SEO Research in 2019 - A Summary of Rank Ranger's Findings
December 18, 2019 |
The best part of my job is the ability to spend time pondering, analyzing, and researching anything and everything within the world of Search. It's in these moments where I feel I'm making a difference by helping advance the SEO conversation. With this sentimentality, let's have a look back on 2019 and revisit some of the insights we've uncovered here at Rank Ranger over the year.
Before we get into the research itself, I want to thank everyone on the Rank Ranger team from the editing crew to the development team. Though it's my name on these studies, it's far from a one-man show. Without the help of the entire team, there is no way I would be able to bring these SEO insights to life.
I specifically want to thank Levi Genesove for all of this diligent work, thoughtful advice, and countless hours helping with everything from number-crunching to editing. I'd also like to thank our SEO manager Tzvi Barish for all of his help and insights as well as Rank Ranger founder Shay Harel for being a sounding board par excellence!
Trends, Insights, and Data on Google's Core and Unconfirmed Algorithm Updates
I've often mentioned that there is nothing as fertile than Google's core updates for the latest insights into the underpinnings of Google's relationship to web content and sites per se. At the same time, isolating site-level trends can be both laborious and enigmatic. With each of 2019's core algorithm updates, and at times with Google's unconfirmed updates, I've tried to offer a per niche rank volatility breakdown along with any site-level patterns that may have presented themselves. To the latter, I've at times have had great success while in other instances have ended up grasping at straws. So it goes in the great game that is analyzing Google's algorithm.
That said, here is a short synopsis of what I've found over the course of the year.
Data on the March 2019 Core Update
Believe it or not, but the March 2019 Core Update was the first confirmed update to follow the famed Medic Update
. As such, it only made good sense to compare the power and scope of the two core updates to each other.
Overall, the Medic Update was significantly more volatile than the March 2019 Core Update. While certain niches, such as the Travel and Retail niches, did show more volatility at the top of the SERP in March, the general trend clearly showed that Medic was the more potent algorithm change. This was particularly true for the Health and Finance niches.
See below for a full analysis of the relative volatility seen during the two updates:
With the March update, there were all sorts of speculation indicating the update was a reversal of the Medic Update. From my analysis, this was not the case. While there were some sites impacted by Medic that displayed a reversal trend, it was not the overall pattern. More than that, these sites did not see a full return of their rankings but rather a slight increase that put them (on the whole) nowhere near their former ranking levels.
Niche Trends from the June 2019 Core Update
Just three months after releasing the March 2019 Core Update, Google blessed us with the second confirmed update of 2019... the June 2019 Core Update. Isolating site-level trends for this particular update was unusually difficult as Google unveiled its new domain diversity construct
at the exact same time as the June 2019 Core Update.
That said, at the niche level the June update continued the trend where Your Money Your Life (YMYL) sites were left more impacted by a core update. (Read my analysis for more on why the core update YMYL trend
exists.) Looking below, the data indicates the Health and Finance niches being far more impacted by the update than the Travel and Retail industries:
Nuanced Content Profiling with the September 2019 Core Update
If I keep mentioning that finding a trend among the sites hit by a Google update is difficult, it's because it is. However, every once in a while you find a gem. The September 2019 Core Update was one of those times.
The overall volatility seen with the September update was a bit on the low side, at least compared to the June update. That said, this was a core update, and as you can see below, it brought a significant amount of rank volatility with it:
That said, the interesting news out of this update was the way Google looked at content. One of the themes within this update (as there are many trends and themes within any update) was the way Google almost seemed able to understand tone. When looking at the data that came out of the update I noticed that more than a few loan sites lost a significant amount of their rankings.
My initial thought after looking at some of these sites was that they were victims of having thin and templated content. However, a deeper investigation showed that sites with a prolific amount of content were still negatively impacted by the update. Upon closer investigation, it became apparent that the use of marketing language in inappropriate places was an issue for these sites (along with not offering enough information on the loans themselves and while offering too much information on how to secure one).
In other words, and generally speaking, sites that discussed loan details from a purely informative point of view maintained stable rankings while those sites that spattered their informational content with marketing-like phraseology were subject to being negatively impacted by the update.
Analyzing the July 2019 Unconfirmed Update
Yes, I am well aware that July comes before September. However, being that the July update was not a confirmed core update, I'm placing it out of sequential order for our purposes here.
The reason we're talking about this unconfirmed update versus the dozens of others Google releases is due to the extremely high levels of rank fluctuations presented by the update:
Accordingly, the rank volatility among the various niches during the update rivaled that which was seen during the June 2019 Core Update:
Of course, the major takeaway here is that unconfirmed updates can rival the confirmed core updates
and should not be ignored.
As far as the sites impacted themselves, there was not a particular YMYL bent to this update (which furthers the theory that the core updates have their own agenda). In fact, unlike a core update, it was extremely hard to find a consistent site-level pattern (far harder than usual that is).
There were some interesting "reversals" however. Specifically, I noticed that there were a nice number of sites that saw slow sustained growth from the start of July which then saw these ranking gains washed away by the July update:
At the same time, there was also a group of sites that saw massive ranking increases as the update rolled in only to see those gains eradicated during the second day of the update's rollout:
Featured Snippet Research in 2019
Though I didn't plan it as such from the onset, I spent a lot of time focused on Featured Snippets
in 2019. As I said, this was not intentional. It was more a case of doing one study that got me thinking about something else which led to another study and so on.
That said, what we found with our series of Featured Snippets studies was some of the most interesting "stuff" we've produced.
On Featured Snippet URL Stability
For our first study of 2019, we took 350 keywords that produced Featured Snippets and analyzed them over a 30-day and 90-day period to determine how often Google swapped out URLs within a Featured Snippet. As part of this analysis, we also looked at how many URLs Google is making use of within the average Featured Snippet over both a 30-day and 90-day period.
The data showed that on average Google uses 1.7 URLs making 2.2 swaps over a 30-day period, with the data being relatively proportionate over a 90-day period:
That said, different intents and keyword variants yielded slightly different results. For example, keywords that contained the word best
when associated with a product or service (as in best online broker
or best freezer
) underwent more URL swaps within the Featured Snippets:
Featured Snippet URL Market Share
Our study on Featured Snippet URL swaps left me wondering how much market share was given to each URL utilized for a particular Featured Snippet. If Google is using roughly 1.7 URLs within a Featured Snippet over a 30-day period (as was shown by the study) how much time was each URL getting within a Featured Snippet?
To answer this we analyzed 300 Featured Snippet producing keywords that utilized only two URLs over a 30-day period (since you can't actually find a Featured Snippet making use of 1.7 URLs over 30-days).
What we found was that when a snippet uses just two URLs there is a clear favorite with one URL garnering 77% of the market share over 30 days:
Of course, we still wondered what happened when a Featured Snippet made use of more than two URLs over 30 days. So, we took a look at Featured Snippets that made use of three URLs over a 30-day period and found Google still strongly prefers one of the URLs implemented:
Naturally, there are a ton of implications from this data related to how and when to target Featured Snippets!
Mobile vs. Desktop Featured Snippet Consistency
Is a Featured Snippet win on desktop an automatic win on mobile (and vice versa)? It's a good question that I was a bit surprised to find had not been tackled until our study released on October 23, 2019!
Surprise, surprise, the synchronicity of Featured Snippet URLs across devices is not as strong as you would intuitively think. If you consider instances where only one device shows with a Featured Snippet on a given day... there is a 30% schism! That is to say, if you win a desktop Featured Snippet that win will not translate to mobile 30% of the time (over the course of 30-days).
Even considering days where a keyword produces a Featured Snippet on both desktop and mobile there is a 10% divide. Meaning, 10% of the time Google will show a different Featured Snippet URL on mobile than the one being shown on desktop. That's still pretty high in my estimation!
How Long are Featured Snippets?
Did you know the average paragraph Featured Snippet is 269 characters while the average list Featured Snippet is 315 characters? It's true. However, it's only half the story as list Featured Snippets are 9% longer than they once were while paragraph snippets are about 5% shorter than they were previously!
What's the commonality between Featured Snippets getting longer when in the list format and shorter when shown as a paragraph? Directly answering the query. A longer list means more complete information which means fewer clicks. At the same time, a shorter paragraph snippets means less content bloating. It means the snippet is more refined and can better directly answer the user's query... which also means fewer clicks!
Intent within the People Also Ask Box
No, the People Also Ask (PAA) box is not technically a Featured Snippet but they do use the same format... so I'm throwing this study together with the ones we did on Featured Snippets per se.
The PAA is box is everywhere as it appears on roughly 50% of all page one SERPs! With that, I was curious to know how diverse are the intents reflected by the initial four questions shown in the PAA box?
And the answer is... pretty diverse as the average PAA box reflects nearly three unique intents!
Interestingly enough, the first intent reflected by the first question within the PAA box received a double portion. That is, the initial intent within the PAA box was reflected within an average of nearly two questions (whereas the other secondary intents each received a signal question apiece):
Getting a Grasp on Google's Entity Understanding
Entity understanding is the future of SEO. I know that might be both a bold and in a way premature statement, but I firmly believe it. How well Google is at understanding entities directly correlates to how well Google understands web content and the sites that content resides on.
To this, we published two case studies: One showing the limitation of Google's entity understanding and one showing how Google has made great strides in the area.
Google's Ability to Understand Secondary Entity Profiles
I'm a bad news first kind of guy, so here's the bad news... Google has shown that it has a very hard time understanding an entity's secondary profile. What do I mean by a secondary profile? I mean entities that have multiple identities. For example, Ronald Reagan was both a famous actor and
a US president - two separate profiles. How do I know Google is not great at picking up on secondary profiles? Well, here's the Related Search boxes you got for Ronald Reagan a few month's back:
Not only do these results not reflect the entity's secondary profile (i.e., being an actor) but the second box is pretty bizarre in its own right.
The good news? Now when you search for Ronald Reagan you get a Related Search box entitled Conservative Actors
Meaning, since I published this case study Google has indeed gotten far better and understanding an entity's secondary profile! It's still far from where it should be, as many of the examples I highlight still don't touch on the entity's secondary profile, but it's a good start.
Google Leaps Forward with Entity Sub-Classification
Whereas understanding an entity's multiple profiles has proven difficult for Google, drilling down into an entity and subclassifying it has been a relative cinch! In this case study, I went through how Google is getting mighty good at classifying segments of an entity's identity. To highlight this we looked at a plethora of Knowledge Panels related to sports teams highlighting the various ways Google has segmented the topic of "players" within these panels.
What we found Google doing was taking the 'players' element within these panels and classifying them further. For instance, as opposed to just indicating a tab for 'players' within the Knowledge Panel, Google has started showing a subclassification entitled "Famous Players." At times, Google gets extremely specific by even showing a tab for a specific position of a player that is highly relevant to the team in question as it does here for the NY Giants American football team:
SEO Research: What's Interesting for 2020
A busy street grows no grass and since my head grows no hair then it must be busy. Heading into 2020, there's a lot to look into and I already have a lot on my mind. Of course, you can expect us to keep up with Google's core updates. Beyond that, however, and with the emergency of BERT, Natural Language Processing (NLP) has started to really fascinate me. How good is Google at understanding sentiment within a sentence? Where does it thrive when understanding language like a human might? Where are its hangups? As Google tries harder and harder to mimic human understanding of language the answers to these questions are going to become increasingly important for SEOs to know. I think we're just starting to incorporate Google's linguistic tendencies into the SEO conversation which makes the area ripe for new research and insights!
Here's to SEO research in 2020!