Gathering heaps of customer data is well and good, but data alone will not increase your bottom line.

For Brian McCloskey, hpshopping.com's Director of DM infrastructure, the answer was to go from a product-centric focus to one that focuses on the customer.

Right now he is half-way through a three-year project to rebuild hpshopping.com's marketing infrastructure in order to improve customer interactions. In this exclusive interview, he reveals what he's learned so far, and why CRM alone is not the answer.

“A couple of years ago people were buying killer applications to attack CRM. We didn’t do it that way,” McCloskey explains.

"We knew that to build e-commerce relationships, it was all in the data about the customers. We recognized over a year ago that we were building a lot of info about our customers; but, we did not have a good understanding of who our customers were, and we weren’t cultivating relationships with them.”

-> Step 1: Rebuilding the marketing database

McCloskey's first step was to build a better data management infrastructure. One that allowed for merge capability, data cleansing, and, ultimately, aggregation, so that he would have a single view of each customer within HP and hpshopping.com.

Then he tied in data feeds from four systems HP was already using to manage accounts:

a. Sales from multiple channels: Whether the customer ordered online or through a print catalog, the sales were now counted by the same system.

b. Lifetime value data: HP already had a customer valuation process which looked at the customer based on all of the customer’s past purchases and HP’s investment in that customer, and then predicted future purchases and built an individual profit score.

c. CRM data: McCloskey fed in data from E.piphany's CRM system which HP used to get light analytics, as well as campaign management data for both offline and online campaigns.

d. General customer groups: HP had already grouped customers into three primary categories: General store orderers, business employee purchasers, and academic purchasers. Each account in the new system was tagged with this category code.

“We know what our customers buy, we know where they come from, and, to a certain extent, we know why they buy,” McCloskey says of the new infrastructure.

Next he ran his masterfile of buyers against the Claritas database in order to append even more demographic and psychographic/lifestyle data to each account. “They have mapped every single zip-plus-four in the US into a lifestyle segment,” he says. “All of that information is appended to our customer records.”

-> Using the data to improve email results & track DM

McCloskey and the hpshopping.com marketing team have already begun to use the new database to create a campaign segmentation model for more tailored, relevant email communications.

For example, instead of sending just three different email newsletters (one for each main category of customer), now hpshopping.com sends multiple types of newsletters, each tailored to various audiences in terms of frequency, offers and content.

Each individual's newsletter recommends appropriate solutions based on that customer’s specific purchase profile.

“The result is that we have seen a 30% lift in our email click-through as well as in our revenue associated with email communications,” McCloskey says.

In addition, the new database has helped HP track multi-channel sales; so when a customer gets a postcard or catalog in the postal mail and goes online to order, the postal campaign is assigned a portion of the credit for the sale.

End result: Now HP is expanding its postal mail marketing budget because the company feels safer knowing it is really paying off. "Closed-loop metrics quantifying the impact of DM has helped us increase DM investment," says McCloskey.

-> Next steps: Further segmentation and modeling

McCloskey's also continuing to invest in the new infrastructure program (which is more than many big companies can say about their CRM programs).

For example, he has contracted Prevision Marketing Group to provide segmentation, predictive conversion modeling, product propensity modeling, seasonal modeling, and campaign response modeling.

McCloskey plans to use this information, coupled with a “test and learn” campaign management strategy to enhance the relevance of interactions between hpshopping.com and its customers.

In the future, McCloskey hopes to apply his new infrastructure to more touch-points, such as call centers and HP's total customer Web experience.

-> 5 Lessons learned

In the meantime, here are five key lessons McCloskey and the hpshopping.com team have learned along the way:

Lesson 1. Get the data right and make sure you have as broad a view of customer as possible. You will need it to effectively communicate with those customers in a two-way dialogue.

Lesson 2. See what already exists (especially if you work for a big companies with multiple data silos) before you go out and reinvent the wheel.

Lesson 3. Make sure what you are doing is going to have a payback. You can not just build a system without thinking first how it is going to improve the customer experience and through that yield a return.

Lesson 4. Get executive commitment. Let them know how improving customer relationships will transform the business.

Lesson 5. Finally, ask yourself: Will it truly provide a competitive advantage? “I believe that for us it has and will continue to do so,” McCloskey says.