Google search ranking factors no longer apply universally but must be analyzed on an industry-specific basis, says Searchmetrics in releasing its latest annual search rankings study [download page]. Still, it’s important to understand overall trends and general rankings factors, with this year’s final industry-wide edition indicating that “the most important ranking factor is content that is perfectly aligned to the user intention, together with an optimized page architecture.”
In other words, user signals – such as click-through rate – are tremendously important as a general search rankings factor. A new element introduced this year – content relevance – reflects a “dynamic between individual content relevance and user intent,” while technical factors are considered still a “prerequisite” for high rankings.
Following, some key results from the study, organized by area. Enlarge the chart above to see the full correlation values.
A quick note on interpreting correlation in this context: as explained by Searchmetrics, “a high correlation indicates large differences in terms of the given ranking factor amongst the analyzed URLs, when ordered from position 1 to 20. This is not to be confused with the ranking factor’s importance.” Readers are encouraged to review the methodology note at the end of this article as it contains more details from Searchmetrics about how to interpret the data.
- While content relevance does not have a strong correlation within the top 20, the results indicate that content relevance for both the entire page and the specific area decrease alongside search result positions, with positions 3-6 being the URLs with the highest content relevance. Searchmetrics is also of the opinion that (keeping in mind that correlation within the top 20 doesn’t necessarily signify ranking importance – see methodology note) that this is an area that’s becoming increasingly influential.
- The importance of individual keywords – whether in the title, description or H1 – continues to drop, and only 53% of the top 20 URLs this year included the keyword in their title.
- Click-through rate emerges as the ranking factor with the strongest correlation (0.46) overall among the top 20 positions.
- Pages occupying positions 1-3 have an average click-through rate (CTR) of 36%, with the top position having an average CTR of 44% and the third position of 30%. Notably, CTRs are higher for the top of the second results page than the bottom of the first results page.
- The average bounce rate for URLs appearing in the first page of results is 46%, while average time on site for the top 10 URLs is 3 minutes and 10 seconds. Both are up from 2014, with time on site having a higher correlation coefficient (0.11) than bounce rate (0.03).
- While existence of H2 in source code doesn’t have a correlation within the top 20 results, the analysts note that using them can offer a competitive advantage.
- Page encryption using HTTPS is becoming more important, and has the highest correlation (0.19) of the technical factors examined, followed by the existence of H1 headers (0.1).
- Neither .com domains, file size, or site speed appear to have a positive correlation within the top 20, though that doesn’t mean that they’re not important (and they certainly are for the user experience). Interestingly, the study found that the load time for top-ranked mobile pages are about a second faster than their desktop equivalents.
- The top 100 domains by SEO visibility all have a mobile-friendly solution such as a mobile sub-domain, responsive design or mobile app. By contrast, more than one-fifth of those outside the top 100 are not mobile-friendly.
- While internal links are very important, the importance of link volume appears to have dropped in recent years, supplanted instead by a more relevant and logical structure. Mobile URLs were found to have roughly 40% fewer internal links than desktop URLs.
- While video integration doesn’t correlate with rankings in the top 20, the number of images does have a positive correlation. In fact, websites within the top 20 averaged 1.67 images (larger than 200 pixels) per page.
- Higher-ranking pages seem to contain more interactive elements, though fewer are found on smartphones.
- The use of flash elements in top-ranked sites has seen a marked decline, and the analysts argue that “on mobile, the need for universal compatibility across devices has won and flash is dead.”
- There continues to be a strong positive correlation for social signals within the top 20.
- Facebook appears to have the strongest correlation, and has easily the highest level of user interactions. It is also the network with the highest signals across the first page, with Google+ and Tweets both showing steep declines between the first and second positions.
- The correlations for backlinks remains quite high, but these factors are trending down in importance, leading the analysts to say that they have “now become just one of many contributing signals.”
- The backlink factors with the highest correlations this year are the number of NoFollow backlinks (0.22), the number of backlinks (0.18) and the number of DoFollow backlinks (0.17).
The full report can be downloaded here.
Methodology Note: This explanation has been provided to MarketingCharts by Searchmetrics:
“Correlation coefficients are a starting point for understanding the importance of individual ranking factors ”“ for example a high positive score means that more of a factor is present at successively higher positions. However, in each case, Searchmetrics takes a closer look at the relationship between the ranking factor and how it varies across positions 1 to 20 to help get a clearer sense of how important it is.
For example, certain technical factors may be present in nearly all results in the top 20, meaning there is a correlation coefficient of close to 0. This does not necessarily signify that these technical factors are unimportant ”“ but rather that they are ever-present and potentially are a prerequisite for ranking in the top 20 positions.
For other factors (such as social signals) we measured huge positive increases in the factor at successively higher ranking positions. However, although the correlation coefficient is high ”“it does not necessarily mean that it’s a strong signal. The correlation coefficient just measures the differences of the ranking URLs with regard to the examined factor – it does not mean that Google is ranking these URLs more highly because of this factor.”
About the Data: Searchmetrics notes the following in its report:
“As in previous years, the general ranking factors and rank correlations are based on a set of 10,000 relevant keywords. For some factors, a more in-depth analysis required the definition of specially-defined keyword sets.
All correlations are always based on the complete dataset. In the past, we excluded Wikipedia results from some mean value calculations. We have now done this across the board for all factors, because Google’s ranking algorithm seems to apply non-standard criteria to the online encyclopedia.
Furthermore, median values are often provided, as these give a more accurate impression of the real trend, whereas mean values are sometimes disproportionately skewed by outlying values. Any exceptions are clearly indicated on the appropriate graphs. Wherever relevant and useful, we have also included a comparison with the previous year’s results.”