How to Increase Online Store Profits with Database Marketing - How PC Mall Does It

CHALLENGE
Like many online retailers with roots in the traditional print catalog world, PC Mall, Inc. (NASDAQ:MALL) has always been to some extent a database-driven company.

From day one the company was less focused on heavy traffic and more on profitable traffic. They were also media agnostic.

Advertising and Database Marketing Campaign Manager Alan Paggao says, "We made a specific goal to start advertising based on direct marketing, not on media. For example, back in 2000, when most companies were just driving traffic, we drove it based on point of sale transactions - by determining what the ROI was, based on margins."

How do you figure out the ROI of every campaign if leads may be driven to your store from a variety of sources such as print catalogs, email newsletters, or pay-per-click ads? What if a buyer saw one campaign in one media, but weren't convinced to act until they saw it for a second time in another media?

Database marketing, never a science for dummies, just got a whole lot more complicated.

CAMPAIGN
Over the past year PC Mall invested in a substantial database upgrade. "We'd been wanting to do it for years," Paggao says, "but it took getting the right people in place to implement the technology. It was hard. We needed the data to be manageable and fully integrated." Which can be a nightmare when you are coping with previously silo-ed systems.

The key was to be able to track these elements for every transaction:
- original source
- ancillary source if multiple channels were involved
- conversion to sale
- ROI of sale
- customer behavior pattern
- email responsiveness
- lifetime value

By April 2000, the first new parts of the database were ready. "We knocked it off little by little, test, by test." Eagerly the marketing team immediately began analyzing the data with three key objectives:

Objective #1. Behavioral analysis to determine media ROI

Their first goal was to determine how each campaign truly affected sales, no matter what channel the message was delivered in or what channel the shopper chose to place the order through.

"For example, when we mail a catalog I know that's affecting Web traffic hands down. Instead of a source code selection, what we do is run matchback routines. We try to figure out the original source and how to target them."

What if a buyer was touched by two different campaigns? "It's a matter of understanding their behavior and allocating a proper percent of the contribution each media made toward the sale." The team analysed this contribution in several ways:

a. Timing of the campaign. "Each ad has a lifetime that's roughly a bell shaped curve. An online campaign lasts five to eight days, a catalog lasts three to four weeks." At the peak of the curve, Paggao will assign a higher percent of the sale to that media, and vice versa.

b. Number of times buyer has purchased in past. "If our catalog dropped to that buyer, but they purchased through [pay-per-click shopping engine] DealTime for the first time, I'd probably assign 80-90% of that ROI to DealTime. If they repeat, I'd start scaling back the percentages in terms of ad contact."

c. Shopping-style preferences of shopper. If a shopper has a proven history of responding to certain types of offers or shopping in a particular way (for example, mainly price comparison shopping) that would also affect where Paggao assigns ROI.

After backing in sales margin expense, Paggao puts thresholds on "what I'm willing to live with to acquire a customer" and then helps portion out future media investments accordingly. He notes, "I'm willing to pay more to acquire a customer versus just an order. You have to separate out lifetime value."

Objective #2. Pay-per-click SKU analysis to raise profits

Pay per click advertising can get out of control easily when you have thousands of SKUs to place on shopping sites such as DealTime. Paggao says, "I needed to understand the behaviors of SKUs by click source. For example I isolate what SKUs people were clicking into from DealTime, and slam them against the database internally to identify what SKUs are actually being converted and what ones are not."

"Then I do SKU suppressions, and on my next data upload I won't send those SKUs up."

Objective #3. Email response analysis

Next Paggao and the team started running segmentation analysis against the email names on the Company's database. Aside from the obvious data (open rates, conversions to sales, lifetime value) they have also begun tracking metrics by ISP, product category, and manufacturer (for example how well campaigns featuring Sony products do compared to HP product campaigns).

RESULTS
PC Mall achieved record third quarter sales this year up 34% from third quarter in 2001, and up 12% from second quarter 2002. Profits were also up.

As Paggao puts it, "We are badass here!"

However, he is careful to note this is not so much due to technology as it is to having the right team in place to analyze and use the data it produces. "A lot of companies have built it, but they don't know what to do with it."

"It took a while for it to get looked at correctly. You can look at data left, right, upside down. You need someone experienced in looking at data. In September we hired a VP Database Marketing versed in this and he's taking us to the next level."

Notes:

- After analyzing print catalogs' contribution, PC Mall has made "absolutely no cutbacks" in the amount printed and mailed. "We've been able to mail more efficiently based on data, but we aren't mailing less," remarks Paggao.

In fact he notes that catalogs are back in as media-de-jour now that shoppers are so saturated with email and the USPS is holding on rate hikes for a while. "It's so much easier to click delete on an email than to toss a catalog at home."

- Paggao is thrilled with the results of pay-per-click SKU transaction reports. He says, "By understanding how people click in those SKUs, how they convert, and the ROI-level by SKU, that has saved us thousands of dollars."

"For example, you could spend $6000 with DealTime and make $100,000 on it without knowing what SKUs worked best and which didn't. If you have data sets integrated to identify the ROI in reality, you could have gotten that $100,000 for $3,000. Other merchants may be happy with what they are getting, but I wouldn't be."

- For PC Mall's lists, the response gap between HTML and text-only campaigns still exists with HTML as the winner. However that gap is narrower than in years past. "HTML is a little better," notes Paggao, not a whole lot.

Also, Paggao has noticed some differences between ISP responses (for example customers with Hotmail accounts vs. customers withAOL accounts) as well as responses to various manufacturer offers, but wants to build up more data before reporting on them.

http://www.pcmall.com
http://www.ecost.com (a PC Mall-owned site)
http://www.macmall.com (ditto)
http://www.dealtime.com