As a long time conversion optimization consultant, I have asked this question dozens of times, for each client I have worked with: "Should I recommend starting the conversion optimization journey with a test or with a deep data analysis?" The answer, inevitably, would contain the expression "it depends". But with time, I understood an important differentiation that helps finding the best formula for each type of company.
I believe the answer to the question differs radically between two segments: companies that are new to analytics and testing & companies that are veteran conversion optimizers. Below I will go over these two segments and what I would recommend as the best way of action for each. This may be very useful for both consultants and company employees looking to start (or improve) a Marketing Optimization program.
Companies that are new to the conversion optimization field tend to start their journey with the implementation of a Web Analytics tool. Even though this sounds logical and is often the recommendation given by consultants, I find this path very ineffective. It feels like marriage a few decades ago: first you marry and then you get to know each other.
I firmly believe that before you start to implement tools and commit to implementations, it is important to understand what you can get out of a conversion optimization process (in the same way that couples go out, get to know each other and live together before marrying).
A few questions that are worth answering before investing too much into tools and/or implementations:
The easiest way to get a taste of what your efforts will bring is to start with testing. Website testing is a pretty simple technique that can be implemented very easily in a small part of the website, maybe your landing page. If, for example, you create a few designs of your current landing page and test them, you will have the opportunity to prove to yourself and to your boss that testing (and later also Web Analytics) can increase the number of leads and sales you get from your marketing campaigns. (Tip: you can start testing for free using Google Analytics Content Experiments)
By showing real numbers to management you will have an easy job convincing them to allocate the necessary budget/resources to your Conversion Optimization efforts. And everyone will be happy.
Analytics minded companies have already gone through the process described above. They have already implemented and used their analytics solution to improve their conversion rates to some degree. Now what?
When a Web Analytics solution has been running for a while in a company, data is abundant, and analysis can bring important insights for the website optimization. For example, we will be able to understand which pages (or which kind of pages) are underperforming; we will also be able to understand if these pages are failing for a specific segment of visitors only, or for a specific metric. This understanding will help you plan and specify your test segmentation and Behavioral Targeting campaigns.
Therefore, for these companies I believe the best way of action would be to start with analysis in order to help with the prioritization of Testing efforts.
There is an old Jewish saying (from the Talmud I think) that truth is in the middle, i.e that usually a mix of two point of views is better than each of them separately. The same is truth in this case; if there is data (even if the implementation / configuration is not 100%), it should be used to maximize testing success; and testing results should be used to help following data analyses.