Web Analytics: A New View

Web analytics conjures the image of an IT specialist poring over log stats to tally page views and hits. Not so many years ago, this was the reality.

Today, the person studying Web analytics data is more likely to be a marketing expert, or perhaps a business development analyst, or even a customer service leader. The data has changed and expanded far beyond a company Web site. Today, analytics packages track mobile data, e-mail campaigns, information from cloud-based services, and even data coming from external sites such as social networks. All that is combined with Web site information to understand the online customer--not just site traffic.

Straight Web analytics, in fact, has become a commodity. Businesses can get free, powerful Web analytics tools from Google and Yahoo, as well as solid open source options from Internet service providers. Many software tools--including blogging applications, content management systems, and online collaboration platforms such as Microsoft SharePoint--have very good analytics built in. So to stand out, vendors today are touting their tools as something more, such as marketing analytics.

And many businesses require quite a bit more than just commodity Web analytics. They need to know whether their investments in mobile applications and platforms are paying off, and if they should spend more. They need to find out if Facebook and Twitter are useless time wasters or key links to would-be customers. They need to fully combine data from all these online platforms with other stats, such as brick-and-mortar sales and CRM packages, to get a holistic view of the business.

Jing Suk, the audience analytics manager at media and information company Thomson Reuters, says in the bad old days--say, eight years ago--the main criterion for buying a Web analytics product was how many types of reports it could churn out. The more, the better. "Now people who manage analytics, as well as the business stakeholders, they aren't asking for reports, aren't asking for metrics," Suk says. "The key question is, 'How can this data help me take an action? What does it mean? What are the actionable insights?'"

State Of Analytics

In the earliest Internet days, a Web server would generate a massive log file of activity, the analytics package would crunch the data in the logs, and hours or a day later you'd have a report on what had been happening on your Web site.

Next came products using JavaScript embedded in Web pages to send activity information to a central analytics application. Most important, this allowed near-real-time data about what was happening on a site. It also allowed for Web analytics through a software-as-a-service model, which is now the most common way to deliver analytics.

As analytics became vital to any Web business, it sparked a wave of acquisitions. WebSideStory was bought by Omniture, which in turn was acquired by Adobe. Google snapped up Urchin in 2005, a step toward releasing an industry-shaking free analytics package. ClickTracks was bought by Lyris. And earlier this year, IBM acquired the analytics vendor Coremetrics. The deal wasn't a shocker. Even before the acquisition, Coremetrics had built its analytics products to integrate directly with IBM's WebSphere Portal and Commerce servers, to provide visitor traffic reporting and usage analysis.

With Adobe, Google, and now IBM having added analytics to their product portfolios, it won't be a surprise if other major vendors, such as Oracle, SAP, and Microsoft, do deals to boost their analytics offerings as well, perhaps with one of the only companies to remain independent--Web analytics pioneer Webtrends, which was one of the first Web analytics companies and has a history stretching back to the days of log analysis.

The Google Factor

While mergers and acquisitions have changed the face of Web analytics, the biggest shake-up came from a company that has made a habit of causing upheaval in technology markets: Google. When the search giant released Google Analytics, based on Urchin technology, it was much more than a free, no-frills analytics service. This was an enterprise-class offering that provided many capabilities every bit as good as the products being sold by analytics vendors. And most companies didn't have to do much to get the benefits of Google Analytics. If they used a Google service, like AdWords or Google Search, it was a pretty simple step to enable Analytics as well for a site. Since Analytics was well integrated with AdWords and Search, it was possible for businesses to easily track their investments in search keywords and find out how potential customers were finding their online services. That data often led companies to spend more.

Google's free offerings didn't turn out to be a death knell for analytics vendors. But it did accelerate their moves away from pure-play Web analytics and toward broad packages that might be more properly referred to as online or customer analytics. IBM has dubbed its offering an "online marketing optimization" suite, for example. These tools aim to analyze all digital interactions with customers, whether they access a Web site, respond to an e-mail campaign, go to a site via a mobile Web browser, use a mobile application, or access a company's Facebook page or Twitter feed.

One good example of the changing face of Web analytics is Adobe Omniture. In less than two years, Omniture has switched from a company focused mainly on Web site statistics to the broader goal of understanding online customers.

Its core product now is the Omniture Online Marketing Suite, which uses scripting embedded in Web and application code to provide real-time tracking, analysis, and reports on Web, mobile, social media, and video presences. The suite also allows analysis of acquisition (getting customers to your site) and conversion (getting visitors to a preferred goal, such as a buying a product or downloading preferred content).

Just a few years ago, one of Omniture's top Web analytics rivals was ClickTracks. It was acquired by Lyris, which had been focused on marketing software, especially e-mail list and campaign management. This deal made Lyris more relevant, since its focus on e-mail without Web and mobile missed a huge part of online marketing.

Still, it's often a tough sell to get people inside companies to embrace this holistic approach to the online customer--to not just think about improving e-mail click-throughs, but also whether the company has the Web design to turn that click into a sale.

Like Omniture, Lyris has taken a suite approach to managing the challenge of handling data from the many online sources businesses use today. Lyris HQ can analyze data from multiple sources (Web, social media, mobile) and has deep integration with e-mail marketing tools, useful for tying campaigns into actual actions that visitors took on a company's Web site.

Coremetrics has been strong in marketing and campaign-based analytics, with good capabilities for retail sites and products built specifically for industries such as financial services, education, travel, and media. Under IBM, expect more emphasis on cloud-based offerings--in the form of Coremetrics' products as online services, and of tracking and analyzing cloud-based content.

Google Analytics, like its competitors, has the ability to track not only Web site activity but also site visitors using mobile devices and applications, and also to track visitors to company presences on social networks such as Facebook. Google Analytics adds features to optimize and test sites, such as the ability to do bug testing or head-to-head, A-versus-B analysis of site changes, to see how they perform compared with existing designs. Google Analytics can also do segmentation to compare visitors and improve campaigns.

Twiddy & Co. is a real estate firm that also helps vacationers find rental homes on North Carolina's Outer Banks and provides renters with information and services once they're on vacation. It uses Google Analytics' integration with AdWords to know if search keywords it's buying are worth it. Along with that, the most powerful capabilities for Twiddy's Web site are the e-commerce features that show how would-be customers come to the site, and how they come to book or not book.

"The e-commerce really shows you that funnel," says owner Ross Twiddy. "They're coming from these keywords, they're going to these pages, they're looking up details, but they're not booking. What's the obstruction there? What is the obstacle that we need to overcome?"

As one of the original Web analytics companies, Webtrends has been able to survive all the changes in analytics and adapt to them. With Webtrends Analytics 9, it addresses many of the modern challenges of analytics with the ability to track multiple channels of data (mobile, social networks, Web); optimize Web sites through testing and performance feedback; and improve visitor conversion rates. Webtrends also has been focused on improving usability and helping marketers understand the data they're receiving.

Like many other areas of Web technology, there are a good number of open source options for analytics. One is AWStats, which despite being a log file analyzer still can provide results in pretty close to real time. Other open source options worth looking at include Piwik, which works in real time and has surprisingly strong dashboard capabilities, and the Open Web Analytics framework.

Evaluating Analytics

Factors in choosing an analytics package depend a great deal on the type of online business you run. If you have an e-commerce site, the software's ability to show insights into sales is more important than insight on numbers of visitors; if you run a content provider living off ad clicks, number of visitors and page views are very important.

In the end, this all breaks down to a question of usability--especially when it comes to customization and self-serve options. Sure, a big report can look impressive, but when users try to glean knowledge from it, they can find that difficult and tedious.

At Thomson Reuters, one of the critical value factors Suk and her team look at is getting key data to colleagues in real time, without the complexity of conventional analytics reports. People need actionable graphics, not data tables. "Previously, the real-time data was only in the reporting--basically a big table--and for the editorial team it's actually quite a lot of work for them to check the report throughout the day with the data presented in a big list," Suk says. "It's really hard to make heads or tails out of the data."

To solve this problem, Suk's team used the Webtrends Analytics package's API to build a visual module so editors could track the real-time traffic results of content on the site.

That solved this problem. But in general, Suk would like to see much better ability to build custom dashboards for Web analytics, and easier ways for her colleagues to get their own information. "We're a small team, and we manage a large set of stakeholders," Suk says. "You can never really fully support a big company with a small analytics team unless people can self-serve their own needs."

Put any Web analytics suite through this quick usability triage:

>> How good are the dashboard and alerting options?

>> Can key information be broken out into modules that can be delivered in multiple ways--say, portals, collaboration systems such as e-mail, and even mobile devices?

>> Is it simple enough that people not steeped in analytics software can easily configure it to see at least the information they need most often?

This usability test doesn't hold just for basic information users. Strong usability features like visualizations make it easier for a business to take advantage of advanced aspects of the analytics software as well.

The Revenue Connection

When it comes to business goals for Web analytics, one of the most important is using data to spot ways to increase sales and profit. For this, there are several key areas to look at.

A good analytics package should provide insights on advertising and search keyword buys. Which keywords are sending the most profitable visitors to your site? Which e-mail and other marketing campaigns are resulting in valuable transactions? Your analytics package should be able to provide this information and also make it possible to do comparative testing in order to see in real time which campaigns or site redesigns are having the desired impact on visitors and customers.

Many of the newer analytics packages can go beyond simply providing data and can also--once integrated with content management systems--generate automated responses to visitor actions. Suppose a visitor perusing a product page for a while leaves without a purchase. Can the system send an automated e-mail with special offers to bring her back? Almost all of these types of capabilities rely on shoppers having a login and user account set up and use cookies to know what a person does on the site over time, and in real time. Have someone arrive to your site who's been searching for skiing gear? Make sure the home page he sees emphasizes winter sports gear and not kayaking equipment.

Of course, such automation requires integration with Web systems, so make sure the product you choose can integrate with your content management system or Web development platform and your campaign management applications. Ideally, the vendor already has done that integration.

In general, the more integration options the better. Does the product have import and export options so that its data can be read in other products, or it can import data from other systems? Can it take data from legacy point-of-sale and brick-and-mortar systems? Combining this data with online data can provide a holistic view of a company's customers.

Taking on huge importance today is the ability of these products to handle the mobile Web. Most products have some form of mobile tracking, but this can vary widely. Mobile measurement can be very difficult as it isn't always possible to use cookies or JavaScript to track mobile visitors. Because of this, look for products that have APIs and software development kits that make it possible to add tracking directly to mobile applications.

The tracking capability is embedded in the application, typically through a plug-in or instructions for adding the code within the mobile environment. This approach embeds the tracking ability within the code of the application instead of embedding JavaScript within the HTML code of a Web page.

Also of growing importance is the ability to track content from external sites, especially social networks like Facebook and Twitter. Most analytics products will let you add scripts your company's Facebook or Twitter page, while some products have deeper integration into networks like Facebook, making it possible to directly track applications within the network.

Challenges Ahead

While acquisitions and new Internet platforms have reshaped the market the past five years, we expect a faster pace of change in the near future, as social networking and the mobile Internet drive even more interactions, research, and transactions online. Already companies are looking for ways to take into account opinion metrics and chatter on external networks.

Using text analytics, internal analytics can reflect not only what people do on your site but also what they're saying about your company and products on blogs, Twitter, and other Web sites. In addition, as more people use mobile devices for a large part of their online activities, it will be vital for analytics products to handle this data every bit as well as they handle Web data.

But the biggest challenge of all may be fragmentation. An increasingly common usage pattern is to be on a smartphone on and off all day, a work computer during business hours, a personal PC for paying bills at home, a tablet while relaxing in the family room, and an Internet-connected game console once friends come over. One person may access your Web site or encounter your brand using any of these devices in the course of a day, but your analytics package won't know that it was all one person. Future analytics packages will need to come up with some way to solve this challenge if businesses are to keep on understanding their customers.

Marketing's becoming an ever-more data-driven discipline, particularly when it comes to online interactions. But quality of insight doesn't necessarily move proportionately with the quantity of data, so be careful, warns Suk at Thomson Reuters. People in marketing and elsewhere are getting increasingly savvy about Web data, but they don't want an analytics system that piles on features and reports. "They just want to know the most important things as quickly as possible," Suk says. That's where marketing and IT need to work together to get their Web analytics strategy right.

Hurdles To Good Web Analytics

1. Canned Reports Sure, the software has a lot of them. But if they're nothing more than lists, pulling useful data out is nearly impossible.

2. Data Overload Not only from the analytics package, but from other packages--content management, CRM, etc. Thus, the need for visualizations, not lists of data.

3. The Mobile Web It's a vital channel, but proliferation of mobile platforms and the difficulty of tracking mobile users make this tough.

4. The Modern User People move across smartphones, work and home PCs, tablets, games. Tracking them as unique visitors is impossible today.

5. Not Getting The Full View If you're not combining data from Web sites with other channels (brick-and-mortar, customer support, CRM) in some analytics or business intelligence, you're not getting the total picture.

6. Failing To Convert Analytics should help you spot which online sites convert visitors to customers, and those that fail to. If you're not using analytics for this, you're missing its most powerful role.

7. No Automation The most effective online businesses use automated tools tied to analytics to respond to visitor actions immediately.

8. Staying On-Site A company's online presence must go beyond its own Web site, with content and touchpoints on Facebook, Twitter, and other external networks.

9. Not Listening From Twitter to social networks to blogs to e-commerce sites, people on the Web are talking about your company and its products. Your business must track online sentiment.