Hack83.Measure the Checkout Process

Hack 83. Measure the Checkout Process

Improvements in the online checkout processes have yielded more incremental revenue than almost any other aspect of web site measurement.

The checkout process is the most important process for any retail site to optimize. Luckily, it is also the easiest to analyze, and the improvement opportunities uncovered through analysis are extremely tangible. Following this simple framework will allow you to identify changes that have allowed retailers to make up their investment in these enhancements during the first day post implementation!

6.4.1. Step One: Establish a Baseline

The first step is establishing a baseline, as well as a method for monitoring the checkout process performance efficiently and on a periodic basis. There are two high-level metrics that should be trended and provide an excellent baseline:

• Same-session checkout completion over a given analysis period (for example, the fiscal month), defined as the number of visits (including receipt page) divided by the number of visits (including first step in checkout process)

• Cross-session checkout completion over a period representative of the average amount of time it takes a shopper to purchase, defined as the number of visitors reaching receipt page divided by the number of visitors starting the checkout process

In terms of monitoring the checkout process itself, you should go one level deeper and look at what is happening at each step. The metrics you need to pay attention to at each step in the checkout process include:

• The number of visits reaching each subsequent step

• The total potential sales at each step (based on the number of items remaining in the shopping cart compared to abandoned carts [Hack #82])

• The number of visits continuing to the next step in the process

• The abandonment rate from step to step (often expressed as a percentage of visits continuing or percent of visits still engaged in the process)

• The percentage of visits that leave the checkout process but remain on the site

• The percentage of visits that leave the checkout process and the site altogether

• The dollar amount of sales departing the checkout process

• The dollar amount of sales continuing in the checkout process

Keep in mind that measuring the checkout process is really just a specific variation on measuring multi-step processes [Hack #59]. Vendors will often provide each of these metrics in a single report, detailing the checkout process step by step (Figure 6-3).

Figure 6-3. Checkout funnel analysis report

For each step, you now know the likelihood of a customer moving forward in the process versus leaving the process or, even worse, leaving the site altogether. In addition, you immediately see the financial impact of losing these customers. Hopefully, most of you are saying, "That's great, but how can I stop this mass exodus from my checkout process?"

6.4.2. Step Two: Diagnose Problems

Next you need to diagnose the problems that are causing the attrition described above. The key to this diagnostic analysis is to uncover where people are going when they are not moving forward. In many cases, this will allow you to form some hypotheses around the root cause of these movements.

The ultimate diagnostic data source for this analysis is clickstream data, which lays out the exact path customers are taking from each step in the process. It is useful to characterize these movements into the following buckets:

• Direct site departure

• Forward to any subsequent step in the process; if there are no optional steps, this is the next step

• Backward to any previous steps in the process

We recommend that you generate a flow chart of the checkout process with the metrics above included for each step. You'll already know where the big attrition points are, but now you can get closer to what is driving the attrition and see whether it is backward or non-process movements that are causing the problem. The next question you should ask is "Which is worse?"

Most web measurement vendors provide some type of clickstream analysis designed to provide you with the data you need to fully diagnose abandonment in your checkout process in a visual way (Figure 6-4). These reports are variations on the table of information described above and presented in Figure 6-3, designed to highlight problems around a single page in the processin this case, exploring how many sales and orders are driven by people clicking from the "Checkout1: Sign In" page.

Figure 6-4. Clickstream report of page flow on a per-session (visit) basis

Reports and visualizations like this one can help you better diagnose problems that are highlighted by tables like the one in Figure 6-3. You'd be surprised at how valuable the ability to visually explore this information can be, yielding key insights like:

• Discrepancies in the number of visits moving forward and the total number of orders taken, similar to the 9.2 percent of sessions that go from step 1 to "View Cart" although only 3.5 percent of these yield orders

• Movement in unexpected directions, such as the 1,400 visits that leave the site and the 162 visits that return to the home page (resulting in only 0.1 percent of all orders)

• Errors in the process, like the 4,436 visits to the bad-password page and the 979 visits to the forgotten-password page.

6.4.3. Bringing It All Together

In summary, the methodology for measuring the checkout process is quite straightforward:

1. Establish a baseline and monitor high-level trends.

2. Monitor what action customers are taking at each step in the process in order to identify key areas of focus.

3. Dig a little deeper and determine where customers are going if they are not moving forward and create some hypotheses around why they exhibit this behavior; characterize the magnitude of these "leaks" in terms of volume.

4. It often makes sense to analyze individual customer segments [Hack #48] separately to provide further insight into customer behavior as well as to ensure that any changes that are made won't adversely impact one segment for the benefit of another. Typical segments include new versus repeat customers, most valuable customers, customers driven by marketing campaigns, customers of a particularly high price point category, etc.

5. Assess the revenue impact of customers taking a given path, using both conversion rate and dollars per buying session.

6. Prioritize opportunities for improvement based on potential return and cost of enhancement.

7. Test any changes that will help to plug the leaks.

8. Measure the impact of your changes and start all over again!

Assuming you become adept at using the continuous improvement process [Hack #2], you should be making small but important incremental improvements in your checkout process before long.

Brett Hurt and Eric T. Peterson