Upstream, Downstream, Clickstream

OMV analyzes Web usage with SAS^® Web Analytics solution

OMV is the leading oil and gas group in central and eastern Europe, with 2002 annual sales exceeding €7 billion (US$8.7 billion), and 6,200 employees worldwide. Its businesses are active both upstream and downstream, with exploration and production worldwide and 1,700 petrol filling stations in 12 countries. OMV's BroadVision-based company portal plays an important role in the company's communications strategy, providing information to a range of interested parties."Although we are hardly in e-business – it is difficult to sell petrol over the Internet! – as an organization we are very interested to find out who is visiting our company Web site and why," comments Martin Robl, project leader of the Portal Analytics Solution (PAS) at OMV.

Running an effective and successful portal depends on getting timely information to broadly three types of information consumer. Content managers want to know how many and what sort of people have seen the content they have published. They want to know which are the most popular pages and how long visitors stay to read them. Portal managers need figures on the general traffic on the portal: how many people visited, where they came from, which campaigns were most and least successful, and other relevant aspects of portal usage. Finally, portal administrators want to see figures on current data volume, and they want to be able to forecast trends to ensure that it has the capacity to meet future traffic loads.

Planning Ahead for Volume
OMV's enterprise portal was launched in September 2002. It is quite unusual in that it successfully integrates the company's intranet, extranet and Internet pages through a single site, www.omv.com. "Many portal and Web site operators underestimate the size and scope of the task of providing appropriate Web statistics and reports," says Robl. "We did not want to make that mistake."

Therefore, early in 2003 OMV teamed up with SAS on the PAS project to provide standard and custom statistical reports together with sophisticated analysis of portal activity. This would provide the information necessary for the strategic development and planning of the portal – for example, by enabling OMV to structure content according to the profiles of particular target groups (personalization) and by developing the portal in line with the e-business requirements of both OMV and target groups.

These business partners are not necessarily just other companies, but rather interested parties or groups of people with whom OMV comes into contact through the portal: customers, shareholders, employees, suppliers, subsidiaries and joint venture partners, governments, banks, journalists and interested members of the general public. In other words, OMV's portal attracts a highly diversified audience, with different information requirements and – another key variable – speaking a wide range of languages.

"Web use has grown very rapidly over recent years," says Robl. "Visitors do not necessarily want more information than they did in the past, but they do expect higher quality of information, and better service."

Personalization is key. It's only when you know who the visitors are, where they come from and what their main interests are that you can really meet their expectations.

Building on Past Successes
All the raw data to answer these questions exists in the Web content management logs and in observation log files. The basic requirement of the PAS project was to integrate that data and to present it through standard and ad hoc reports. Robl explains why OMV selected SAS Web Analytics for the project. "Essentially SAS offered the best proof-of-concept for OMV's PAS project. Our two organizations have worked together successfully on various projects since the 1980s, so we have built up considerable SAS expertise in-house. SAS Web Analytics offers all the key strengths of SAS software with which we are very familiar, such as expandability, scalability and ease of integration."

The project started on 7 January 2003 and was completed by 20 March 2003. PAS now delivers reports on 50 content groups (internal and external) to 15 content managers, covering statistics on the number of sessions, page views, average dwell time and top 25 content pages. It also delivers a set of standard reports meeting portal managers' needs for information on traffic, session duration, forecasts, searches, referrers and so on. Finally, IT reports give the portal administrators the information they need on current volumes and trends.

OMV archives these reports over time, organized in time-period hierarchies so that the most recent reports are most rapidly accessible.

Easy-to-Access Reports
The reporting process is two-tiered. Each day an extract, transform and load (ETL) process moves data from the database and observation logs into SAS data marts. SAS Web Analytics then performs a second extract and load to integrate the data from the Web server logs and to separate processed data into two Web marts – one for traffic on the external portal and one for the intranet traffic. These processes generate standard reports that can be easily accessed from an ordinary Web browser; if the user wants to dig deeper with visual analysis and ad hoc reports, he or she can easily do so with the SAS OLAP tools or SAS Enterprise Guide, accessing the Web marts directly.

"All of the data processing and report generation is performed automatically each night so we get refreshed reports every morning," says Robl. "It sounds simple, but that's only because SAS software is so effective at bringing order to complex environments. In fact, the OMV portal had a highly complex technical infrastructure featuring several Web and application servers, load balancers and a clustered Oracle database in a high availability environment. So OMV PAS was a challenging project that played to the strengths of SAS for integrating large quantities of disparate data."

However, the project has already delivered key benefits and return on investment. For example, search-string analysis is enabling OMV to find out what sort of information the different categories of internal and external visitors are looking for, while clickstream analysis reveals how quickly and easily they can find it. By acting on this information, OMV can increase the satisfaction of their external target groups as well as the portal usability for employees. "We soon found that one of the most frequent searches among employees was for an organization chart, so we have now made this available through a single click," says Robl.

Portal statistics are also helping OMV to direct and control its marketing activity. Analysis of referrers shows OMV the effectiveness of online banner advertising and similar activity. "We are now able to verify the success of these campaigns and independently determine whether we are getting value for money," says Robl.

The next steps in the PAS project are e-mail campaign analysis, analysis of the success in converting guests to registered users, cost analysis by page view, user and employee, application-specific analysis and much more.

"PAS has already saved us a lot of time and resources, showing us where we can invest our future creative and technical efforts in the portal to maximum effect. More importantly, it has turned up a lot of useful information that will help us serve customers better by improving the usability and quality of our Web pages. We now have to focus on exploiting that information for greater competitive advantage," concludes Robl.

Copyright © SAS Institute Inc. All Rights Reserved.