When Moz undertook this year’s Ranking Correlation Study (Ranking Factors), there was a desire to include data points never before studied. Fortunately, SimilarWeb had exactly what was needed. For the first time, Moz was able to measure ranking correlations with both traffic and engagement metrics.

Using Moz’s ranking data on over 200,000 domains, combined with multiple SimilarWeb data points—including traffic, page views, bounce rate, time on site, and rank—the Search Ranking Factors study was able to measure how these metrics corresponded to higher rankings.

These metrics differ from the traditional SEO parameters Moz has measured in the past in that they are primarily user-based metrics. This means that they vary based on how users interact with the individual websites, as opposed to static features such as title tag length. We'll find these user-based metrics important as we learn how search engines may use them to rank webpages, as illustrated in this excellent post by Dan Petrovic.

Every marketer and SEO professional wants to know if there is a correlation between web search ranking results and the website’s actual traffic. Here, we’ll examine the relationship between website rankings and traffic engagement to see which metrics have the biggest correlation to rankings.

You can view the results below:

Traffic correlated to higher rankings

For the study, we examined both direct and organic search visits over a three-month period. SimilarWeb’s traffic results show that there is a generally a high correlation between website visits and Google’s search rankings.

Put simply, the more traffic a site received, the higher it tended to rank. Practically speaking, this means that you would expect to see sites like Amazon and Wikipedia higher up in the results, while smaller sites tended to rank slightly worse.

This doesn't mean that Google uses traffic and user engagement metrics as an actual ranking factor in its search algorithm, but it does show that a relationship exists. Hypothetically, we can think of many reasons why this might be the case:

• A "brand" bias, meaning that Google may wish to treat trusted, popular, and established brands more favorably.
• Possible user-based ranking signals (described by Dan here) where uses are more inclined to choose recognizable brands in search results, which in theory could push their rankings higher.
• Which came first—the chicken or the egg? Alternatively, it could simply be the case that high-ranking websites become popular simply because they are ranking highly.

Regardless of the exact cause, it seems logical that the more you improve your website’s visibility, trust, and recognition, the better you may perform in search results.

Engagement: Time on site, bounce rate, and page views

While not as large as the traffic correlations, we also found a positive correlation between a website’s user engagement and its rank in Google search results. For the study, we examined three different engagement metrics from SimilarWeb.

• Time on site: 0.12 is not considered a strong correlation by any means within this study, but it does suggest there may be a slight relationship between how long a visitor spends on a particular site and its ranking in Google.
• Page views: Similar to time on site, the study found a small correlation of 0.10 between the number of pages a visitor views and higher rankings.
• Bounce rate: At first glance, with a correlation of -0.08, the correlation between bounce rate and rankings may seem out-of-whack, but this is not the case. Keep in mind that lower bounce rate is often a good indication of user engagement. Therefore, we find as bounce rates rise (something we often try to avoid), rankings tend to drop, and vice-versa.

This means that sites with lower bounce rates, longer time-on-site metrics, and more page views—some of the data points that SimilarWeb measures—tend to rank higher in Google search results.

While these individual correlations aren’t large, collectively they do lend credence to the idea that user engagement metrics can matter to rankings.

To be clear, this doesn’t mean to imply that Google or other search engines use metrics like bounce rate or click-through rate directly in their algorithm. Instead, a better way to think of this is that Google uses a number of user inputs to measure relevance, user satisfaction, and quality of results.

This is exactly the same argument the SEO community is currently debating over click-through rate and its possible use by Google as a ranking signal. For an excellent, well-balanced view of the debate, we highly recommend reading AJ Kohn’s thoughts and analysis.

It could be that Google is using Panda-like engagement signals. If a site’s correlated bounce rate is negative, that means that the website should have a lower bounce rate because the site is healthy. Similarly, if the time that users spend on-site and the page views are higher, the website should also tend to produce higher Google SERPs.

Global Rank correlations

SimilarWeb’s Global Rank is calculated by data aggregation, and is based on a combination of website traffic from six different sources and user engagement levels. We include engagement metrics to make sure that we’re portraying an accurate picture of the market.

If the website has a lower Global Rank on SimilarWeb, then the website will generally have more visitors and good user engagement.

As Global Rank is a combination of traffic and engagement metrics, it’s no surprise that it was one of the highest correlated features of the study. Again, even though the correlation is negative at -0.24, a low Global Rank is actually a good thing. A website with a Global Rank of 1 would be the highest-rated site on the web. This means that the lower the Global Rank, the better the relationship with higher rankings.

As a side note, SimilarWeb’s Website Ranking provides insights for estimating any website’s value and benchmarking your site against it. You can use its tables to find out who’s leading per industry category and/or country.

Methodology

The Moz Search Engine Ranking Factors study examined the relationship between web search results and links, social media signals, visitor traffic and usage signals, and on-page factors. The study compiled datasets and conducted search result queries in English with Google’s search engine, focusing exclusively on US search results.

The dataset included a list of 16,521 queries taken from 22 top-level Google Adwords categories. Keywords were taken from head, middle, and tail queries. The searches ranged from infrequent (less than 1,000 queries per month), to frequent (more than 20,000 per month), to enormously frequent with keywords being searched more than one million times per month!

The top 50 US search results for each query were pulled from the datasets in a manner that did not account for location or personalization in a location- and personalization-agnostic manner.

SimilarWeb checked the traffic and engagement stats of more than 200,000 websites, and we have analytics on more than 90% of them. After we pulled the traffic data, we checked for a correlation using keywords from the Google AdWords tool to see what effect metrics like search traffic, time on site, page views, and bounce rates—especially with organic searches—have upon Google’s rankings.

Conclusion

We found a positive correlation between websites that showed highly engaging user traffic metrics on SimilarWeb’s digital measurement platform, and higher placement on Google search engine results pages. SimilarWeb also found that a brand’s popularity correlates to higher placement results in Google searches.

With all the recent talk of user engagement metrics and rankings, we’d love to hear your take. Have you observed any relationship, improvement, or drop in rankings based on engagement? Share your thoughts in the comments below.