Optimizing website navigation on your e-Commerce site is a complex project. Choosing the right implementation tools and methods is essential to your success.

Key Takeaways:

  • Maintain balance between scripted and server-side variables
  • You need a multivariate and full funnel optimization tool
  • Understand your results before rushing to implement

The goal of our Optimization of e-Commerce Site Navigation series (Part III, & III) was to help you analyze the state of your current on-site navigation and to help you design an effective optimization campaign.

““People won’t use your Web site if they can’t find their way around it.” ”

— Steve Krug, usability specialist and author of “Don’t Make Me Think, A Common Sense Approach to Web Usability”


In this final installment of this blog post series, we’ll focus on various aspects of campaign implementation.

Variable Creation

Different optimization tools offer a variety of implementation options. Most of the time the work requires the creation of custom scripts which, when executed in the browser (or proxy server), will change the navigation options, layout, or functionality.

When it comes to optimizing your website’s navigation, we think that it is important to consider both server-side and client-side implementation options:

  • Client-side Scripted Variables: The main advantage of client-side changes to your site navigation is independence from your e-Commerce IT team. In theory, you should be able to create all desired changes purely on the client side; however, considering the number of navigation links and options, this could quickly become a very complicated coding exercise. Additionally, manipulation of hundreds of navigation links through a client-side script can potentially get slow and impact the page rendering speed.
  • Server-side Variables: Market leading e-Commerce platforms offer template-based languages that allow you to create different navigation options on the server side. Instead of spending days to custom script navigation options, your e-Commerce programmer can create these options within hours on the server side. To make it available to the optimization system, these variations will be injected in the server-side code and delivered as part of the regular web page. During the optimization process, the tool will render one variation of the site navigation invisible and the other visible.

In determining the spread between client- and server-side variables, you will most likely be influenced by the availability and cost of server-side programming resources. We recommend that heavy-duty changes to the navigation are made on the server side. Changes to layout and style, however, are better suited for client side implementations.

Technology Requirements

The optimization of e-Commerce site navigation is a very demanding undertaking. In order to succeed, you have to design variables that have upside potential and use technology capable of performing heavy lifting. The following is a short list of requirements that the optimization tool has to satisfy:

  • Multivariate: There are almost an infinite number of ways in which visitors can navigate through your e-Commerce site. This is because a huge number of combinations get created when you start to configure different main-, sub-, and secondary navigation options. For that reason, running A/B tests to find a working combination will get you nowhere. To draw a parallel, the A/B approach is equivalent to searching for a needle in the haystack – there are simply too many options. To simplify this process, many companies are running a series of tests focused on one aspect each time. For example, one wave of tests will be focused on main options, followed by sub-navigation options, then on layout options, and so on. Individual options might produce a lift in isolation, but combining many of them into a single version of the page may actually produce negative results. The issue is created by an interaction between individual changes.
 This is why you need to use a multivariate optimization solution capable of changing multiple navigation options simultaneously.
  • Full Funnel: Site navigation has a site-wide presence and your focus has to be on end-to-end results. As you are designing your optimization experiment, you need to be flexible enough to maintain or remove navigation options from certain pages (i.e., checkout pages) in order to create changes that co-vary with elements on other pages (keeping style consistent throughout the site is important). These kinds of variables are only possible if the tool enables multi-page campaigns and multi-component variables.
  • Real Time: One often-misunderstood dynamic of e-Commerce sites is that visitor behavior changes over time. This implies that an optimization tool needs to be able to make decisions in real time and adapt to changes in web visitor behavior. There isn’t one static solution for dynamic visitor behavior, and this applies not only to general site content, layout, and functionality, but also to site navigation. (To illustrate the time-varying nature of shopper behavior, check out this recent post: What the S&P 500 and e-Commerce have in common)

How to Use Results

Legacy testing methodology is thinking in terms of which statistically significant combination changes in navigation have produced a maximum lift. For example, the report shows that combination #184 is the best and the data sample has 99.75% statistical confidence.

e-Optimizer® Campaign top combinations: Revenue per Visit

Based on the results above, the client would just proceed and implement the winner. The problem is that visitor behavior is changing and there is not one static solution that will consistently produce a positive lift.

That is why we advise our clients to go a layer deeper and focus on the individual navigation options included in the optimization campaign. For example, the report below shows that the Cat Nav – D variable was one of the positive navigation options.

e-Optimizer® campaign variable importance bar chart: revenue per visit

Next, we would advise our client to examine if that option had a positive impact during the duration of this campaign. The example below shows this variable was positively impacting revenue per visitor each time:

e-Optimizer® campaign variable importance ling graph: revenue per visit: cat nav D

Based on the results above, we would advise our client to permanently implement that variable. The variables that have a positive cumulative impact, but vary over time, should remain in-the-mix for ongoing optimization, and variables that were cumulatively negative should be replaced with alternatives.


We hope that this series of blog posts have helped you better understand the importance of e-Commerce site navigation optimization. At some point, this optimization initiative will have to be performed on an e-Commerce site. From our point of view, we see this as a continuous improvement process where you will run waves of optimization campaigns and periodically find evergreen changes that deserve implementation, while also learning from the results to create optimization variables that will continually move the revenue needle upward.