Peas In A Pod: Behind Personalization

Do you ever wonder what happens when you go to an e-commerce web site and input your information to receive personalized recommendations? What wheels start to turn behind the scenes? The short answer is Many! In order to complete your personalized transactions, the system taps into one or more databases.

Let’s take a look at how Peapod, the online grocer, uses databases to facilitate its unique personalized grocery shopping and delivery service.

About Peapod

Few things are more personal than food. So when it comes to personalizing the web experience, what could be more personal than shopping for groceries from the comfort of your desk chair?

Peapod Peapod delivers groceries to more than 40,000 households in six markets. Peapod isn’t your garden-variety web site that passes orders on to local grocery stores. Its professional shoppers take your personal grocery list and scour the aisles of major grocery stores to do what harried consumers don’t have the time to do — or interest in doing: Selecting exactly the products wanted by the consumer, and delivering them to the shopper’s home.

Your Personal Profile

In order to make Peapod grocery shopping more convenient, shoppers can store their favorite products, including favorite brands, in its database. This allows Peapod to remember a consumer’s preferences — just like your favorite butcher or product manager at your current supermarket does.

What? Your current grocery store doesn’t remember who you are each time you come in, and doesn’t have that special cut of meat waiting for you? You can already see one way Peapod is winning customers — with personalization.

Another rung in the personalization ladder is this: Shoppers returning to the Peapod site to shop again may notice that the screen has changed ever so slightly, with a popup menu describing the functions of the new buttons that have appeared. One example of this changing user interface is the ability to use the previous shopping session as a basis for the current shopping session.

What many Peapod shoppers may not realize is that the special offers, electronic coupons, and advertisements they see on Peapod are not shown to everyone. In fact, there is a chance that other people with exactly the same profile won’t see the same promotions. This is because Peapod uses a custom-designed software system called the Universal Event Processor (UEP) to determine which promotions will be seen by which shoppers.

Events Lead To E-Commerce

Peapod takes the view that practically everything that occurs on the web site and in the fulfillment of orders can be considered events that can be managed by the UEP, says Thomas Parkinson, executive vice president-chief technology officer at Peapod.

Content Events – The UEP is constantly looking for ways to provide content to the consumer, such as presenting electronic coupons and what grocers call “deals” — such as discounts, two-for-one offers, tie-ins, and other ways to provide consumers with extra products and extra value. By monitoring the products purchased, the UEP can trigger the display of content (or deals) that might appeal to the shopper.

Customization Events – The UEP enhances the shopping experience by providing opportunities to customize the shopping through a personal profile that the UEP can draw upon to trigger promotional activities. Just as special deals can be triggered by the actual purchase of products, the personal profile can trigger the display of deals as well. However, there is more to customizing content than just displaying opportunities for additional purchases. The act of redemption is just as important to monitor and track in order to provide the feedback to the system that one-to-one web marketers need to judge the effectiveness of their promotions.

Peapod’s UEP is used in a variety of ways to personalize the content for its shoppers. For recent members, Peapod encourages the use of personal list functionality by automatically creating personal lists for them from first and second orders. The Peapod application records user preferences such as preferred sorting criteria (unit price, alphabetical, fat content, etc.) or payment method, and presents the user session with these preferences as default settings. By tracking the purchase behaviors of shoppers, it can offer manufacturers the opportunity to test a wide variety of promotional activities, including:

  • Promotions aimed at consumers loyal to another brand, without targeting their own brand for deals.
  • Comparing the effectiveness of ads versus samples.
  • Promoting a product when a complementary product is being purchased.

Commerce Events – The actual purchase of products through the Peapod web site involves the UEP even more than the selection of products, because of the complexity of multiple deals accepted by the shoppers combined with their personal profile specifying their individual preferences. Peapod’s style of shopping is similar to the way people shop at traditional grocery stores. If you’re used to pushing a cart up and down each aisle, letting the products on the shelf remind you of what you need, then Peapod will seem very familiar. On the other hand, if you go to the grocery store with a detailed list in hand, then you, too, will find Peapod familiar.

Integrating Databases With Fulfillment

Parkinson sees closing the loop on orders, requests, and other actionable activities — or, as grocery people call it, “redemption” — as one of the most difficult aspects of operating Peapod. Think of all the systems that need to be looked at, and the “back-end” activities that are needed in order to provide end-to-end integration of the systems needed to full each customer’s order:

  • Products are displayed based on the preferences of the customer.
  • An order is placed using whatever special deals the consumer qualifies for.
  • Batches of orders are transmitted to each fulfillment center (i.e., grocery store).
  • Reports are printed at fulfillment centers for each department (produce, meat, packaged goods).
  • Professional shoppers acquire or obtain the products for each shopper, noting any difficulties in locating products for a specific customer (e.g., “bananas too ripe, skipped on this order”).
  • Checkout and packing each customer’s order.
  • Delivery and payment collection.
  • Entry of order completion.

Database Marketing

Peapod’s UEP uses database marketing techniques on a real-time basis, customizing content based on a number of factors. Parkinson and his team of developers use a traditional relational database system from Ingress as the core technology, with a number of “add-ons.” The UEP uses a real-time tiered logic system that handles five different types of events that customize content:

  • Promotions that are presented to every shopper.
  • Specials targeted at easily identified groups of shoppers, such as those within a certain market.
  • Groups of shoppers based on characteristics, such as geography.
  • Profile characteristics.
  • Complex sets of attributes such as those that depend upon a combination of profile and products purchased.

By allowing for multiple types of events to be handled by the UEP, Peapod can add an event to practically any part of the system, from web site display through the delivery of products to the consumer.

Data Warehousing

Managing a business requires many management reports that summarize the activities of the business from many different angles. Typically today, businesses use data warehouses that create and store summary types of data that are needed to analyze the condition of the business.

At Peapod, computer systems are automatically run weekly to create snapshot databases with detail and summary data that is needed by management to analyze the trends of the business and plan for the upcoming week. Since the UEP and its underlying database management controls all aspects of the Peapod business, the data warehouse has a complete picture of all operations for the past week.

Next Week: The future is now. Let’s take a look back to see where we’ve come, where we are, and where we are going with one-to-one marketing on the Internet.

Related reading