Fulfillment : Your Data Holds the Key

Improve average order value with successful online upsells and cross-sells

April 2009 By Will Hakes, Ph.D.

singlepage

It’s a pretty simple formula: Drive customers from high-cost channels like call centers into a low-cost channel like your Web site and save money. You likely will lower the cost to serve, give customers what they want and create a winning scenario for everyone. One problem—that’s generally how it works in a textbook but not in the real world.

In the real world, online and offline channels are not perfect substitutes. They are very different challenges, indeed. Perhaps it’s time to take a fresh look at how you are cross-selling/upselling in an online environment. It doesn’t really matter if you are a telco, an office supply company or sell pet supplies, your competition is trying to do it better than you are. So don’t just do it better—do it differently.

To put some structure around developing an approach, think about your online environment in the following steps.

1. What’s the Strategy?
This is the most challenging, and the most nuanced, to each business. Without diving into strategy formulation here, suffice it to say this is the first, biggest step. The online strategy must be integrated with the offline approach.

Your customers aren’t really going to care if they came to you online or offline. Therefore, your strategy for sales and support must use both channels to enhance the customer experience as well as provide the organization with essential data about customer interaction as a whole and, more specifically, individual customer data and behavior.

For example, a customer coming in through a call center may have called in with a particular promo code or to a unique toll-free number tied to a campaign. Likewise, an online customer may have responded to the same campaign. Both customers’ interactions should be tied to the campaign at the campaign-performance level as well as the individual-lead level, so you can learn how to improve your targeting at the macro and micro levels.

2. What Information Should Be Collected?
The most important piece—how will you identify a person who comes to your Web site? User login? IP address? Cookie? The more granular your level of recognition, the better you’ll perform when making the “best” offers.

If you have a recurring relationship with your customers, then you should be leveraging every bit of that rich history. Moreover, each online interaction, when possible, should be logged and stored to further enhance that historical view. In other words, your customer datamart should speak online/offline languages equally well, with one channel feeding the other.

Some of the most obvious information, such as keywords, lead source, etc., is often discarded. The right Web analytics tools and proper analysis should help you further understand what’s driving traffic and provide key input into analytics/targeting.

3. What Analytics/Targeting Should Drive Offers?
The simplest form of analytics to deploy is some type of collaborative filtering technique. You’ve likely seen it on sites like Amazon.com (“people who bought X also bought Y …”). It may seem like a crude technique, but it is comfortable and noninvasive. That’s why it works.

But the offers themselves should be based on good analytical models. Think of it this way: When you capture someone’s attention, just for a moment, you have a split second to show her something relevant. Don’t show her every offer you can imagine, but consider presenting more than one. The right offers will drive both average order value and conversion rates, two of your most important key performance indicators (KPIs).

These models can take one of three different forms:

• Behavioral: Watch the movement, navigation, path, etc., of an individual, and then serve what works best for those following a similar path. Across industries, I’ve seen 20 percent to 50 percent lift via good behavioral models.
• Predictive: Take what you know about that customer (both historical as well as implied data such as geography, keywords, etc.) and make the offer(s) that are most compelling. In some cases, I’ve seen literally an overnight jump of 100 percent in online orders when smart models are deployed. Moreover, conversion rates tend to increase as well because you are selling the “right” things.
• Hybrid: If you have the database systems in place, why do only one of the above, when you can do both? Your prior knowledge of a customer drives a series of best offers, and the path taken to and then on your site drives refinement.

4. How Can Analytical Insights Be Leveraged at Various Customer Contact Points?
The key to success here is a tiered approach. First, once you have
accomplished the above, and online offers are targeted, you’ll have compelling results. This promotes confidence. Use initial findings to get additional buy-in.

Each customer contact point—be it a return visit where the visitor is known (cookie, login or otherwise) or a post-sale e-mail confirmation—is an opportunity to collect more data and possibly make another sale.

When a customer returns to your Web site and you can identify him, leverage the opportunity to test offers via a targeted, multivariate approach. Multivariate testing involves presenting a variety of changes—from look and feel to offers—and seeing what combination works best to drive a particular outcome (such as a sale). By inserting targeting into the mix, you’re able to filter the offers down to a set that the customer is most likely to respond to. This technique delivers not just higher sales, but higher revenue.

Often overlooked is the post-sale experience. If you’re shipping merchandise to a customer, you’ll likely send her an e-mail confirming the order and shipment. This is an opportunity to apply what you know about the customer and send a specific offer to her. Maybe it’s for free shipping on a product that is complementary to what she just ordered. If the order was a birthday gift for someone else (collect this data during checkout), perhaps you offer her a feature to remember that person’s birthday in the future and make recommendations as the event gets closer. Even though this isn’t a cash incentive, you may find that convenience is equally motivating.

5. Ensure Continuous Improvement
To ensure continuous improvement means knowing what you were trying to improve in the first place. The first step here is to define your goals as an organization, as well as for each campaign. If you haven’t been collecting data consistently (meaning the same data, the same way, every day and every campaign), then you cannot accurately compare campaigns or time periods to gauge improvement.

To establish baselines for improvement, you must first establish your KPIs. These are a set of tools that help you organize data in context of your business goals. Developing KPIs isn’t always an easy task because of multiple stakeholders and opinions on establishing measurable KPIs. Expert advice in this area often is recommended. However, once KPIs are established, you consistently can measure campaigns, time periods, etc., against each other, determining improvement areas and which levers to pull to affect change in your marketing.

This data also should provide a closed feedback loop when tied to individuals and their lead sources. Feeding the data back into modeling, multivariate tests and touchpoint strategies will drive incremental lift as you experiment with different offers, creative design and cross-channel communications.

One final point to keep in mind: Getting this right is a complex endeavor. Consider bringing in an outside expert, one who has seen online offer targeting implemented across many industries. A very basic audit from an expert can help you avoid costly missteps.

William Hakes, Ph.D., is president, analytics at Aspen Marketing Services, an integrated marketing services firm with headquarters in Chicago. He can be reached by phone at (678) 597-0218 or via e-mail at whakes@aspensms.com.

• MORE ON
• ONLINE MARKETING
• PRINT PRODUCTION & FULFILLMENT