Prev
The following blog post is actually a thesis written by Josh Lavine, a student at Princeton University who's task was to interview a start-up company, preferably in the hi-tech area for an Entrepreneurship class. It is quite long, but describes how Pear Analytics was started and where it is going, the challenges they face, and more.
(It is quite an interesting view of SEO from outside. Gregory PS)
...
SEO consultants try to determine what Google's algorithm is, or at least, what aspects of a website it weighs as more important than others. Every SEO expert believes something different from the next one about what is most important, and many of them get good results using completely different methodologies. Kelly has his own biases and beliefs regarding Google's algorithm, and he advises his clients based on them with measurable success.
The Evolution of Pear Analytics
In early 2008, Kelly's consulting business was going well. He had a number of customers who were willing to pay him about $3,000 dollars/month for six months for his personal SEO services. After a few months, however, Kelly realized that he was doing exactly the same thing with each customer. He would gather data about their website—keywords, number and quality of inbound links, sitemap structure, webpage layout, load time, etc. and then diagnose the problems and develop a plan to fix them. The problem, as with every SEO consulting business, was that gathering data usually took an incredibly long time. The time was spent passively diagnosing problems instead of actively fixing them. But because the process was essentially the same, Kelly's engineering background kicked in and he started asking himself if the process could be automated.
He opened a blank PowerPoint document and wrote a series of mathematical if-then statements that would later become the algorithm for his automated website data gathering program called SiteJuice. His algorithm would tell a computer to "crawl" a website (scan it), and by gathering and analyzing a vast amount of data that would normally take a human days to collect, within minutes pop out a score, from 1 to 100 that indicated how optimal the website was. It would also provide an analytic breakdown of the website in many different SEO related categories: keywords, website structure, tagline, etc. Kelly describes the algorithm as "a series of nested weighted averages", which means that it analyzes many separate factors, scores them based on his rubric, and then weights each of them in a final average to give one complete score. The weights Kelly determined are based on what he believes is important for SEO. (As aforementioned, there is speculation and debate among SEO experts as to what is most important. Kelly's own beliefs bias the weighted averages in his algorithm.) The final breakdown assesses the site for the biggest weaknesses and helps Kelly determine what needs to be done to fix the website.
Read more.