The Big Integration

This week, I am very happy to welcome Jonathan Levitt, Vice President Marketing of iPerceptions, one of the very top applications in attitudinal analysis, often referred to as Voice of Customer analysis. Since iPerceptions and ForSee Results are generally seen as the leaders of the field, and since I interviewed Larry Freed, CEO of ForeSee, last week, I thought I would ask Jonathan the same questions, and see how iPerceptions the behavioral/attitudinal mix can bring more added value than each of them taken separately.

JW – Could you tell us about how online attitudinal analysis has evolved in the last 2, 3 years, and where it is now (adoption, evolution, etc.)?
 
I think online attitudinal analysis went mainstream in the past year or so. We’ve all seen the hallmarks of its maturation: widespread penetration in all verticals, integration with behavioral research, firms specializing in specific sub-disciplines like open-ended quantification.

Basically, at some point everybody recognized the limitations inherent to clickstream web analytics and by now most of the big players have augmented their research toolbox with some kind of VOC solution. Having said that, I think the honeymoon period for attitudinal analysis may be drawing to an end. In 2007 and 2008, it was enough to have a VOC program ongoing without worrying too much about ROI; listening to real visitors was basically seen as an end in itself.

Now, with the scary economic situation, the game has changed. There is pressure on attitudinal analysis vendors to justify their worth, to prove their technology’s contribution to corporate bottom lines. As a practitioner, simply collecting and reporting out bland dashboard metrics–things like customer satisfaction scores, net promoter scores–and then patting yourself on the back isn’t going to be enough anymore. We’ve seen how this painfully deep consumer recession and the Wall Street crisis have wrecked companies like AIG, Wachovia, Bank of America, and CitiGroup–firms that had rock-solid customer satisfaction scores for years. So, metrics alone–even if they are directly sourced from real visitors–won’t save you.

Instead, I think the focus is shifting towards visit success and task completion. Companies want to know what tasks their visitors are seeking to accomplish, how they’re doing in helping them complete these tasks, and, most importantly, what they need to fix to boost visit success. When I talk to C-Level decision makers, I keep hearing the same thing: provide me with that one golden insight that will help squeeze more revenue from my site or help me stall the churn bleeding. Online attitudinal analysis is still better positioned that clickstream analysis to deliver this, but it requires vendors to think like business owners instead of ritualistically reciting dry and impersonal metrics.

JW - We are interested in data integration, can you tell us how iPerceptions can make its data available? Any API, or export schema?

Without infuriating our development team too much, I can promise you that APIs for both our enterprise-class and free 4Q services are most definitely on their way. I think this will be far and away the most seamless way to blend and mash-up iPerceptions data with data from other sources. Right now, we’re relying on a series of ad hoc measures–much as other vendors are. We’re using parameter-passing, FTP and direct downloads, and online file exports to do the job, but the flexibility of an API cannot be matched and we’re very eager to roll that out.

JW - We would like to know if it’s possible to integrate iPerceptions data with applications such as WebTrends or Omniture.

It certainly is. We have bilateral relationships in place with all the major web analytics vendors, making it very easy for an iPerceptions client to mesh together data sets based on a unique identification variable.

JW - Is it possible to somehow link respondents to what they actually did on the site? If yes, how do you technically accomplish that?

This is the first question that companies put to us. It’s important to note that a lot of interactive marketers and web analysts don’t have backgrounds in offline marketing research, where surveys have been going on since time immemorial. As such, some of them get a little bit nervous about whether or not their respondents are being fully truthful; they want some way to calibrate this with actual activity on the website. We’re more than accommodating and it’s actually a very easy process. If we’re dealing with a client who’s plugged into a BI platform like Omniture’s Genesis, then it’s really just a matter of setting up certain routines and then the machine basically runs itself.

JW - Without naming the client, could you give us an example of a company that actually linked their behavioral and attitudinal data? What was the most interesting finding that could not have been possible with only doing the survey?

I think most of the success stories in these types of situations all follow a similar tack. Company A has a particular series of behavioral segments that they want to analyze and contrast. Say they want to measure offline purchasing likelihood for visitors linking from paid search vs. display ads. So, essentially, they will graft survey data onto pre-established behavioral segments. They’ll do this until they’ve answered a question, and then the integrative research will reach a terminal point.

Most success stories tend to feature some variation on this schema. Our most exciting success story, however, came from a client that flipped this model on its head and did things the other way around. They took all the visitors would had indicated (through surveys) that they had successfully completed their primary tasks and those who didn’t and they grouped them into two distinct segments. They then built a composite sketch of these segments: age, gender, intent, top referrers, click paths, recency, frequency, etc. In the end, they discovered that the segment representing visitors who DIDN’T complete their tasks contained a huge proportion of several key target demographics. Conversely, their site was working at peak efficiency among visitors for whom they weren’t even optimizing. It was a massive culture shock for that company, but it smashed some unhealthy tenets that had been dearly cherished there. 

JW - How do you see a solution like yours evolve in the near future in terms of better integrating its data with other enterprise customer data systems?

We’re always working to iterate, to improve, and to innovate in this area. Ultimately, I think the future will bring a broader reach for integration. We’ll go beyond marrying behavioral and attitudinal silos. The reach of integration will expand to include CRM, POS, and ERP systems (and many, many more acronyms!), as well social data coming from new standards like Facebook Connect. And then, we’ll all have to figure out a way to grapple with integrating the ocean of data coming from ubiquitous mobile web access. We are definitely entering an interesting and challenging era for integration.

JW – Thank you very much Jonathan!

Today I am very happy to welcome Larry S. Freed, President & CEO of well-known ForeSee Results, one of the leaders in attitudinal analysis, often called Voice of Customer analysis. Larry was kind enough to find some time in his very busy schedule to answer a few questions about how behavioral and attitudinal analyses can be integrated to offer extra added value.

JW – Could you tell us about how online attitudinal analysis has evolved in the last 2, 3 years, and where it is now (adoption, evolution, etc.)?

 LF – We had our first beta client right after September 11, 2001, so we’ve been doing this work for seven or eight years now. I think five years ago, even two or three years ago, analyzing attitudinal input or voice-of-customer feedback was a nice-to-have. Only the really forward-looking, advanced companies did it. What we see now, especially with the weak economy and fierce competition for share of a smaller and smaller wallet is that attitudinal analysis is a must-have. There are a lot of academic studies showing that customer satisfaction is predictive of future financial performance, and companies really can’t afford to ignore it anymore. So we see more and more companies starting not only to collect and pay attention to attitudinal data, but when they can apply science to it, they are also using it to measure success and drive decision making. There is huge demand for actionable analytics that actually help companies make decisions rather than just sitting there in a spreadsheet, showing what already happened. Adoption of attitudinal analysis is fast and growing every day. The companies that aren’t doing it will be left behind.

JW – We are interested in data integration, can you tell us how ForeSee Results can make its data available? Any API, or export schema?

We make our data available through an FTP file transmission or clients can download data through our online portal. We can provide all the schema that people will need– file export schemas are available.  

We’ve actually created an exchange mechanism to take other sources of data in, be it behavioral or financial data; we can import those other sources into the ForeSee Results portal and marry those sources with our online satisfaction data.

JW – We would like to know if it’s possible to integrate ForeSee Results data with applications such as WebTrends or Omniture.

Yes, we do integrate with companies like Web Trends, Omniture, Coremetrics and other clickstream analytics tools by providing a common linkage point at a respondent level between ForeSee’s system and the clickstream analytics’ system. We’ve done this numerous times across various platforms.

JW – Is it possible to somehow link respondents to what they actually did on the site ? If yes, how do you technically accomplish that?

Absolutely, we have the ability to link either by integrating with the clicksteam analytic tool, with a company’s internal behavioral/financial applications, or through passing of parameters from the website to the ForeSee system.

JW – Without naming the client, could you give us an example of a company that actually linked their behavioral and attitudinal data? What was the most interesting finding that could not have been possible with only doing the survey?

There are a lot of great examples, and it’s been really instructive to see the kinds of things people can learn by linking behavioral and attitudinal data. It can be really simple—we had one client who was able to look specifically at browsers who abandoned a cart to find out what they were likely to do next—whether they were planning to buy from that same company in a store, buy from a competitor, return to purchase later, etc. This company was really able to put the cost of an abandoned cart into perspective when they saw why people were abandoning. Another great example is a company we worked with that was able to understand the value of improving satisfaction by linking satisfaction to purchase data. We know that satisfaction is proven to predict future financial performance, but integration actually allowed this company to prove that out and show how revenue increased as satisfaction increased. A third example is a company that was able to segment satisfaction and likely future behaviors based on referring site and keywords. They were able to see which referring sites resulted in the best and worst quality visitors. They could actually see that visitors who came to the site via one keyword were more likely to buy than those that came through another key word. This is invaluable information when it comes to marketing decisions and resource allocation.

JW –  How do you see a solution like yours evolve in the near future in terms of better integrating its data with other enterprise customer data systems ?

We’re actually just coming off a period where I feel we’ve made tremendous strides in integration. But the industry is always evolving, and our goal is always to make it as easy as possible to extract data from the clickstream tools.

JW – Thanks a whole lot, Larry.

Today, we got a top guest for our tbi interview : Barry Parshall, VP of Product Strategy at WebTrends. Barry has been at the heart of product evolution for several years. Prior joining WebTrends in 2001, he ran Crystal products for 4 years. Barry talks to us today about WebTrends Connect within the context of web analytics and BI integration.

I’m very happy Barry could take the time to answer the following questions. Each one of them would have stimulated many others, but this is a blog after all!

JW – Tell us more about WebTrends Connect. It doesn’t seem to be an actual product, nor a module of WebTrends Analytics.

BP – Connect is a collection of components (APIs, libraries, data models, drivers, etc.) that are an intrinsic aspect of all WebTrends Products. Connect is geared to fully expose the underlying data that drive our customer’s businesses and our analytics products. More importantly, Connect is the manifestation of our corporate philosophy predicated in customers being active participants in the development of their and our solutions to improving their businesses. A good illustration of this is our upcoming customer conference in which we provide a forum for customers to help design the agenda and content.

JW – What do you see as more common? Companies exporting aggregate data from WT and dumping it to their warehouse after some ETL, or using their BI application to connect directly into WT?

BP – Intuitively I’d say it’s a 50/50 split, as we haven’t performed a formal analysis. We know many of our customers use our ODBC driver to build their own Excel scorecards. More and more, though, customers are directly loading their Teradata and other warehouses with our WebTrends Marketing Warehouse data to garner a more complete view of the customer’s behaviors and transactions.

JW – What are the respective roles of WebTrends Analytics and Marketing Warehouse/Visitor Intelligence in the Connect/BI environment?

BP – Both products are supported with the same drivers, APIs, etc. – the difference is in the data they provide. WebTrends Analytics is an aggregate reporting solution that it provides operational reporting needs to help direct all aspects of your online marketing strategy. As such it contains much of the data to populate dashboards, scorecards and daily reports in external spreadsheets, portals or reporting systems.

The Marketing Warehouse provides a detailed accounting of every action taken by every visitor on the site. Just like a CRM system maintains a record of every phone call, every email, indeed every interaction with a customer, Marketing Warehouse maintains a physical data record of everything the visitor has done on your site. The data itself is stored in a SQL Server 2008 database with a fully documented and exposed database schema (as well as a library of abstraction objects). The role of the Marketing Warehouse is to provide the non-aggregated detail visitor-level behavioral data needed to drive the specific audience segmentation, data integration and rich analysis requirements of a given organization.

Visitor Intelligence is a packaged business intelligence solution extending the Marketing Warehouse. It utilizes a true OLAP data store and OLAP slicer/dicer component, along with other BI components and or course pre-built cubes and reports. It’s built on top of the exact same data models and APIs exposed in Connect and is a great tool for companies that don’t have an existing BI system they wish to utilize instead.

JW – How do you see the role of a high-end product such as WebTrends within the BI suite in enterprises? In other words, do you see WebTrends as becoming the application of record for Web – BI analyses? Doesn’t Analytics run the risk of becoming a kind of Web data pre-process application ?

BP – We believe that both approaches will be used for the foreseeable future. Many organizations will continue to operate their online marketing optimization within the silo of web analytics information – it’s a proven model and it works for basic marketing and operational needs. But for an increasing number of organizations, online data is seen as one piece of a broader business intelligence solution. These are organizations with firmly established practices, processes and investments in enterprise BI systems. For this organization, the reporting solution provided by the web analytics vendor has increasingly marginal value. It’s the data that’s gold, along with what can be done with it. There’s a reason why there isn’t a universal business intelligence solution … every organization has different BI needs. Packaged solutions like we see in the web analytics market can provide much of an organization’s needs out of the box, but as organizations become more sophisticated, requirements grow beyond what a typical vendor can provide in a “packaged” offering. This is when the data (un-aggregated) and the ability to leverage entrenched BI systems becomes vital.

So will web analytics turn into a pre-process application? To some extent, yes, in that some organizations will want nothing else from the web analytics provider (“just give me the data”). But I think a better way of thinking of this is that web analytics will adapt to greater business value which we believe is richer forms of analysis. Basic reporting tools, which are well down the path of being commoditized, will give way to more advanced analytical applications, such as those that mine relevance with high-value audience segments, statistical analysis of campaign attributes across all prior touch points, and so on.

JW – Can you comment on what WebTrends Connect is bringing to the table compare to Omniture and Coremetrics’ data integration efforts?

BP – What makes WebTrends unique with Connect is our commitment to industry-standard technologies, open and published data models and both deployment options (hosted and software). As a rule, data warehouse and business intelligence savvy organizations are frustrated by proprietary data stores, flat data dumps or any hindered access to the original detail data via rich data objects. All vendors provide APIs, but if the data has already been aggregated, it thwarts many of the integration and analysis needs.

JW – Thanks a lot, Barry.

I just saw an Omniture’s press release announced on Twitter for the launch of Discover OnPremise for Retail. You go and read it if you have a minute. What I find interesting is that Omniture is aggressively positioning themselves as the application at the center of all the online/offline integration. This is very important of course. In an integrated customer data world, what will be the application of record for analyzing all that? Obviously, Web Analytics applications need to be assertive here and demonstrate they could be that application if they don’t want to become mere web data pre-processing mechanism that ETL the data to the data warehouse, while a BI app takes care of the valuable insights.

Since Discover OnPremise is the new name for Visual Sciences, which was the most expensive Web Analytics applications out there at the time (hence the two dozens clients they had), I wonder how much that product is for retailers. I would tend to think that beyond a few hundred thousands dollars, retailers will prefer to connect to the web data and take care of it within their existing BI installation.

Well, well, WebTrends confirms its strong orientation toward data integration with the nomination of its new Vice President of Products, Casey Carey, who’s coming from the database world.

You can read the press release here.

Do you have already laid out your data integration plan for 2009? No? Hmm…

From Kevin Hillstrom:

Web Analytics experts must integrate their tools and data with all company data, or risk becoming irrelevant as the Business Intelligence community integrates data for them.

I would suggest that by “web analytics experts” here, Kevin meant the application vendors in particular (WebTrends, Omniture, Coremetrics, etc.). Obviously, as practionners, we do need to seriously start doing the big customer data integration soon. It should be at the very top of our list in 2009.

Today I finally got my toll free number on my business site. I have wanted one for some time for a simple reason: measuring how many phone calls I was getting from people on my Web site. You probably remember I touched that topic some time ago, and I wanted to do what I preach.

I have so many times heard Web teams telling me they knew their site had “echo” effects on the business; sales that were closed offline, with all the grunt work (i.e. persuasion) done online. I would tell them to publish a number exclusive to the site, so they would know it would came from the site every time the phone rang.

Is it that hard to get? Well, you’d be surprised to see how people think it’s complicated. Rather depressing…

So, with the launch of my new site, I thought it was finally time to take care of that little detail. I am a one-person operation, but not alone if you get my drift. There are thousands, millions of companies in a similar situation. You’d think there would be a service to accommodate a simple need such as a toll free number.

Well, well, not in Canada. At least that’s what I thought until today.

I called Rogers (my current carrier), who could not change my individual account into a business one, and I needed to be a business account to get a toll free number. I called Telus, who said they would only provide one if I had at least $100 worth of fees a month. I said I would be willing to pay a hundred bucks minimum, regardless of the number of calls I would get (you see? I wanted it that bad! I’m obsessed with measuring. I guess I’m sick). They never called back. I then went to Bell Canada’s site, placed a demand online, and never heard from them.

I am actually naming them because I want to see if they do any kind of reputation monitoring and analysis. Guess what…

No big thing you probably think. Why make such a fuss? Again, I wanted to know, I need to know if that darn site works, and I can’t measure it only from the number of emails I get. OK, I could ask when people call (or plainly guess they were on my site when they are calling from Europe, I imagine), but I wanted hard data.

Being left alone in the dark by the big 3, I had given up hope.

But today, tah dah!, I came across My1Voice from my current fax supplier MyFax (disclosure to traditional marketers: I’m talking about them not because they give me any incentives, but because I like their services). It took me 5 minutes to setup an account and publish my new toll free number.

One has to appreciate the disconnect between my need, I believe a common one for all of us working independently, and the traditional phone carriers’ ability to address it. I’m no specialist of the telecommunication market, but this new mobility (or nomadism to use a new buzzword) the Web is bringing to hordes of workers is the fundamental dynamics of an unbelievably huge market.

What percentage of visits to my site will generate a phone call versus an email? That remains to be seen…

My my! You know when you read something, and it expresses exactly what you have had in mind, but could not put into words with precision and eloquence? Well, it happened to me today when I read Jim Novo’s latest post.

Notice the fundamental shift from analyzing Visitors to analyzing Customers.

Go read it now.

I just wanted to redirect you to a new series of posts Gary Angel is starting about the problems related to integrating Web Analytics data into a datawarehouse. Certainly the deepest text I’ve read so far on the topic!

You can read it here.

If I may, I’d like to suggest that you put what Gary says in light of this, and I believe you will notice that a new web data analysis landscape is slowly forming…

It was an important week in terms of Web Analytics and Customer Analytics data integration. Omniture announced its new Omniture Developer Connection, which seems to be quite promising. Surprisingly, it didn’t stir much attention in the WA blogosphere (For reactions, see here, and here). Could that be any indication of how low customer data integration is on the WA community radar? I hope not; it may be that yet another Omniture news slipped through the crack of our everyday hectic life.

WebTrends is joining the dance with its coming release of WebTrends Analytics 8.6 (not yet announced). From November, its On Demand clients should be able to more easily export large quantities of data to other systems through the new WebTrends Connect component.

I really look forward to playing with both systems.