E-commerce Abandoned Carts

Recapturing lost revenue

Summary

I coordinated data across databases and worked across multiple departments to solve a longstanding issue for our business. In the process, over $200K of revenue has been recaptured through online ordering. Rates of order completion among targeted customers rose from 31% to 40%, and customers were more likely to order online than through fax or phone.

Organizations Entered per Client

$240K

Regular Users per Organization

9%

Defining the Problem

Abandoned carts had been a problem within our business for several years. While we could see which carts were created within our ecommerce system and never checked out, we were unable to see if that customer had then ordered by another method, such as fax or phone. The call center understandably did not want emails to be sent to customers who ordered another way, because it would likely confuse them and cause them to call in. As a business, we were leaving potentially hundresds of thousands of dollars on the table by not following up on these carts, and the revenue that did come in from another method required manual processing.

In the past, various departments had been somewhat siloed, which had prevented this problem from being solved. Additionally, the online ordering system maintained its own seperate database for various business reasons, so information did not exist in a single spot. Finally, since it crossed several departments, no one person or department had taken ownership of this problem, and so no one solved it.

Gathering Requirements

I conducted a series of meetings with the various stakeholders to nail down the problem. I met with people from our call center and exhibitor service center, our marketing team, and our product management and development teams to define what problems I would need to solve at each step in the process.

We needed to be able to cross reference customers online with other methods to see if customers had placed an order or not so that we could know who to contact. Then, we needed a clear plan of what content we would send them to help them through the process, how we would support them if they had questions, and how we would measure the success or failure of the project.

Technical Research

The first problem to tackle was targeting the correct people. I knew that somehow the online ordering system DB connected to the system of record in order to push orders over, and that both databases would have a unique identifier for that record. I worked with my technical team to write a SQL query that grab the identifier and whether an order was associated with it from the system of record, and another query that would report on abandoned carts and include that unique identifier.

From there, I worked in excel to compare the two. I grabbed ID's with no orders in the system of record, and then used a formula to determine which of those had an abandoned cart. This way, we only matched abandoned carts with no orders placed, and knew who to contact. I circled back and demo'd this to the service center to get their blessing.

Excel formula to manually identify duplicates

Building the Email

Once we were able to define who we would send the abandoned carts emails to, the next step was defining what we would send them, and how. For this phase, I worked with our marketing department to get an email sent out of Marketo.

Since the customer had previously abandoned their cart, I wanted to eliminate as much confusion as possible. I designed an email that would walk them through the process of checking out to eliminate as much confusion as possible.

Additionally, I wanted to make sure that the primary actions appeared above the fold. Several of our marketing emails included large header graphics, which, while attractive, could serve to push the action items below the fold. I wanted to make sure that as soon as the user opened this email, they could see where to go to complete their task, and that help was available.

Measuring Impact

The final problem was how to measure the success of the project. Around this, there were two main questions to answer. The first was how to gather and track metrics relative to the immediate performance of the project, such as revenue. The second was the impacct on the overall business - Did the abandoned carts email really increase the rate of ordering among targeted exhibitors, or did it merely pull orders online from fax and phone?

To answer the first question, we created a UTM tag that we applied to the link in the email. Any transaction made after the user clicked on that link would be reported to Google Analytics, and could be segmented out in order to analyze it on its own.

The second question was more difficult. We don't know what people would have done in an alternative universe where they did not get an email, but we do know what they had done in the past. In order to analyze this, we would have to examine the ordering rate of exhibitors with abandoned carts before we started sending emails to the rate at which they ordered after. I was able to do this after we had been sending the emails for about two months, since we needed to gather sufficient data.

Results

The abandoned carts emails performed well beyond my initial expectations. In just over 6 months since deploying them, we have generated over $240K in revenue, a significant chunk of which we were not getting before. We increased the rate of ordering among exhibitors targeted by 9%, gathering new orders rather than just moving them online.

Retrospective

This project challenged me to grow above and beyond my role as a designer, and to learn new skills. The cross department ties forged would lay the groundwork for future projects, both in terms of how to collaborate and how to track effectiveness. In addition, I learned quite a bit about writing SQL queries, which allowed me to get even more insights on historical data.