Measuring Customer Experience Online the HP Way
For that purpose, the company has devised recipes: optimizing lead generation, merchandising and navigation, tuning pages and entry points and analyzing the sales funnel.
"Our charter is to generate additional online revenue, which we track every week, and optimize ROI of marketing campaigns and increase customer satisfaction and self-service," Seth Romanow, director of research and e-testing for worldwide business at HP, told attendees in his keynote session.
The company specifically measures four areas: customer behavior from traffic and click-stream data, conversion, usability and satisfaction. Proper measurement is critical because combined Compaq.com and HP.com have 2 million pages, Romanow said.
To make life easier after the Compaq-HP merger, Romanow started work long before that process to integrate the Web operations, such as the top 250 search terms on compaq.com and hp.com.
"The sites are not merged, but a lot of the search is merged," he said.
Based on his experience, Romanow advocated fewer links on a Web site for more conversion. He recommended a range of techniques to meet goals such as decreasing service cost and increasing traffic and conversion rate.
HP techniques for these objectives include support services such as warranty management and path analysis, sales efforts like merchandising and search analysis, and marketing like ad placement, next purchase prediction and profiling.
Romanow cited three case studies. In the case of site design, analytics applied to the iPaq handheld area of compaq.com led to an 83 percent rise in conversion and 25 percent increase of iPaq options. The company had "buy" links that now go directly to the specific item in the store at smb.store.compaq.com. This increased conversion and the iPaq option pages revenue.
In another instance, HP wanted to raise its click-through rate for its Compaq desktop system. So it set up rotating ads -- called "desk," "math" and "running" in the organization -- on Yahoo with different themes to find the highest click-throughs. It then standardized on the best theme.
After analytics were applied to optimizing lead generation, it found that the "math" ad had twice as many clicks, plus more future orders. Among things analyzed were customer segments and technographics, pogo-stick backtracking, current and future orders, and help and search use.
In the third case study, for Pocket PC, Romanow said 18 merchandising ads were placed and their placements optimized. The incremental performance of each of those ads was analyzed, and negative performers were removed to increase the overall merchandising effort.
In addition, a model was created to identify the fewest ads that account for the highest incremental sales. The remaining ad spots were used for other non-Pocket PC ads to maximize the total revenue from merchandising, Romanow said.
In the end, the optimized Pocket PC ads increased incremental sales 10 percent. At the higher level of sales, half the ads were given up to merchandising other products to create additional incremental revenue. The overall conversion rate was two times higher in 2001.
Reducing clutter was key to higher conversion rates, Romanow said.
"When you've finally made the sale, don't confuse them with another shirt -- you've already sold them a shirt," he said.