What does Google consider “quality content"? And how do you capitalize on a seemingly subjective characteristic to improve your standing in search?
We’ve been trying to figure this out since the Hummingbird algorithm was dropped in our laps in 2013, prioritizing “context” over “keyword usage/frequency.” This meant that Google’s algorithm intended to understand the meaning behind the words on the page, rather than the page’s keywords and metadata alone.
This new sea change meant the algorithm was going to read in between the lines in order to deliver content that matched the true intent of someone searching for a keyword.
Write longer content? Not so fast!
Watching us SEOs respond to Google updates is hilarious. We’re like a floor full of day traders getting news on the latest cryptocurrency.
One of the most prominent theories that made the rounds was that longer content was the key to organic ranking. I’m sure you’ve read plenty of articles on this. We at Brafton, a content marketing agency, latched onto that one for a while as well. We even experienced some mixed success.
However, what we didn’t realize was that when we experienced success, it was because we accidentally stumbled on the true ranking factor.
Longer content alone was not the intent behind Hummingbird.
Content depth
Let’s take a hypothetical scenario.
If you were to search the keyword “search optimization techniques,” you would see a SERP that looks similar to the following:
Nothing too surprising about these results.
However, if you were to go through each of these 10 results and take note of the major topics they discussed, theoretically you would have a list of all the topics being discussed by all of the top ranking sites.
Example:
Position 1 topics discussed: A, C, D, E, F
Position 2 topics discussed: A, B, F
Position 3 topics discussed: C, D, F
Position 4 topics discussed: A, E, F
Once you finished this exercise, you would have a comprehensive list of every topic discussed (A–F), and you would start to see patterns of priority emerge.
In the example above, note “topic F” is discussed in all four pieces of content. One would consider this a cornerstone topic that should be prioritized.
If you were then to write a piece of content that covered each of the topics discussed by every competitor on page one, and emphasized the cornerstone topics appropriately, in theory, you would have the most comprehensive piece of content on that particular topic.
By producing the most comprehensive piece of content available, you would have the highest quality result that will best satisfy the searcher’s intent. More than that, you would have essentially created the ultimate resource center for everything a person would want to know about that topic.
How to identify topics to discuss in a piece of content
At this point, we’re only theoretical. The theory makes logical sense, but does it actually work? And how do we go about scientifically gathering information on topics to discuss in a piece of content?
Finding topics to cover:
- Manually: As discussed previously, you can do it manually. This process is tedious and labor-intensive, but it can be done on a small scale.
- Using SEMrush: SEMrush features an SEO content template that will provide guidance on topic selection for a given keyword.
- Using MarketMuse: MarketMuse was originally built for the very purpose of content depth, with an algorithm that mimics Hummingbird. MM takes a largely unscientific process and makes it scientific. For the purpose of this case study, we used MarketMuse.
The process
1. Identify content worth optimizing
We went through a massive list of keywords our blog ranked for. We filtered that list down to keywords that were not ranking number one in SERPs but had strong intent. You can also do this with core landing pages.
Here’s an example: We were ranking in the third position for the keyword “financial content marketing.” While this is a low-volume keyword, we were enthusiastic to own it due to the high commercial intent it comes with.
2. Evaluate your existing piece
Take a subjective look at your piece of content that is ranking for the keyword. Does it SEEM like a comprehensive piece? Could it benefit from updated examples? Could it benefit from better/updated inline embedded media? With a cursory look at our existing content, it was clear that the examples we used were old, as was the branding.
3. Identify topics
As mentioned earlier, you can do this in a few different ways. We used MarketMuse to identify the topics we were doing a good job of covering as well as our topic gaps, topics that competitors were discussing, but we were not. The results were as follows:
Topics we did a good job of covering:
- Content marketing impact on branding
- Impact of using case studies
- Importance of infographics
- Business implications of a content marketing program
- Creating articles for your audience
Topics we did a poor job of covering:
- Marketing to millennials
- How to market to existing clients
- Crafting a content marketing strategy
- Identifying and tracking goals
4. Rewrite the piece
Considering how out-of-date our examples were, and the number of topics we had neglected to discuss, we determined a full rewrite of the piece was warranted. Our writer, Mike O’Neill, was given the topic guidance, ensuring he had a firm understanding of everything that needed to be discussed in order to create a comprehensive article.
5. Update the content
To maintain our link equity, we kept the same URL and simply updated the old content with the new. Then we updated the publish date. The new article looks like this, with updated content depth, modern branding, and inline visuals.
6. Fetch as Google
Rather than wait for Google to reindex the content, I wanted to see the results immediately (and it is indeed immediate).
7. Check your results
Open an incognito window and see your updated position.
Promising results:
We have run more than a dozen experiments and have seen positive results across the board. As demonstrated in the video, these results are usually realized within 60 seconds of reindexing the updated content.
Keyword target |
Old Ranking |
New ranking |
---|---|---|
“Financial content marketing” |
3 |
1 |
“What is a subdomain” |
16 |
6 |
“Best company newsletters” |
32 |
4 |
“Staffing marketing” |
7 |
3 |
“Content marketing agency” |
16 |
1 |
“Google local business cards” |
16 |
5 |
“Company blog” |
7 |
4 |
“SEO marketing tools” |
9 |
3 |
Of those tests, here’s another example of this process in action for the keyword, “best company newsletters.”
Before:
After
Assumptions:
From these results, we can assume that content depth and breadth of topic coverage matters — a lot. Google’s algorithm seems to have an understanding of the competitive topic landscape for a keyword. In our hypothetical example from before, it would appear the algorithm knows that topics A–F exist for a given keyword and uses that collection of topics as a benchmark for content depth across competitors.
We can also assume Google’s algorithm either a.) responds immediately to updated information, or b.) has a cached snapshot of the competitive content depth landscape for any given keyword. Either of these scenarios is very likely because of the speed at which updated content is re-ranked.
In conclusion, don’t arbitrarily write long content and call it “high quality.” Choose a keyword you want to rank for and create a comprehensive piece of content that fully supports that keyword. There is no guarantee you’ll be granted a top position — domain strength factors play a huge role in rankings — but you’ll certainly improve your odds, as we have seen.
Nice work, Jeff!
Content is King - a lot of SEOs already noticed that content that is indepth and valuable to the reader also ranks better. It also did work for my blog posts.
But here I also have to mention that good indepth content is not everything, as you also have to promote your content in order to get seen by others. (good backlinks are very important here).
completely agree, by order content and diffusion either by social networks or backlinks, are one of the important factors to take into account. I am working I have been for a few months and it shows a lot. I recommend doing it to everyone if they still do not
Regards
No argument there! Strong inbound link profiles are No.1
THANK YOU for finally writing this article. I've been telling people for months that the "content length" data was a false flag (multiple "search insight" reports claimed that it was a factor).
Humans are weird, especially when we analyze data. SEOs are lazy, especially when we analyze data. Instead of thinking about what the data means, we try and find the answer that ties to the data the prettiest. The authors of those reports saw great rankings and high character counts and immediately went to "content length is a factor!"
Of course it isn't... good grief.
It was never the length, it was what you actually wrote about. You could have a shorter article than everyone else, but if you covered "all the things" then you would get the reward.
I have a feeling your use of the words "depth" and "breadth" will stick.
Hey Jeff!
I've always thought that this was "the way" to success, but I had a hard time quantifying and explaining a process that makes sense.
Glad you enjoyed the article.
My pleasure. Keep up the good work.
Then again, the original content on that page was very long too.
Interesting read - and great that you shared real examples and real rankings. Thanks.
I'd be interested in an update in say, 3-4 weeks? I have a hypothesis that Google 'tests' what it sees as new things, with elevated rankings - then based on clicks/engagement it will then settle to a more solid position in the SERPs.
E.g. I checked just a few of your examples, and already some are slipping back slightly from their 'new' (higher) position.
Of course, it's a machine with lots of moving parts (competitors likely not standing still just for the sake of your experiment) but I think an update to this would be really interesting - please. :)
Thanks for the comment, Andrew!
Most of the keywords in that table were actually tested weeks, if not months ago. And they do fluctuate on a daily basis, but they generally stay in an elevated position.
Did you use a keyword tracking tool, or incognito window.
Will do! Could be a good case study on RankBrain, huh?
when you publish really good content Google sometimes draws an "N" ranking line: first up because of the good content, then down to re-evaluate the keywords and then up again if the content was really good.
Also suggest using Search Console to see what searches are triggering those previously existing pages, then optimize that page. Search Console queries are a great way to see how Google has determined what your content is relevant to.
Great Suggestion Daniel..
Interesting study.
I would like to see more studies like this one, as many SEO professionals claim that the longer the better.
I believe that we can never deduce a linear equation for the positioning, because other aspects such as the vision for the reader, the writing if it is of quality and not robotic, generate controversy or debates on that topic, links, etc....
It's true that a long, good and quality content will make the rate of rebound is lower, but also if the content is too long and they come out in a few seconds, it can hurt the site more for the positioning of that entry, right?
I fell like a lot of SEO professionals are not just simply claiming to post longer content but to update you content to make it more indepth - of course the bottom line of that is that your content gets longer because you add more valuable information.
Yes, a high bounce rate is not good for your website - that's why you have to hook your readers within the first few sentences/paragraph so that they are interested and keep on reading. If you feel like your content is too long, one option might be to add a table of contents so people can jump to the parts that they are really looking for.
Thank you Abel.
You're welcome, glad my answer is of help to you!
This post is really about the ranking algorithm for "freshness of content" - compare the other ranking websites and you'll see they're outdated blog posts. This is not about length of content. Just freshly updated content.
If he changed any major elements (like title page/subheadings) this would not be possible.
And it's not a "study" since the test is false. The searches in the before and after "tests" are not the same.
Thanks for the great post. It is a very interesting case study that I will share for sure.
I only have a question:
I've seen rank brain put me in page 1 right after the indexation of my blog posts, and then gradually send me back to page 2 after a week or so. I've had this situation few times already.
Have you kept the ranking you acquired over time with this technique? I wonder if it might be that Google just "try" your content to see if users respond well to it.
I would really like your comment on this.
Thanks again for your case study.
Hi Jean,
Thanks so much for reading!
I completely agree with your assessment. I think that if you create content that is sufficiently "in-depth" and does a better job of covering related topics than the competitive landscape for a given keyword, you will get a shot at a new ranking. From there, you are probably at the mercy of RankBrain.
So as you mentioned, I think Google is "trying" you out in your new position, and whether or not you stick there will depend on your CTR and dwell time. So this speaks to the importance of how your content is displayed in SERPs (strong title tag and meta description) and super engaging content on page. This is to say, a huge block of text isn't going to satisfy your readers and generate sufficient dwell time.
With our extremely short attention spans, content needs to be highly visual and easily consumable. Otherwise, you will probably be booted to a lower ranking.
Jeff
Hi Jeff,
Thanks for the quick answer. Did you kept your new improved ranking shown in this case study long after the publishing of the new version?
Thanks
Hi Jeff,
Interesting article, thanks for writing it! I really like the idea of listing the competition’s content (page 1 and 2 of SERP) and to categorize it into topics to find patterns and cornerstone topics.
Personally I would add to your method: that you first categorize your piece of content to a certain searchers intent. Then categorize the competition’s content into searchers intent as well and only analyze the content that serves the same intent as your content.
Especially for broader keywords, the searcher intent can vary. Take your example of the keyword ‘best company newsletters’ for instance. Are people searching for email templates? Or are they searching for creative email marketing examples of other companies? In this example you could maybe satisfy both intents with one article, but I do think this is not possible in all cases (i.e. transactional intent and informational intent).
Furthermore what I noticed is that in both of your examples (financial content marketing and best company newsletters) you changed the meta title to exactly match the keyword you are aiming for:
Old meta title: Content to click through: The best email newsletters of 2016 | Brafton
New meta title: Top 8 best company newsletters of 2018 | Brafton
Old meta title: 7 impressions of content marketing in the financial services…
New meta title: Financial content marketing that works: 5 ways the finance industry is..
As you probably know most SEO experts still agree that the meta title is one of the most important ranking factors there is. So, how do you know that changing the meta title did not play a crucial role in the improvement of ranking for these particular keywords?
Don’t get me wrong: I like your hypothesis and I do believe you got something here, but I would love to see a better test (only change body content, no meta data).
What you did not share, but what I am very curious to know is: did it help your organic traffic as well? It is nice to see the improvement of ranking on one keyword, but you could have theoretically lost other rankings. Did you see an increase in organic traffic for the pages you improved?
With regards,
Peijke
Thank you for your well thought-out response.
Indeed we changed some metadata, and that likely had an effect. I published a similar story in which I isolate each factor and measure results. For example: I took a measurement of performance prior to on-page optimization (metadata), optimized the metadata and took an observation after a month, then performed content depth and took a final observation:
My finding was that (no surprise) optimizing on-page factors still matters, but no NEARLY as much as the results we saw with content depth optimization.
https://www.brafton.com/blog/analytics/how-marketm...
Hi Jeff,
Thanks for sharing this! Interesting read!
What I am still curious about: did this case study (the one you posted here on Moz) help to grow your organic traffic for the pages you optimized?
Yes, increased rankings resulted in increased CTR.
Really interesting case study. Definitely gives a clearer meaning to "high quality content". Even long-form content that is written well and is of genuine interest to readers may not always be considered by Google to be so, it seems, if it fails to tackle popular subject matters that competing web pages do cover. Writers and bloggers should be conducting content analysis beforehand across competitors' articles/pages targeting the same keyword sets, to ensure that they can provide the same information as other pages and more... This is just good writing practice anyway.
Curious if your process would change if page 1 results indicate Google is displaying content serving different intents. Would you look at more than 10 results, knock out the ones serving an intent different from your own, and only then inventory and compare topics discussed? Seems logical...
Great post btw. Thanks for sharing.
I would definitely knock out those posts that seemed to serve a completely different intent.
Thanks for reading, Donna!
Awesome article Jeff, love the real life examples and success you have seen from focusing on Content Depth, very excited to try this out for ourselves and see what results come. I would also add that using the "People also search" keywords at the bottom of Google (when they make sense for your page) to your content helps as well.
Yes, you are right at that place. Content is the king but if you can't place it on right direction then all are waste. I like your content delivery method. I am completely satisfied with your words "Longer content" for staying visitor on our website but it is also compulsory to make your content's words easily readable. Bounce rate should not be high is also good point for any website in case of traffic on our website.
Agreed. You have to balance readability with depth in order to keep people on the page.
GREAT JOB, Jeff! If you want to see the hard-core data study on this, please check out the "1 million search result"rank study i oversaw for Brian Dean using MarketMuse data. Pulled over 20,000 urls, and those that were "comprehensive" - based on top 10 in serp per kw- ranked far better than those that were not.
Great article Jeff, I think one of the biggest factors is the authority of the website and its links.
I have had great success updating content to be more informative, engaging, and up to date, but I was not using the fetch as google tool. I did not realize the rankings could change that quick.
Thanks for the valuable post.
Give it a try! Immediate feedback.
Great content, Jeff! This is an actionable piece we can use today. I always wondered about updating the publish date while keeping the url. I'm glad to see you've had good results using that approach in tests/real world.
Thanks Blair!
Great article, I totally agree with him. I also think that the content update is very important and the linkbuiling. Thanks for the input :)
Nice Jeff! I'll have to add this to my list of Case Studies as well. I like how well you explained each step though too. Thanks a lot!
Really interesting case study. Completely agree! Thanks for sharing.
Jeff, kuddos on this post. One of the best tactics I've seen in years related to content and LSI (do we still call it that?). Looking forward to giving it a try.
Thanks Jason!
Really interesting post Jeff!! I´ll try to use it as soon as posible, because my content needs a refreshment if I want to keep ranking well.
Interesting and Great post.
Going to implement it soon on recent blog named 5 Reasons Why Your Ecommerce Website Is Not Ranking In Google.
Intuitionsoftech - ecommerce development company
Congratulations on the article Jeff, really opened my mind.
Glad to hear it!
Great Jeff, a very complete article with a practical case that is very good. I loved the post. Thank you very much for sharing your experiment
Regards
Thank you sir!
Very interesting article, no doubt, make a post for a corporate blog requires many techniques, and which more complicated. In your article you have helped us a little more. Sorry for my English, maybe I should write it in Spanish. A greeting!!
Thanks for the comment, and hope you are able to benefit from this!
I have yearly versions of the same story on my domain. Like Berlin-Marathon 2015 and Berlin-Marathon 2016 and so on. Following your example it would be best to combine all these into one long piece of content?
The benefit of doing this is that you would maintain your link equity!
Do not get rid of the other pages.
Hi Jeff, - Excellent post - We are about to update a lot of content on our site, so this post came at a great time. I just have two small questions:
1) Could you recommend a freemium tool which fulfils a similar function to marketmuse?
2) Did you find the ranking jumps you experienced were sustained? Did they increase even more?
thanks a lot
Hi,
Thanks so much for reading. To address your questions:
Thanks again,
Jeff
It's only the "freshness of content" that is relevant here. The other ranking pages are all blog pages so "freshness" is the only real algorithm here.
He also probably selected "display results from last day" ... since he's hiding all the content above.
There is nothing hidden about this test. Watch the video. Again, try it for yourself.
Great post!
We see similar results when updating existing posts. We use Searchmetrics Content Experience for that purpose. It's awesome! It fully integrates with their Searchmetrics Research Cloud which is very powerful already.
I'd love to hear more about your results! Thanks for reading!
Great post! I'm definitely going to experiment with this myself, thanks for sharing.
Thank you. Great way for me to go through a debug some content that I thought should have ranked higher. Moz keeps expanding my toolbox, for which, much appreciation.
Besides ranking, in depth content also plays a key role in influencing another important aspect of SEO; the bounce rate.
I will speak from my own experience. One of my blog posts is a massive 3000+ words guide which has resulted in having a bounce rate of about 40% and average session of 3 minutes.
This does make a huge amount of sense! Great post, I think I need to work through this a few times. But it does make sense. I'm always saying to my colleagues: Google is up to relevancy. Not a huge amount of similar words. Not the most pictures and views, at least not mainly. They are catching the relevancy-train. So, going for all these different "topics" for each keyword depending on what the top ten do is totally awesome! Thanx for this advice / whitepaper
A very enjoyable post and I will definately be having a play with some of my older posts. Just a side query, how much does the freshness/update timestamp figure into these rankings?
Great study.
It also suggest that Google try to move from only link related signal. Links are important to determine authority but depth is also important. If we think in Google's way then it is understandable, Google want to provide answer that satisfiy all need of searcher. Only depending on link may not give best result.
And I also experience same with my site, when I update my ranking suddenly increase and I rank for numbers of other keywords.
This is great research, Jeff, and thanks for showing the community your technique. Just curious if you can go into a bit more detail on topic selection - especially the differences between using SEMRush, MarketMuse, and manually identifying topics. While I don't yet have a MarketMuse account, the SEMRush report you brought up doesn't seem to be all that useful at first glance for this purpose.
I ran their Content Template report for the "financial content marketing" keyword you're studying, but the topic gaps you talked about were largely absent from my results (so SEMRush wouldn't have been helpful to identify those for the copywriters to target). What I mean is, out of your four identified topic gaps, for me SEMRush only surfaced something connected to "crafting a content marketing strategy". There was nothing of note related to tracking/goals, marketing to existing clients, or marketing to millennials. So in this case I think I would've had to go manual or use MarketMuse to get useful/relevant topics ...
Or maybe I'm just not looking in the right place here? This seems to be the only part of the SEMRush Content Template report that deals with topics: "Enrich your text with the following semantically related words: financial institutions, digital marketing, piece of content, good content, banks and credit unions, financial market, case studies, blog posts, insurance companies, financial services companies, technology and innovation, creating content, build trust, financial advisors, engagement consumer, content marketing strategy, content types, financial services industry, social media, online financial"
Thank you for taking the time to include the video. It was very helpful to actually see the before and after of your page.
Hello Jeff,
Thank you for a wonderful and very helpful article. Yes, content marketing is the ultimate solution to getting those targeted high quality organic traffic that really converts and the points which you have provided above are very helpful.
Thank you.
Hi Jeff Baker,
Thank you so much for this case study. Frankly, I was not clear about how to write content better then my competitor. But now after reading your full case study it's clear to me. I have got really some special tips from your post. I will definitely try to follow your study and wait for the result. :)
Cheers
Jemee
Yup totally agree that word count alone does not qualify as great content. It's about having related keywords and context that actually answer a user's search intent. Loved the process of updating existing content and rewriting it for enhanced relevance. Sometimes its the smaller details that make all the difference.
Another way to identify topics to write about is through a content gap. Analysing what worked for the top ranking content pieces in your chosen topic, either by highest number of referring domains (ahrefs) or social signals (buzzsumo). These 2 factors have so far generated most engagement results for me.
Thanks for sharing!
Interesting read Jeff, I'm certainly going to revisit some older page 2 posts and pages to see what I can do with them. I've always been nervous of knocking them down rather than up but using semrush and other sites to research the content is a great idea, one that i've not used before.
How did your long term serp fair? did it end up on a higher average than before. I've experienced a positive high then drop on changing before (not as quick as that though it has to be said)
Changing the date is a good idea too!
Thanks for an educating read, every day is a learning day.
All of the pages we have tested have resulted in better rankings, other than 1. In this particular instance, it was totally my fault because I didn't properly brief the writer on how to write the content to address the topic gaps.
Thanks for reading.
Jeff
And what about your before/after test?
Great read! Thank you for posting. We have been trying to further improve our website's ranking but the information here has put everything into perspective. Thanks
Very interesting test Jeff. Your approach was well-thought-out and systematic - and the video was a very nice touch! Regarding the sudden jump in ranking position, I couldn't help but think of Google's QDF (query deserves freshness) ranking algorithm.
Publish date aside, I believe Googlebot is able to instantly recognize that a recently crawled page is different from the version it has cached. Because Google hasn't had time to evaluate user engagement signals like SERP bounce or time to long click (potential byproducts of making a page more comprehensive), to me it feels like some type of "freshness" signal is at play here.
Hi Brian,
I would agree. I have to imagine that freshness is definitely a factor, especially for the blogs we have tested. From what I have seen, the blogs get a sudden jump in rankings, then they will be left to the fate of RankBrain.
Landing pages, on the other hand, also have benefitted from substantial increases in rankings, but over a longer period of time. For example, "SEO content writing services" and "Content marketing agency." They saw jumps, but definitely not over 60 seconds.
This is one of more useful experiments I've seen in a while. I just have to address two things:
First is your assumption about 'has a cached snapshot of the competitive content depth landscape for any given keyword'. I am not quite sure that I understand what you mean by 'content depth landscape', but whatever it may be, given my knowledge of IR I doubt they create landscapes based on phrases. It is more likely that they order results by diversity of information they contain, and these algorithms are phrase agnostic, so to speak. The 'central' query vector does need to align with the content vector though.
Second is your topics listed and keyword targeted.. not sure I could draw lines between all of them, i.e. some keywords don't seem to correlate to topics listed. Also, as we all know, it is quite difficult to have controlled experiments in SEO, and here it appears to be happening, you got some rank changes, but some of them may have happened due to external factors, i.e. some new links appearing with related atexts.
On the other hand, would be interesting to see rankings for some other significant concepts from top Google pages, like 'financial marketing agency'.
Would be also fun to see similar experiment on some more obscure topic and on some low authority blog.
To address your points:
Agree on your point of trying to rank on low authority blogs. It's a bit easier when you have a website with a decent Domain Authority.
Good article Jeff! I have noticed position hikes only by updating the articles. But to reach the top positions also depends on other factors, such as the authority of the website, links, etc.
Updating the articles helps, but if the competition is tough it's just one of the things you have to do to overcome it.
No argument there!
Its really interesting and informative article.. Thanks Jeff for shared your experience with us..
Its Really Helpful post... Thanks again buddy..
Regards:
Few issues with this post:
So just update the date of your blog post, put up 1-2 pieces of fresh content (an image, or links at the very top) and that's it guys.
The only thing you're right about is completeness of content. That's always been a ranking factor.
Thanks for the comment. Let me provide a few points of clarification:
Totally cool if you don’t believe this test, but maybe, give it an honest try for yourself, then come back to me.
I would be interested to see the exact same test for updating publish dates ONLY for blog posts AND landing pages.
Jeff