Creating Conversion Funnels via Segmentation

Regardless of what type of website you manage, it is bound to have some sort of conversion funnel. If you are an online retailer, your funnel may consist of people looking at products, selecting products, and then buying products. If you are a B2B company, your funnel may be higher-level like acquisition, research, trial and then form completion. Many of my clients want to model their conversion funnels in Adobe Analytics (SiteCatalyst) so they can se where visitors fall, in what percentages and how these buckets change over time. Unfortunately, this isn’t one of Adobe Analytics’ strong suits. In this post, I will share why the out-of-box conversion funnels are not ideal and how you can use segmentation to help build your conversion funnels.

Conversion Funnel Report

As I described in my old blog post on Conversion Funnels, the Conversion Funnel report is merely a graphical representation of whatever Success Events you happen to add to the report. This works if you have discrete Success Events related to each of your conversion funnel steps, but it does not show you what percent of your population is currently at each step of the funnel. For example, if I visit an online retail website, view a product, then add a product to cart (Cart Add Success Event is set) and then order a product (Purchase Success Event is set), the conversion funnel would have a value of “1” for me in each of the rows of this conversion funnel report:

While this may be useful in the context of seeing what percent of visitors make it through each step of the funnel, what if my question is “What percent of my population reached a specific step in the overall conversion funnel this week or month versus last week or month?” In this situation, the out-of-the box conversion funnel report can show you a time-based comparison, but as I will show later, this doesn’t give you the full picture:

In the next section, I will show you how segmentation can be used to improve upon this…

Using Segments to Create Funnel Populations

To address the aforementioned questions in Adobe Analytics, it is best to use the segmentation features of the product. Using segmentation, you can place each website visit (or visitor) into one of your high-level conversion funnel buckets and then create a different type of funnel in Excel using the ReportBuilder tool. First, you have to identify what criteria you are going to use to determine if a visit is in bucket #1, #2, etc. In this case, let’s imagine that you work for a B2B company and that your first bucket is “Awareness” and it is defined as people who have come to your website, but never seen a product, attempted to download a trial of it or purchased it. The second conversion funnel bucket is “Researchers” and this includes visits where people have looked at one or more products (or clicked on demos/videos and other product-related actions), but have not added a product to the cart or purchased (or filled out a lead form if online purchase is not possible). The third conversion funnel bucket is “Interested” and this includes visits in which people have either added to cart of filled out a lead form, but have not purchased (if available online). Our last conversion funnel buckets is our “Buyers” who have successfully purchased a product or committed to the product in some way (if purchase is not available online).

With these four conversion funnel buckets in mind, your next step is to subdivide all of your visits (or visitors) into one of these four buckets. While this may seem easy, it is actually a bit tricky, because you have to make sure that the same visit is not present in more than one bucket. Doing this requires some fancy Adobe Analytics segmentation skills. To create the first conversion funnel bucket, you would want to create a Visit segment that excluded any visitors who had viewed products, added products to the cart or purchased:

Next, we want to create our Researchers segment for visits that viewed products (you can also add other research events here with an “OR” clause), but excluding visits where a cart addition or order took place:

Next, we want to create our Interested segment for visits that added products to cart (you can also add things like lead form completions here), but excluding visits where an order took place:

Finally, we have our Buyers segment to see visits where visitors completed an order:

If you add up the various Visit counts in the above segments, you can see that they are mutually exclusive and add up to the total 40,089,255 showing in the segment preview area. This is a quick way to verify that you have built your segments correctly.

Applying Conversion Funnel Segments

Now that you have your conversion funnel segments defined, there are many ways you can use them. First, you can apply each segment to see any report for visits at that stage of the conversion funnel. For example, you could look at what internal search phrases are used by Researchers vs. Awareness folks. You could view the different pathing behaviors by conversion funnel segment or see what campaign codes drove each type. But the most interesting thing you can do (in my opinion) is to create a conversion funnel report in Microsoft Excel using ReportBuilder. For example, if you were to build a Visits data block with the “Awareness” segment applied, you would be looking at Awareness visits for the specified date range. Then you could do the same thing for the other three segments and then trend the percentages over time. Once you have separate data blocks, you can use formulas to combine them into a percentage-based conversion funnel and see the progression over time like this:

In the preceding example, there is not much of a spread when it comes to the last two funnel steps, but if we use some different [fake] data, let’s see how cool the reporting of this might look:

What I like about this type of analysis, is that it provides an opportunity to see where YOUR website problems lie. Every website is different. Some websites are great at getting top of funnel visitors to get to stage three or four of the funnel, but then they struggle to get them across the finish line. Others are the opposite in that they don’t get many people to stage two or three, but when they do, they convert very well. Knowing where your website’s problems lie, allows you to identify practical ways to improve your funnel. This can be done by focusing your testing and design efforts in the right places, instead of wasting time in areas where your website is doing well. As you can see, this is a different type of approach to conversion funnel analysis, but one that I think can help your organization better understand how visitors are flowing through your conversion path at a high level and provide benchmarks of this over time. If you already have most of your key conversion funnel KPI’s set, then this solution requires no tagging, just the creation of some new segments, so there is no reason to not give it a try!