Recently, Moz announced the results of our biennial Ranking Factors study. Today, we’d like to explore one of the most vital elements of the study: the Ranking Factors survey.
2015 Ranking Factors Expert Survey
Every two years, Moz surveys the brightest minds in SEO and search marketing with a comprehensive set of questions meant to gauge the current workings of Google’s search algorithm. This year’s panel of experts possesses a truly unique set of knowledge and perspectives. We’re thankful on behalf of the entire community for their contribution.
In addition to asking the participants about what does and doesn’t work in Google’s ranking algorithm today, one of the most illuminating group of questions asks the panel to predict the future of search β how the features of Google’s algorithm are expected to change over the next 12 months.
Amazingly, almost all of the factors that are expected to increase in influence revolved around user experience, including:
- Mobile-friendliness
- Perceived value
- Readability
- …and more
The experts predicted that more traditional ranking signals, such as those around links and URL structures, would largely remain the same, while the more manipulative aspects of SEO, like paid links and anchor text (which is subject to manipulation), would largely decrease in influence.
The survey also asks respondents to weight the importance of various factors within Google’s current ranking algorithm (on a scale of 1-10). Understanding these areas of importance helps to inform webmasters and marketers where to invest time and energy in working to improve the search presence of their websites.
On-page keyword features
These features describe use of the keyword term/phrase in particular parts of the HTML code on the page (title element, H1s, alt attributes, etc).
Highest influence: Keyword present in title element, 8.34
Lowest influence: Keyword present in specific HTML elements (bold/italic/li/a/etc), 4.16
Titles are still very powerful. Overall, it’s about focus and matching query syntax. If your post is about airplane propellers but you go on a three paragraph rant about gorillas, you’re going to have a problem ranking for airplane propellers. |
Keyword usage is vital to making the cut, but we don’t always see it correlate with ranking, because we’re only looking at what already made the cut. The page has to be relevant to appear for a query, IMO, but when it comes to how high the page ranks once it’s relevant, I think keywords have less impact than they once did. So, it’s a necessary but not sufficient condition to ranking. |
In my experience, most of problems with organic visibility are related to on-page factors. When I look for an opportunity, I try to check for 2 strong things: presence of keyword in the title and in the main content. Having both can speed up your visibility, especially on long-tail queries. |
Domain-level keyword features
These features cover how keywords are used in the root or subdomain name, and how much impact this might have on search engine rankings.
Highest influence: Keyword is the exact match root domain name, 5.83
Lowest influence: Keyword is the domain extension, 2.55
The only domain/keyword factor I’ve seen really influence rankings is an exact match. Subdomains, partial match, and others appear to have little or no effect. |
There’s no direct influence, but an exact match root domain name can definitely lead to a higher CTR within the SERPs and therefore a better ranking in the long term. |
It’s very easy to link keyword-rich domains with their success in Google’s results for the given keyword. I’m always mindful about other signals that align with domain name which may have contributed to its success. These includes inbound links, mentions, and local citations. |
Page-level link-based features
These features describe link metrics for the individual ranking page (such as number of links, PageRank, etc).
Highest influence: Raw quantity of links from high-authority sites, 7.78
Lowest influence: Sentiment of the external links pointing to the page, 3.85
High-quality links still rule rankings. The way a brand can earn links has become more important over the years, whereas link schemes can hurt a site more than ever before. There is a lot of FUD slinging in this respect! |
Similar to my thoughts on content, I suspect link-based metrics are going to be used increasingly with a focus on verisimilitude (whether content is actually true or not) and relationships between nodes in Knowledge Graph. Google’s recent issues with things, such as the snippet results for “evolution,” highlight the importance of them only pulling things that are factually correct for featured parts of a SERP. Thus, just counting traditional link metrics won’t cut it anymore. |
While anchor text is still a powerful ranking factor, using targeted anchor text carries a significant amount of risk and can easily wipe out your previous success. |
Domain-level brand features
These features describe elements that indicate qualities of branding and brand metrics.
Highest influence: Search volume for the brand/domain, 6.54
Lowest influence: Popularity of business’s official social media profiles, 3.99
This is clearly on deck to change very soon with the reintegration of Twitter into Google’s Real-Time Results. It will be interesting to see how this affects the “Breaking News” box and trending topics. Social influencers, quality and quantity of followers, RTs, and favorites will all be a factor. And what’s this?! Hashtags will be important again?! Have mercy! |
Google has to give the people what they want, and if most of the time they are searching for a brand, Google is going to give them that brand. Google doesn’t have a brand bias, we do. |
It’s already noticeable; brands are more prominently displayed in search results for both informational and commercial queries. I’m expecting Google will be paying more attention to brand-related metrics from now on (and certainly more initiatives to encourage site owners to optimize for better entity detection). |
Page-level social features
These features relate to third-party metrics from social media sources (Facebook, Twitter, Google+, etc) for the ranking page.
Highest influence: Engagement with content/URL on social networks, 3.87
Lowest influence: Upvotes for the page on social sites, 2.7
Social ranking factors are important in a revamped Query Deserves Freshness algorithm. Essentially, if your content gets a lot of natural tweets, shares, and likes, it will rank prominently for a short period of time, until larger and more authoritative sites catch up. |
Social popularity has several factors to consider: (1) Years ago, Google and Bing said they take into account the authority of a social profile sharing a link and the popularity of the link being shared (retweets/reshares), and there was more complexity to social signals that was never revealed even back then. (2) My experience has been that social links and shares have more power for newsy/fresh-type content. For example, a lot of social shares for a dentist’s office website wouldn’t be nearly as powerful (or relevant to consider) as a lot of social shares for an article on a site with a constant flow of fresh content. |
Honestly, I do not think that the so-called “social signals” have any direct influence on the Google Algorithm (that does not mean that a correlation doesn’t exist, though). My only doubt is related to Twitter, because of the renewed contract between Google and Twitter itself. That said, as of now I do not consider Twitter to offer any ranking signals, except for very specific niches related to news and “news-able” content, where QDF plays a fundamental role. |
Page-level keyword-agnostic features
These elements describe non-keyword-usage, non-link-metrics features of individual pages (such as length of the page, load speed, etc).
Highest influence: Uniqueness of the content on the page, 7.85
Lowest influence: Page contains Open Graph data and/or Twitter cards, 3.64
By branching mobile search off of Google’s core ranking algorithm, having a “mobile-friendly” website is probably now less important for desktop search rankings. Our clients are seeing an ever-increasing percentage of organic search traffic coming from mobile devices, though (particularly in retail), so this is certainly not an excuse to ignore responsive design β the opposite, in fact. Click-through rate from the SERPs has been an important ranking signal for a long time and continues to be, flagging irrelevant or poor-quality search listings. |
I believe many of these will be measured within the ecosystem, rather than absolutely. For example, the effect of bounce rate (or rather, bounce speed) on a site will be relative to the bounce speeds on other pages in similar positions for similar terms. |
I want to answer these a certain way because, while I have been told by Google what matters to them, what I see in the SERPs does not back up what Google claims they want. There are a lot of sites out there with horrible UX that rank in the top three. While I believe it’s really important for conversion and to bring customers back, I don’t feel as though Google is all that concerned, based on the sites that rank highly. Additionally, Google practically screams “unique content,” yet sites that more or less steal and republish content from other sites are still ranking highly. What I think should matter to Google doesn’t seem to matter to them, based on the results they give me. |
Domain-level link authority features
These features describe link metrics about the domain hosting the page.
Highest influence: Quantity of unique linking domains to the domain, 7.45
Lowest influence: Sentiment of the external links pointing to the site, 3.91
Quantity and quality of unique linking domains at the domain level is still among the most significant factors in determining how a domain will perform as a whole in the organic search results, and is among the best SEO “spot checks” for determining if a site will be successful relative to other competitor sites with similar content and selling points. |
Throughout this survey, when I say “no direct influence,” this is interchangeable with “no direct positive influence.” For example, I’ve marked exact match domain as low numbers, while their actual influence may be higher β though negatively. |
Topical relevancy has, in my opinion, gained much ground as a relevant ranking factor. Although I find it most at play when at page level, I am seeing significant shifts at overall domain relevancy, by long-tail growth or by topically-relevant domains linking to sites. One way I judge such movements is the growth of the long-tail relevant to the subject or ranking, when neither anchor text (exact match or synonyms) nor exact phrase is used in a site’s content, yet it still ranks very highly for long-tail and mid-tail synonyms. |
Domain-level keyword-agnostic features
These features relate to the entire root domain, but don’t directly describe link- or keyword-based elements. Instead, they relate to things like the length of the domain name in characters.
Highest influence: Uniqueness of content across the whole site, 7.52
Lowest influence: Length of time until domain name expires, 2.45
Character length of domain name is another correlative yet not causative factor, in my opinion. They don’t need to rule these out β it just so happens that longer domain names get clicked on, so they get ruled out quickly. |
A few points: Google’s document inception date patents describe how Google might handle freshness and maturity of content for a query. The “trust signal” pages sound like a site quality metric that Google might use to score a page on the basis of site quality. Some white papers from Microsoft on web spam signals identified multiple hyphens in subdomains as evidence of web spam. The length of time until the domain expires was cited as a potential signal in Google’s patent on information retrieval through historic data, and was refuted by Matt Cutts after domain sellers started trying to use that information to sell domain extensions to “help the SEO” of a site. |
I think that page speed only becomes a factor when it is significantly slow. I think that having error pages on the site doesn’t matter, unless there are so many that it greatly impacts Google’s ability to crawl. |
The future of search
To bring it back to the beginning, we asked the experts if they had any comments or alternative signals they think will become more or less important over the next 12 months.
While I expect that static factors, such as incoming links and anchor text, will remain influential, I think the power of these will be mediated by the presence or absence of engagement factors. |
The app world and webpage world are getting lumped together. If you have the more popular app relative to your competitors, expect Google to notice. |
Mobile will continue to increase, with directly-related factors increasing as well. Structured data will increase, along with more data partners and user segmentation/personalization of SERPs to match query intent, localization, and device-specific need states. |
User location may have more influence in mobile SERPs as (a) more connected devices like cars and watches allow voice search, and (b) sites evolve accordingly to make such signals more accurate. |
I really think that over the next 12-18 months we are going to see a larger impact of structured data in the SERPs. In fact, we are already seeing this. Google has teams that focus on artificial intelligence and machine learning. They are studying “relationships of interest” and, at the heart of what they are doing, are still looking to provide the most relevant result in the quickest fashion. Things like schema that help “educate” the search engines as to a given topic or entity are only going to become more important as a result. |
For more data, check out the complete Ranking Factors Survey results.
2015 Ranking Factors Expert Survey
Finally, we leave you with this infographic created by Kevin Engle which shows the relative weighting of broad areas of Google’s algorithm, according to the experts.
What’s your opinion on the future of search and SEO? Let us know in the comments below.