Conversion Rate Optimization of a Buy Pipeline (Part 1) – A Case Study

Conversion rate optimization is a very important part of an online business.  As noted in earlier posts, we at SpyFu enjoy sharing the challenges that we face as a business, in addition to creating great software applications.  Continuing this series, we will shed some light on the way that we have analyzed our buy pipeline, and what changes we have taken to produce positive results.

My name is Brian, and I perform business and operations analysis at SpyFu.  It is my goal to demonstrate the process of how to do quick business process analysis, in a way that allows readers to use the techniques later, even if the problem is slightly different.

Leaky Pipeline

Background:

A buy pipeline is the series of pages that a customer must complete in order to purchase from a company online.  SpyFu started with a relatively simple buy pipeline, but as our selection of products and options grew, we started adding pages into the buy pipeline which offered our customers the opportunity to purchase other services.

Generally, offering a client an upsell or another service is seen as a way to increase revenue, by adding additional dollars to a customer that you’ve already worked to attract to your site.  But it is important to remember that there is always a cost to everything.  In our case, the additional option pages were adding extra time and complexity into the pipeline, and some potential customers were not making it all the way through the purchasing process.

SpyFu actually has three pipelines on our site, and each is slightly different, but here is an example of the sequence of pages in our former main pipeline:

List of Products -> Upsell page -> Cross-Sell page -> Create Account page -> CC Info Page -> Confirmation

As you can see, our pipeline consisted of at least 6 pages.  There is also an optional page which can fit anywhere into these steps, if the client requested additional information.  The upsell page was originally added to allow customers to purchase Recon credits at a discount, and the cross-sell page offers the client the option to add UK services to their package.

Problem:

The longer a pipeline gets, the more hoops you are making the customer jump through in order to purchase from you.  At each point, a certain percentage of them give up and leave.  It may be a small percentage each time, but the longer the pipeline is, the more meaningful the effect on your conversions.  I’m sure you’ve had a few instances yourself where you are in the process of purchasing, and something about the purchase experience made you say “To hell with this crap!”, and leave the site.

Impact Measurement:

The old adage is “You can’t manage what you don’t measure”.  Well, the beauty of an Internet business is that we now have wonderful things like Google Analytics (GA) which allow us to measure the direct impact of everything… including each step of a buy process!  The challenge in this day and age is not just measuring the data, but making sure that your data is of good quality, and complete enough to do the analysis you need.

In our story, we first went through and re-mapped our “Goals” in GA so that we made sure that each pipeline was tracking all the pages in the same order that they were presented in real life.  Secondly, because we have multiple pipelines, we made sure that we eliminated any links which would feed a buyer from one pipeline into another (and cross-contaminate the data we were collecting).  Finally, we repeated the process for each pipeline, to ensure that we did not lose any sales data, and we could also determine if any changes had a profound effect on one of our sub-audiences.

Baseline:

With the “goals” section properly set up in Google Analytics, we then established a baseline of how many people were leaving the pipeline at each stage of the purchase process.  This allowed us to establish accurate numbers of how many visitors reached each step of the pipeline, as well as the percentage of people who were leaving at each stage.  In our case, we had an effective conversion rate (completed sale) of 3.71% when comparing the number of people who visted our product list, the official start of our buy pipeline funnel.  However, this metric is not the most useful ratio we can use.  If we jump ahead to the “cross sell” page (we are ignoring the “plan list” page, as it is an optional page), then we have a better starting point.  People who have reached the cross sell page are not just casual lookers, they are visitors who have actively moved beyond simply looking at products and prices, and have indicated a desire to buy.  Our conversion rate for these visitors was 35.4%, or roughly 1 in 3.  Definitely room for improvement.

Here is a screenshot of the original buy pipeline (Follow along to Part 2 to see the changes and results):

Old Pipeline:

Customer pipeline flow chart

Go to Part 2 for results of the test.