November 18, 2015

How Google Would Use Maths in SEO

Andreas Voniatis is a Data Scientist at Artios. He explains the benefits of taking a mathematics-driven approach to SEO

Google are pioneers. They’ve shifted cultures and transformed the way we use the internet, all on the foundation of data-driven analysis. Their algorithms for assessing and ranking pages are mathematically complex, highly sophisticated, and impressively accurate.

So why is it that most search engine optimisation (SEO) consultancies take a very different approach to their work? They attempt to outsmart Google with an approach that’s based on second-hand knowledge and estimated expertise. They don’t leverage the same pinpoint precision and mathematical accuracy that Google uses to rank client websites.

And that means they can’t scientifically audit client websites, provide advice on SEO practices, and deliver the best results.

The role of maths in analysis and auditing

Most SEO consultancies offer a monthly audit for their clients. These audits use a single tool or set of tools to assess various criteria across a website. The findings are then presented in a report, which is often accompanied by a set of SEO recommendations – the steps clients should take to improve rankings and increase traffic.

However, a mathematical approach shows these audits to be limited, even if they happen to be accurate.

An SEO audit looks at numerous criteria, from mobile page loading speeds to the reading age of content. But a snapshot of any variable doesn’t necessarily represent the bigger picture.

The speed at which pages load will change through the day, as the server experiences different loads. Similarly, the average reading age of your content will vary as you add, remove, or revise individual pages.

A statistician understands that a thin slice of data isn’t enough. Instead, a successful audit begins with carefully designed data experiments for each and every variable. It’s only when you understand how often variables change that you can calculate how often they should be measured. And it’s only when your measurements are accurate that your findings hold real value.

The role of maths in understanding best practice

For the most part, the established best practice for SEO comes from third-party sources. While Google and other search engines publish their own advice, an entire industry has sprung up around publishing articles, blog posts, and guides to the latest algorithm changes and what businesses can do to keep up.

However, this has led to best practice that is generalised and sweeping. In simple terms, what works for one client won’t always work for another.

Take meta title tags, which recommendations often state should be less than 60 characters long. That blanket guidance doesn’t consider:

– Type of content
: a long title may not perform well on a product page, but could be useful in ranking a blog post or article
· Industry: some industries use longer title tags on a regular basis, with no real proof that a shorter title would perform any better

Truly scientific SEO isn’t merely based on best practice and guesswork. If Google ran an SEO firm, they’d make the calculations themselves, segmenting content types, industries, and much more.

As a result, they wouldn’t need to settle for broad best practice. Instead, they’d understand content in its proper context – the very thing that search engine algorithms analyse.

The role of maths in predicting real results

Finally, a mathematical approach allows you to identify the specific factors that really make a difference to your traffic. Fundamentally, this is based on statistical analysis across client websites and, crucially, competitors.

Through data-driven analysis, it’s possible to get a complete picture of client websites – every page, across various times of the day. Then, this can be compared to competing websites. Together, this analysis can be used to predict the mathematical probability of increasing traffic through a specific change.

And it doesn’t stop with implementing the appropriate changes. The same robust analysis can then be used to measure:

· Which SEO recommendations actually delivered
· The amount of increased traffic
· The accuracy of predictions
· The length of time required for changes to deliver ranking value

It’s SEO with clear, structured predictions and measurable impact. What’s more, when coupled with an artificially intelligent algorithm that learns over time, it’s SEO that can grow more sophisticated with every cycle of audit, alteration, and analysis.

Sophisticated SEO – driven by mathematics

For Google, maths has allowed them to develop the most accurate algorithms for assessing and ranking web pages. It’s led to an algorithm that’s so sophisticated it can’t be tricked into preferential treatment.

But maths also offers big opportunities to those on the other side of the table – the companies that want to maximise search performance. Using the same attitude, approach, and ethos that Google themselves adopt, data-driven processes lead to more intelligent analysis, smarter predictions and clear, concrete results.