Applying the Scientific Method to SEO

Search certainly isn’t rocket science, but it can be much better understood with a little bit of basic science background.

For those who don’t work in the discipline, the world of search engine optimization is usually perceived as dark and mystifying. The field itself sometimes gets a negative image within the industry because it is greatly misunderstood (the ugly duckling, if you will), but when executed correctly, it displays all of its glory (a la beautiful swan). Search certainly isn’t rocket science, but it can be much better understood with a little bit of basic science background. Over the past several years, I began to notice that the perplexities of the search process could be distilled when paralleled with the scientific method.

It was all the way back in elementary school when we first learned about the analytical process of the scientific method. At the most fundamental level, the method consists of five main steps: observe, research, hypothesize, experiment, and draw a conclusion. The notion of “always be testing” will yield the most value in search when these five steps of the scientific method are applied.

The Five Steps to SEO by the Scientific Method

1. Make observations. To translate this into search terms, think of this step as an overall characterization of your website’s health. The characterizations are the measurements and baseline performance statistics (e.g. CTR, conversion rate, etc.) that will be used as testing variables and controls in forthcoming steps, as well as measures of success. This step also includes an in depth characterization of the client’s business objectives and revenue goals. By defining these elements, we now have a solid starting point to proceed with a closer examination.
2. Conduct background research. This step includes a quantitative data analysis (tools such as Omniture, Forrester Research, Quantcast are great to have access to here) in order to obtain deeper insight into the types of visitors that are coming to your website and its competitors. This data is also useful in revealing any untapped pockets of consumers within the market and identifying the key need states. In conjunction, a comprehensive site audit will need to be performed to uncover anything that may be obstructing conversion. Finally, keyword research can be conducted based upon correlations with the gathered data.
3. Form a testable hypothesis. With the data acquired from the observation and research phases, you can now postulate a sound hypothesis. This ties back to the very first step in which we characterized baseline performance statistics. For example, “Using branded keyword phrases yields a 35% higher CTR than using unbranded keyword phrases” is a testable hypothesis that uses a key performance indicator, CTR, as the dependent variable.
4. Test your hypothesis. The testing phase in the search process involves technical integration and development, content strategy and development, link strategy and deployment, and a lot of patience. It’s during this testing period where we begin to bridge the connections between keyword phrases and the consumers searching for them in order to effectively provide optimization recommendations. By explaining these connections via market research and need state analysis (as opposed to marketing jargon), we’re able to build a multi-channel bridge that will apply its positive affect across all capabilities (paid, social, UX, etc).
5. Analyze and draw conclusions. The final step of the process involves measuring results and optimization. Results should be reported in a tailored fashion based upon the KPI’s and business objectives. Although rankings and traffic values are important to consider, they are not the determining metric of success. The true indicator is if conversion was actually attained. At this point, it is crucial to inspect all possible variables that are affecting that output. This is where the optimization comes into play – usability and A/B tests are performed, variables are adjusted, and the entire process repeats itself in order to achieve more desirable results (i.e. greater conversion rates).

In the case of search and the scientific method, a “true” hypothesis will never truly be reached because of the sheer erraticism of search engines – you should and will always be testing. Google is well aware that like with any experiment, too much consistency will burn it out because systemic configurations can’t be sustained by a dynamic structure (that structure being the Google algorithm). Although no definitive process can be constructed to layout the optimal way to perform search engine optimization, the scientific method does provide an organized approach to studying and better understanding this imperative phenomenon.