"People love testing, but it's expensive and often unguided. Survey research is a great way to test persuasive messages with little or no cost or time. Research can lead to more effective site testing"
HP has been faced with some of the most dramatic shifts in corporate and product landscape imaginable:
Their goals is to show how online customer research that focuses on more than Website navigation can be used to understand the impact of major business changes, drive better understanding of customer dynamics, and focus business strategy
In the course of a few weeks, HP announced a huge, fundamental shif in its business, with dramatic and potentially negative impacts for the online store:
Sales took a significant dive. They hypothesized that the news caused the drop in consumer purchases, but were unsure as to how exactly and how much the news caused the drop in sales, so they mapped a response.
They deployed an online survey, using Vovici, and collected over 3,000 customer responses from website shoppers in just one week, using very aggressive tactics that are often frowned upon. The cost was less than $10,000, and inception to delivery of analysis was less than 3 weeks.
Data mainly reflected customer sentiment. The survey was very short in order to encourage people to complete the survey. There were 11 questions, three or four of which were demographic. Then they asked people whether they were aware of the announcement, and how deep their awareness was. Knowledge of the announcements was probed to determine how it affected customer sentiment. They wanted to see, where there was an impact, who was the most negative.
Since only those who have a good enough sentiment to want to come to the company store are ever presented with the survey, you're not getting a representative sample. As a result, it's very important to understand those with a negative sentiment. These are the people surveyed who most closely represent the demographic you want to know most about. It's crucial to be aware of the skew with web intercept surveys, and to figure out how to use it.
Overall, they discovered that the news had not affected customer sentiment as seriously as they had believed, and that there were opportunities for mitigating any negative effect, particularly among corporate customers.
Gary Angel points out that in a presidential election 55/45 is considered a landslide victory. A few percentage points one way or the other does make a significant impact, and in opinion research, it's important to look at small differences between groups or in data points.
Careful analysis of the survey data revealed some interesting aspects to the impact of the news. Some Key Take-Aways:
These results showed that the situation was not as direct as initially thought. Although the data is biased and incomplete, the data helped show how the company was affected, leading to a plan on how to modify the company to raise sales again. Instead of implementing an extensive program to help sentiment recover from bad sentiment, it was possible to formulate a different plan which was more effective.
Buying patterns around laptops have changed dramatically. One likely culprit – the Apple iPad. How much cannibalization of laptops is there with HP customers and where are the key drivers of choice?
The big question of the day: Are tablets truly an alternative to laptops? For everyone? When and why are they or aren't they?
The survey asked about 13 questions, beginning with demographics and then moving on to understanding product interest and examining what drives customer choices. Visitors were asked what product type they were interested in, whether they have a model in mind, whether those interested in a laptop might buy a tablet, and if so, what would sway their decision towards a laptop instead. It's important to note that the survey was responsive, meaning if a visitor expressed no interest in buying a laptop, the survey ended with just four questions, meaning that only relevant data was gathered. Anyone not in the group of interest is immediately eliminated. The 'model in mind' question is a good way of gauging how far along in the purchase process they are.
One important thing they discovered is that those considering tablets were people with high income and low price sensitivity, meaning that customers from this segment may well represent a higher percentage of revenue than customers from other segments.
The landscape has changed radically. Laptop penetration has almost reached desktop levels. About half of all laptops purchased in the US are purchased online. Laptops, which were once considered a gift given at holiday time or bought as "back to school' gifts, are now considered "Durable Goods" rather than luxuries, so there is less seasonal change in laptop sales. The US economy has also had difficulty. All of these affect laptop sales, meaning it is a mature market. Desktops have gone through this pattern as well.
As a product matures, speaking to customers about it requires a new language and strategy.
In this market, where overall laptop sales are flat, it's important for marketers to focus on improving branding and focusing efforts not on selling laptops but on selling HP laptops.
The survey asked people to choose how to describe what they're looking for – looking for inexpensive, traditional value, and people looking for image-related products. This language was changed repeatedly. Almost all the interesting findings were discovered by cross-tabulating different variables.
Andrew leads the Analytics organization for HPDirect, HP's WW B2C eCommerce organization, which is #17 on the Internet Retailer Top500. He's spent the past four years building a multi-disciplinary Analytics team with locations in both the US and India. HPDirect is a fully integrated Manufacturer direct online website and is the only place where customers can buy custom configured HP PCs to their exact specifications.
Previously, Andrew has worked in Sales, Marketing, Channels, Business Development, Strategy & Planning, as well as Online & Retail Merchandising for HP. With his extensive business background Andrew drives a very pragmatic approach to Analytics that always starts with a clear understanding of the business problem and ends with actionable insights that drive tangible business results.
Andrew is consistently ranked in the top ten percent of all managers at Hewlett-Packard.
Gary is Co-founder and President of Semphonic, one of the leading web analytics consultancies in the United States. Semphonic’s clients include some of the largest Financial Services, Media, and Health and Pharma companies in the world. Clients he manages at Semphonic include American Express, Charles Schwab, Nokia, Turner Broadcasting, the BBC and Genentech.
Semphonic concentrates on web analytic verticals without a clear ecommerce transaction component and has built substantial practice specialties around lead assessment and integrated marketing communications, web analytics data integration, SEM Analytics, mobile measurement and measuring engagement.
Gary is a frequent speaker at analytics events (and hosts Semphonic’s X Change Web Analytics Conference for enterprise analytics managers), publishes frequently on DM Review, ADOTAS and a variety of print and online publications, and blogs at semphonic.blogs.com/semangel/.
A graduate of Duke University, Gary has 20+ years of experience in data analytics for marketing and operations and lives in San Francisco with his wife and two daughters.