How to Set Up a Basic Conversion Testing Strategy
For many of us, conversion rate optimization (CRO) can seem like a complicated venture. Many people aren’t even sure where to begin. Though each business might need to tailor the testing process to their company and website, the following is some basic insight into how to create a CRO strategy for your company.
If you’d like to start with an overview of CRO, read A CMO’s Guide to Conversion Rate Optimization.
Phase I: Strategic Research
The first step in any project is always research. For conversion testing, this includes things like:
<>Analytics Data Mining
- Determine baseline traffic volumes, top landing pages, bounce rates, exit rates and traffic sources.
- Understand current conversion rates for the site.
Competitive Data Mining
- Examine other sites in the same competitive niche, and see what elements are missing from your site.
- Are you competitive with other sites in terms of offerings and pricing?
- Are your competitors taking the right action for their audience?
- Do competitors follow best practices?
Establishing a Sales Funnel
- Create a map of the overall sales funnel based on the outcomes of the above two steps. This will be used as a reference document throughout the testing process.
- Some sites may have multiple goals for their visitors. In those cases, it would be appropriate to have multiple different Funnels.
Determining Target Page and Testing Method
- Based on the outcome of all steps mentioned above, determine the best page to target for testing.
- Based on analytics findings, also determine whether testing should be A/B or Multivariate.
In deciding your priority for testing, consider using the following order:
- Pages that are already performing to some degree
- Pages with the highest volumes of traffic
- Pages near the beginning of the sales funnel
Phase II: Creative Development
For A/B Split Testing
- Develop a series of redesign “recipes” for the entire targeted page (different ingredients that go into a variation), taking care to include differences between each recipe (typically about three different designs).
- Traffic is split randomly, but equally between the difference recipes, and visitor actions are recorded.
- Low volume of traffic is required on the targeted page due to low recipe count.
For Multivariate Testing
- Develop a series of redesign recipes for specific elements on the targeted page, taking into consideration how elements may interact with each other.
- Traffic (users) shown all possible combinations of targeted elements randomly but equally, and visitor actions are recorded.
- High volume of traffic required on the targeted page due to high recipe count.
Phase III: Testing
We recommend using the Google Marketing Platform (formerly known as Google Website Optimizer) to facilitate A/B split and low-level multivariate testing. For multivariate tests on larger scale sites that will require an excess of 50,000 recipes, some of these options may be suitable solutions:
- License more robust software from a third party (e.g., SiteTuners)
- Develop an in-house suite with more robust capabilities than Google software provides out of the box.
In my experience, Google is usually able to provide actionable results after approximately one month of testing. Results may come in more quickly for sites with larger volumes of traffic, or may take longer for sites with smaller volumes of traffic.
Phase IV: Reporting
Once you are satisfied that substantial results have been obtained, it’s time to generate a findings report.
The report should detail what was changed, how it was changed, why it was changed, and then deduce why the performance increase was achieved. Also, you should report on the estimated lifetime value of increased conversion rates based on the winning version.
Creative Development Tactics: Navigation
Outlined below are a few steps that can be taken to quickly test navigation elements, should the need exist on a particular page:
- Site overlay feature in Google Analytics can provide insight as to where people are clicking.
- Use software to predict visual attention patterns without the need for an actual sample demographic. For example, a heat mapping type of software like HotJar or eye-tracking program such as Attention Insight can help you analyze behavior on a page.
These can help to identify areas that are cumbersome within navigation:
- Outline all current possible navigation steps (aka, task analysis).
- Simplify the task flow, on paper, wherever possible.
- Top landing pages (content overview) report can give insight as to what navigation features are being used most frequently.
- Navigation analysis report provides more details as to how effectively the navigation funnel is performing.
- Requires multiple participants (three – five people).
- Among the quicker methods for spotting usability issues in any navigation system.
- Participants are controlled, and do not interact with each other.
- Controller has the opportunity to explain navigation, which ensures context-relevant reports from participants.
- Controller can note unrelated issues as well, as they are brought to light during the evaluation.
For more information, check out this classic research article on how to conduct a heuristic evaluation.
Happy testing, everyone!
This article was originally written by SEO analyst and CRO specialist Scott Fowles. It is now maintained by BCI staff.