge: Yeah, so if we try to think about how that might work it gets messy because one application can’t read another application’s cookie, right?
Dennis Mortensen: That’s somewhat correct. But, when you use the Yahoo! Web Analytics tool collecting data, and that in the Yahoo! domain we can connect data with everybody else who is visiting a site in the Yahoo! domain. It does not have to be the Yahoo.com domain as such, but any of the properties that are using the Yahoo! cookie. And, we of course have a cookie on every one of our different sites and we are definitely setting a cookie together with the Yahoo! Web Analytics solution as well.
So, yes it is cookie based, and yes it is not necessarily a 100% accurate, but the sample that we have is so big that it becomes reasonably accurate.
Eric Enge: Right. Well, it’s like anything else in the analytics, right? There is always a certain amount of accuracy error.
Dennis Mortensen: Exactly.
Eric Enge: And, you have to know how to read the data you are looking at.
Dennis Mortensen: Indeed.
Eric Enge: Cool. So, another thing that was kind of interesting is you recently wrote that the Web Analytics feature race was over.
Dennis Mortensen: It is just a strong statement to communicate that it is not as important anymore. I think we, as Web Analytics vendors, like to do a couple features here and some there. Let’s look at the hardware, just to give you a good example to compare it. Now I am not saying that the hardware race is over. But, if I buy one laptop or another, it doesn’t really matter too much, my Word, my Excel, and my browser are probably going to run pretty decently in both of them.
I believe if you go out and buy Omniture, or you install Yahoo! Web Analytics, the capabilities that we have are so far ahead of the capabilities of the analysts or better the general users using the applications. So, if I add ten more features, I don’t think we should expect that I am going to create ten new insights.
I believe analytics vendors have taken not only one step ahead of the market, but we are in fact seven steps ahead of the general market, mainly for the reason that there has been such a huge adoption now, where it used to be for a select few experts, that utilized the tools to their full extent.
Now, you have a market of millions of users who are using a very limited set of features. So, it doesn’t mean that we stop innovating, not at all. It just means that we don’t really compete on whether we have 70 features or 75 features. And, most of us have gotten to a level, where we have all the basics in order.
Eric Enge: There are of course new frontiers like Mobile, where there are things that need to be done differently. And you have rich media. Earlier you referred to the distinction between page views and content views, which starts getting at things like Ajax applications.
Dennis Mortensen: Yes. When I say the feature race is largely over, I like to conclude that it doesn’t mean that we stop innovating. It just means that I don’t think we compete on features anymore.
Eric Enge: Right. You compete on service and flexibility.
Dennis Mortensen: Yes.
Eric Enge: Yes, indeed. So, what’s your sense of what’s hot in the market for analytics, what are the frontiers or people who are really going to begin building things out?
Dennis Mortensen: I think that’s one of the major questions as we are moving away from Website Analytics to Web Presence Analytics, meaning that if I produce content whether its a page, a video or audio, that content is likely to be consumed more elsewhere than on my own site. Essentially whatever I track on my own site is a smaller sample than it could be. Or imagine that this could be a presentation and that presentation is going to be shared over email on slide.com, on other blogs, and on so many more places than on your own site.
So, when I just collect information on my own site, that is probably going to be a limited view. I, therefore, believe there will be a move from Website Analytics to Web Presence Analytics. And how do we do that?
Eric Enge: Right. So, for example, if you syndicate content, being able to tell what’s happening with that?
Dennis Mortensen: Yes. Just an example, I did a talk on the difference between a metric and a KPI and did a Powerpoint presentation that I put on my blog. It had 800 views on my blog as far as I remember, and 9,500 views on slideshare.com. That concludes that only a very small part of that piece of content that I produced was actually consumed where I published it. It was consumed somewhere else 90% of the time, which I didn’t track at all.
That was one site. I Imagine all the sites that took that content and put it elsewhere, put it into an email, put it onto intranets, compiled it into a paper, printed it. So, how am I measuring my content consumed beyond my site, that’s what I call Web Presence Analytics. I truly believe that’s a big question.
Eric Enge: Right. So, another flavor might be if you have a business with a couple of hundred locations or maybe a couple of thousand locations. And, if you are on these local directories in yellow page sites and things like that you might want to have data on how many times those pages are viewed.
Dennis Mortensen: Spot on.
Eric Enge: Right. And, that’s not even a content syndication thing, that’s just a different aspect of understanding how your presence on the web is working.
Dennis Mortensen: Yes, exactly.
Eric Enge: And then, you can compare, for example, the 300 visitors in this time period from super pages that viewed the super page’s profile, how many of those turned into clicks, visitors, and what did the visitors do on your site?
Dennis Mortensen: Exactly. Because you don’t have those impressions today. So when you look at your store locator and say we have 10,000 people a week looking at our store locator. What does that number mean if you have 20,000 looking at the yellow pages to find one of your stores?
Dennis Mortensen: You are completely missing the idea behind and value of such a number, which in most cases, at least in my opinion, are much bigger than what you are seeing on your site.
People are not loyal to a URL; they are loyal to a piece of content. So, if I want to look up a phone number, I am just loyal to finding that phone number, it doesn’t have to be on a particular site.
Eric Enge: Yes. That sits very well in with the notion of focusing website development on very high-value content. That’s what creates more of an issue for Web Presence. I mean, you can have just a straight eCommerce site, and it’s not something that a lot of people link to and you don’t have a lot of need for this expanded thinking.
But, if you’ve done something where you do have a significant amount of your presence outside of your existing website, then you should know what’s going on.
Dennis Mortensen: Yes, I agree. But, I would probably even advocate for the fact that even if I am a highly specialized e-tailer, people will buy your products, and they can probably only buy them from your site because that’s where the checkout process is. But, they will blog about them, they will rate them, they will sell used versions on eBay. You will be measured way beyond your own site.
The way you are measured or the sentiment of that measurement will have an impact on your site.
Eric Enge: Well, that makes sense. So, how quickly do you think Web Presence Analytics will begin to emerge?
Dennis Mortensen: I think we are partly seeing it bubble up right now. Not as one holistic solution, but as small solutions here and there. You can have Nielsen and their BuzzMetric help you figure out how people talk about you in the blogosphere, and what their sentiment is of it. You can have YouTube serve your videos, not just on your site. But throughout the internet, and you can figure out how people interact with your videos.
You can do things in silos, but you have to do have an external data approach here and try to collect all that information manually and figure out what the impact of it was. At some point, you might see somebody collect this into a more holistic solution.
Eric Enge: Thanks Dennis!
Dennis Mortensen: Thank you Eric!
