In August of 2019, 1,584 professionals in the field of Search Engine Optimization (SEO) took a survey sharing their opinions on the relative use and merit of various inputs in Google’s ranking systems. This report shares the aggregated results of that survey. Each year, the survey will be repeated to show trends of how ranking factor opinions shift.
How Google Weights Ranking Inputs
Respondents were asked whether they believed ranking inputs were fixed in the algorithm’s weighting system, or whether certain types of queries (or all queries) weighted ranking elements differently.
The responses show that nearly 2/3rds of survey takers believe Google has a wide variance of how ranking inputs are weighted depending on the query words used. This makes analysis of the ranking systems vastly more challenging (as a ranking input might be very important for one set of queries, and relatively unimportant for another). Hence, in the ranking input weighting results below, results should be interpreted with knowledge of this potential variance. I.E. SEOs generally believe certain factors outweigh others in importance, but not necessarily in the universality of this ordering.
Ranking Factors Overview
The question text shown to each participant was the same: “For each of the following factors, enter your opinion of how much weight it receives in Google’s organic ranking systems (the classic, ten-blue-links style results).”
Participants in the survey were given a 0-10 scale to rate each of 26 ranking factors with the following labels:
- 0 – Not Used
- 5 – Moderately Weighted
- 10 – Very Heavily Weighted
The visual below illustrates the results from highest average (8.52/10) to lowest (4.19/10).
Consensus vs. Variance of Opinions
Using standard deviation, we can see how on some factors, survey-takers generally agreed (i.e. their responses were tightly clustered around a number) vs. had more disagreement (i.e. responses were spread out). The visual below ranks the factors by level of disagreement.
As might be expected, there’s relative consensus around the top ranking factors (“relevance of overall page content,” and “quality of linking sites & pages”), and more disagreement toward the bottom of the list. This suggests that some SEO practitioners still feel very strongly that factors like keyword-use in domain name, and age of website are powerful influencers, while others think they have little to no impact.
Most interesting to me was the relatively high disagreement on two factors in particular: “Use of Google” AMP and “Content accuracy with accepted facts.” Given that both of these are, in my opinion, used situationally in the ranking algorithms, the high distribution of responses makes sense. One could argue that Google AMP is “all or nothing” factor — in the AMP box on mobile, it’s essential to even being considered, and in all other cases, it makes little difference. A similar case could be made for content accuracy — that it’s used when Google’s applying a high trust parameter to YMYL-type queries, and not applicable at other times.
Comparison of Self-Described “Top 10%” of SEO Professionals vs. All Respondents
Those taking the survey were asked to rate their level of SEO knowledge and experience from “0” (New to SEO) to “10” (Top 10% of the Field). The distribution of those responses is below.
The visual below compares the responses from the 8.1% (129) of survey-takers who said they were in the “top 10%” of the field to the average across all survey-takers.
Perhaps surprisingly, the comparison is not particularly striking. The most variance comes on keyword use in the URL (-1.06) and age of website (-0.97). In general, those with more self-described knowledge and experienced rated all factors a little lower than those with less, perhaps evidence that those with greater experience find more complexity in Google’s ranking systems.
Which trends will have the biggest impact on SEO in the next 3 years?
Each survey-taker was asked to rate the following trends based on their perceived impact to the field of Search Engine Optimization (SEO) over the next 3 years.
These results show that professional SEOs generally think Google’s own activities, especially their layout choices in the SERPs (Search Engine Result Pages), their decisions to enter more verticals as a publisher and competitors, and their technological/product advancements will have a far greater impact than any government, competitive, or outside influence.
Rand’s Personal Analysis
I compared this year’s data to the aggregated opinions from the previous Moz Ranking Factor opinion surveys across the prior 14 years (2015, 2013, 2011, 2009, 2007, 2005), and found a number of fascinating trends:
- For the first time, content > links & keywords: In the early days of the survey, keywords were the top-voted ranking factors, then, for nearly a decade, links did. Now, content relevance and quality dominate. I think this perception is generally correct (though links are still a powerful #2), and it reflects the great strides Google’s made in understanding content that satisfies searcher intent.
- The perceived value of anchor text is diminishing: Anchor text of links had been a mainstay in the top few ranking factors every year until 2019. It’s not even in the top 10 anymore. I think that’s probably incorrect as an absolute assessment, but I agree that in general, Google has been moving away from an over-reliance on that factor across the prior two decades (1998-2018).
- Mobile friendliness & load speed are bigger than expected: My impression from case studies and Google’s public statements are that these elements are relatively small direct ranking factors (though, I’d posit that indirectly, they nudge things like link earning, engagement, and other important inputs). Seeing them so high was a surprise, and may reflect that modern SEO often conflates correlation and causation (though, as I’ve often argued, correlation in the SEO field is not only interesting, but useful).
- Amount of Content & Age of Website: It’s my belief that, while both are correlated with higher rankings, neither of these are technically used by Google to rank web pages. I’m surprised (and a little disappointed) that they scored so highly.
- Filtering by Geography: One of the biggest misses of the survey this year was my failure to ask where the respondents did the majority of their SEO work. It’s my perception that the importance of various ranking inputs has great variance depending on query language and geography, and I hope to show that in future years.
On the trends front, I found the results initially surprising, mostly because I’m a skeptic on the impact of voice answers (which scored highly), and strongly expect the joint investigations from various branches on the US government to results in significant changes to Google. However, on reflection, I believe the “3 years” timeframe is perhaps responsible for these results. An investigation and subsequent court battle could take significantly longer to resolve.
Methodology & Survey Phrasing
- 1,584 responses were collected via a Typeform survey published from August 6th – August 27th.
- Twitter, LinkedIn, and Email were the primary collection methodologies.
- 920 responses came from desktops, 654 from mobile phones, and 15 from tablets
The text of each ranking factor from the survey was simplified to create the visuals and data charts. Full text as used in the survey is below for those interested in the precise wording (which tended to be more explicit and explanatory).
“For each of the following factors, enter your opinion of how much weight it receives in Google’s organic ranking systems (the classic, ten-blue-links style results).
If you believe signals are differently weighted based on the query, assume this question refers to the average weight of that signal across all queries.“
- Relevance of the page’s content to the query (i.e. is the text topically relevant to the searched-for keywords)
- Quality of the websites and pages linking to the page
- Use of words, phrases, and content Google might deem “highly relevant to” or “crucial to answering” the query (apart from the query term itself)
- Google’s perceived expertise, authority, and trust of the host domain
- Mobile friendliness of the UI/UX
- Exact (or near exact) use of the searched-for keywords in the content, title, and meta data of the page
- Quantity/diversity of the websites linking to the page (i.e. more unique linking domains vs. many links from the same sites)
- Accuracy of the content (i.e. whether, from Google’s perspective, the page’s/site’s content is truthful and correct)
- Link authority of the host domain (based on the quantity and quality of the links that point to the entire website)
- Google’s perceived expertise, authority, and trust of the individual page’s content (and, if identifiable, the author behind it)
- Use of entities relevant to the query (like names, concepts, places, etc) in the page’s content
- Web page load speed
- User & usage data signals such as searchers’ click-preferences, bounce rate relative to other pages/sites in the rankings, pogo-sticking, engagement, etc.
- Freshness/recency of the content’s publication
- Anchor text of links pointing to the specific, ranking page
- The location, frequency, and distance of words and phrases closely related to the searcher’s query in the text of the page’s content
- Total amount of content on the page
- Use of unique images/visuals relevant to the query
- Site accessibility factors (like use of alt text on images, screen-reader friendliness, use of color, design of online forms, header us, resizable text, etc)
- Anchor text of links pointing to other pages on the host domain
- Keyword use in the URL
- Mentions of the host domain or its associated brand in content around the web (aka “unlinked mentions”)
- Age of the website/domain
- Use of Google’s AMP web component framework
- Presence of external links in the page’s content (i.e. linking out to other websites)
- Keyword use in the Host Domain Name
“How much of an impact do you believe the following trends will have on SEO in the next 3 years?”
- Voice search as a query input (i.e. a searcher speaks their search to a mobile or desktop device instead of typing it)
- Voice-answered queries like those offered by Google’s Assistant, Alexa, Siri, etc. (i.e. a searcher receives a spoken-voice answer without a screen of results)
- Zero-Click searches on Google (queries that result in no traffic to the sites that appear on the SERP)
- Advancements in machine learning and artificial intelligence
- Changes to the quantity and presentation of ads in Google’s search results
- Google entering more verticals and competing directly with publishers in their results (e.g. Google Flights, Hotels, Jobs, Events, Maps, Play, Books, etc)
- Government intervention in the technology and web landscape (e.g. EU’s GDPR, Articles 11+13, US Justice Dept. investigation into Google’s antitrust behavior, etc)
- Loss of cookie, visit, and web tracking data (from privacy-focused changes from browsers, tech company changes, and govt. requirements)
- Visual search advances like Google Lens, photo-based querying, or other image-search technology leaps
- Google Discover and other latent, content-nudging/engagement technologies
- Government intervention in Google’s operations from Congressional investigations, Justice Department actions, plans like Bernie Sanders’ or Elizabeth Warren’s to break up Google, or other anti-monopoly activity
The raw data for each factor, including mean, median, and standard deviation, are in the chart below:
For feedback on this document, or suggestions for what to add in future iterations, tweet to @randfish or email Rand at SparkToro.