Correlation and Causation in SEO

Home » News » Correlation and Causation in SEO

Correlation and Causation in SEO

Posted on

This blog post looks at correlation and causation in SEO (search engine optimisation). There is a huge amount of data available for your website performance and a lot of advice out there on how to improve your website traffic and rankings.

First of all let’s not be under any illusions.

There are no magic formulas or easy gains in Search Engine Optimisation.

It’s usually a combination of consistently doing many things well:

  • Producing an excellent website that is kept up to date and accurate.
  • Doing keyword research so you know what potential customers are searching for.
  • Promoting your website using conventional marketing, social media, outreach, PPC etc.
  • Be aware of and optimise the many ranking factors that help boost website traffic and rankings.

As there can often be a relatively long lead time between SEO actions and results you need to choose your actions wisely.

That means knowing the difference between correlation and causation and what is worth spending precious resource (time and money) on SEO initiatives that genuinely cause a boost or simply correlate: appearing to be linked but aren’t. That means that you may do more of one action and it may be a coincidence that it is improving your SEO rather than being a direct cause. Using the available data and your business sense is crucial to making the right choices.

Correlation is not causation and to succeed in anything you need to know the difference.

Correlation versus causation in the wider world – things that don’t

Statistician and author Tyler Vigen highlights many amusing examples of things that appear to correlate but cannot possibly be linked.

His website and book Spurious Correlations shows many graphs of validated statistics with classics such as the number of people falling who drowned by falling into a pool appearing to correlate with films Nicholas Cage appeared in:

Correlation between people drowning in pools versus Nicholas Page films
Data sources: Centers for Disease Control & Prevention and Internet Movie Database

A particular favourite of ours is the close correlation between the divorce rate in US state Maine and the per capita consumption of Margarine:

Divorce rate in Maine vs per capita consumption of margarine
Data sources: National Vital Statistics Reports and U.S. Department of Agriculture

There are many other examples but it does show that you it is all too easy to kid yourself that actions are working well or badly when there is little or no causal link.

Correlation versus causation in SEO and website design

In very crude terms there is a strong correlation between having a poorly designed and maintained website and rankings and traffic. Other factors come into play but look unprofessional and operate your website badly and various key performance indicators will suffer.

Frequency of publication and updates

WordPress developer Robert Ryan conducted an experiment to stop blogging for 251 days and these were his main findings:

  • Overall traffic to the site saw a major decline as it fell by 32 percent.
  • Organic traffic dropped by a massive 42 percent.
  • Traffic to the contact page was down by 15 percent.
  • Overall site conversions fell by 28 percent.

Here is the link to his experiment so you can review in more detail: The no-blogging experiment is over

Whether publishing frequently causes improved web traffic is a subject for debate. If you go for quantity rather than quality then you could be harming your SEO. Not publishing anything new or not improving your content over time reduces the likelihood of getting backlinks and social media mentions.

However If you use aggressive black hat SEO techniques and generate lots of poor quality links on poor websites then it is likely that your rankings and web traffic will be harmed. By how much is difficult to quantify but in extreme cases you may get a manual penalty from Google.

e-commerce websites and speed

If your e-commerce website is slow then there is a strong correlation between sales and website speed:

  • 47% of consumers expect a webpage to load in less than 2 seconds
  • 40% of people abandon a website if it needs more than 3 seconds to load
  • An eCommerce making $50,000 a day could potentially lose $1.25 million sales per year from 1 second page delay
  • Conversion rates increase 74% when load times improve from 8 seconds to 2 seconds

Source: The impact of speed in your e-commerce

Research on the effect of social media on conversions

While social media is important, Danny Richman has done some research into how much resource small companies are spending compared to other digital marketing

By analysing data from the Google Analytics Assisted Conversion data across 200 websites, social media still only contributed to fewer than 11% of all conversions.

Conversion data from various digital marketing actions

Source: Are small businesses wasting time and money on social media?

Joe Wannamker
By Bain News Service, publisher. [Public domain], via Wikimedia Commons
From our client base we could find no clear causation between the number of social media  followers and likes and website ranking.

There was a correlation at times but it was hard to separate out the effects of just good marketing and strong reputation and brand over all forms of media.

Good companies offering a great service with an excellent website tend to generate a strong social media presence with the right professional approach.


John Wanamaker (1838-1922), department-store magnate, once said, “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.”

With good SEO and the latest tracking and analysis software techniques we can do better than that but there is still some element of experimentation. Therefore it’s important to recognise the early signs of an action helping or hindering your business objectives for better rankings and web traffic. Correlation is not causation and to succeed in anything you need to know the difference.

Further reading