My area of interest is the web. Specifically, I’m interested in researching what it takes to make a successful website. This is a relevant question and also one that a lot of people have been trying to answer for years. The industry that’s been built up around this question is of course commonly referred to as search engine optimization (or SEO).

Websites, while they have evolved, haven’t changed fundamentally since the first sites were launched decades ago! People still perform web searches, still bookmark their favorite sites, and still engage with websites. Social media has overtaken a lot of interaction on the web and devices have moved from desktops to mobile…but websites are still the heart of the web.

As far as what some people would call “ranking factors” or “engagement factors”, my emphasis and concern is what can be shown to actually work. (As opposed to what is rumored to work: things like supposed leaked Google ranking factors, expensive SEO tools, or secretive blackhat SEO methods). This – finding out what actually can be demonstrated to work – is the way to empirically arrive at conclusions about what makes a website show up higher in the search engine results.

To measure success, I will make note of various metrics when testing my experimental website. These might include measuring some or all of the following:

• The search engine result numbers for specific keywords
• Total number and total frequency of search engine bots
• Total number, total frequency, amount of time spent on site, and number of return visits for humans
• Total number of the site’s indexed pages in major search engines (i.e. Google, Bing, Yandex)
• Speed of indexation for newly-published pages on the site

Each of these can be quantitatively analyzed.

My approach is inductive. In order to learn what will be successful, a number of unknowns must be discovered. Results from trying new things will influence the research as the feedback loop is altered. I do not have a solid theory as to what will work in each case.

My study will be experimental. My idea for an approach that will work well for a website is to use the “split-test” model (AKA “A/B Testing”). This methodology allows an experimenter to introduce and measure the outcome of treatments on a set of websites, on two different versions of a page, or on two different versions of a site asset (such as a “Buy Now” button for instance).

Obviously, when doing split-testing, the original version of the thing being tested is the control and the altered version is the experiment. The results are measured and recorded. Additionally, it should be noted that only one change should be made to the site at any one time in order to not influence the experiment with any additional outside factors.

To answer the basic questions regarding this type of research, here are my reasons.

• Why? I want to find out the best website characteristics that could be employed to rank well and gain visitors. If I knew these, I could create a successful website – I’d know how to build it in a way that it will rank highly and gain traffic.
• How? I propose starting the experiment with a “blank slate” – and actually use two roughly equivalent domain names (to see if domain name has an effect on indexation or ranking). Going forward, build identical websites and make one change at a time. When a “winner” is determined, make one change at a time on that site (using split-test experiments) and record the results.
• When? Daily, but possibly more or less often. Web/server analytics will be used to record and make sense of the data. This will allow trends to be visible and evident over verious timeframes.
• Whom? Not a “who” in this case, but a “what” –  a website.

I believe this research will lead me to know what works best and will provide some valuable guidelines for bloggers, ecom site owners, and other website producers so they can create a successful site more efficiently.

Suggested Discussion Group Name: TechSearchers