Quality and relevance are different things, yet we often discuss them as if they were the same. SEOs have been optimizing for relevance all this time, but are now being asked to optimize for quality. This post will discuss what that means and how to do it, with a focus on what I believe to be the single most effective and scaleable tactic for improving your site's overall level of quality in most situations.
You need BOTH quality AND relevance to compete these days.
First, let's establish what we're talking about here, which is quality. You can have relevancy (the right topic and keywords) without quality, as shown here:
This MFA site is highly relevant for the phrase "Chairs for Baby." But it also sucks.
"...babies are sensitive, delicate individuals who need cautious. So choosing a right Chairs For Baby is your gentle care." WTF?
It doesn’t matter how relevant the page is. The only way to get that page to rank these days would be to buy a lot of links, but then you're dealing with the added risk. After a certain point, it’s just EASIER to build a better site. Yes, Google has won that battle in all but the most hyper-competitive niches, where seasoned experts still find the reward:risk ratio in their favor.
OK, now that we've established that quality and relevance are different things, but that you need both to rank, how do you optimize for quality?
Quality Indicators |
Tactics (How to Optimize for Quality) |
|---|---|
| Grammar, spelling, depth | Hire a copy editor. Learn to write better. |
| Expertise, Authority, Trust (EAT) | Make content deep and useful. Call out your awards, certifications, news coverage, and use trust symbols. Make it easy to contact you, and easy to find policies like terms of service, returns, and privacy. |
| PageRank (PR) from links | Build high-quality links. There are thousands of great articles out there about it. |
| Reviews | Ask your best clients/customers. |
| Short-clicks | Improve Out-of-Stock pages with in-stock alerts and related products. Be less aggressive with your pop-up strategy. Also see query refinements, pages per visit, and dwell time. |
| Query refinements | Only attract the "right" audience and deliver what you promised in the search results. Choose keywords by relevance, not just volume. Think about query intent. |
| Dwell time on site | Make your pages stickier by improving the overall design. Add images and video. Run GA reports on the stickiest traffic sources and develop a strategy to increase traffic from those sources. |
| Pages per visit | Improve internal linking, suggest related content, customize 404 pages. Split up long posts into a series. |
| Conversion rates | Do A/B testing, make your messaging very clear, follow CRO best practices. |
| Ad placements | No ads above the main content on the page and do not have an annoying amount of ad blocks or pop-ups. |
| CTR from SERPs | Craft better title tags, URLs, and descriptions. Grab attention. Gain trust. Make them want to click. Add schema markup when appropriate. |
| Page speed | Visit https://developers.google.com/speed/pagespeed/insights/. |
| Mobile experience | Responsive and/or adaptive design, less text, easy navigation, loads lighting fast, shorter forms. |
There are many ways to improve the quality of your site. Some are obvious. Others are more abstract. All of these quality indicators together make up your site's overall level of quality. For more, check out the keyword agnostic ranking factors and the engagement metrics from SimilarWeb areas in the Moz Search Engine Ranking Factors report.
We've already established that Google knows the relative quality of a page. Let's assume — because it is very likely — that Google also knows the relative quality of your entire site. And let's call that your sitewide QualityRank (QR) (h/t Ian Lurie, 2011).
What's the most effective, scalable, proven tactic for improving a website's QR?
In a word: Pruning.
Learn more about it here. Pruning requires doing a content audit first, which you can learn more about here. It's nothing groundbreaking or new, but few clients come in the door that can't substantially benefit from this process.
Sometimes pruning is as easy as applying a noindex tag to all pages that have had zero organic search traffic over the course of a year. You may be surprised how many enterprise-level sites have huge chunks of the site that fit that criteria. Other times it requires more analysis and tougher decisions. It really all depends on the site.
So let's look at some pruning case studies.
Three things to remember:
1. Significant portions of the sites were removed from Google's index.
2. Pruning was not the only thing that was done. Rarely do these things happen in a vacuum, but evidence points to pruning as a major contributor to the growth examples you're about to see.
3. You can read more about Inflow's research in our case studies section.
1800doorbell had technical issues that made it possible to identify cruft and prune big chunks of the site quickly. This contributed to a 96% increase in revenue from organic search within six months.
The dip at the end has to do with how the timeline was generated in GA. Growth was sustained.
We're not the only ones finding success with this. Go check out the Ahrefs case study for another example. Here's a compelling image from their case study:
Ahrefs saw amazing results after performing a content audit and pruning their blog.
If you weren't already convinced, I hope by now it's clear that performing a content audit to determine which pages should be improved and which should be pruned from Google's index is an effective and scalable SEO tactic. That being established, let's talk about why this might be.
We don't know exactly how Google's ranking algorithms work. But it seems likely that there is a score for a site's overall level of quality.
Does QualityRank actually exist as a calculation in Google's organic algorithm? Probably not under that name and not in this simplistic form. But it DOES seem likely that something similar would exist, especially since it exists for PPC. The problem I have with the PPC equivalent is that it includes relevance factors like keyword use in their metric for "quality."
Google needs a way to measure the overall quality level of each site in order to rank them properly. It's just probably much more mathematically complicated than what we're talking about here.
The point of discussing QualityRank as a framework for pruning is to help explain why pruning works. And to do that, we don't need to understand the complex formulas behind Google’s ranking algorithms. I doubt half of the engineers there know what's going on these days, anyway.
Let’s imagine a site divided into thirds, with each third being assigned a QualityRank (QR) score based on the average QR of the pages within that section.
The triangle below represents all indexable content on a domain with a QR of 30. That sitewide QR score of 30 comes from adding all three of these sections together and dividing by three. In the real world, this would not be so simple.
I hope the mathematicians out there will grant me some leeway for the sake of illustrating the concept.
This is the same site after removing the bottom 50 percent from the index:
Notice the instant lift from QR 30 to QR 40 just by removing all LOW QR pages. That is why I say pruning is the most effective way to raise your site’s overall quality level for better rankings, IF you have a lot of low-quality pages indexed, which most sites do.
Time to switch analogies
Pruning works because it frees up the rest of your content from being weighed down by the cruft.
"Cruft" includes everything from a 6-year-old blog post about the company holiday party to 20 different variants with their own landing pages for every product. It also includes pages that are inadvertently indexed for technical reasons, like faceted navigation URLs.
Remove the bottom half of this iceberg and the rest of it will "rise up," making more of it visible above the surface (i.e. on the first 2–3 pages of Google).
The idea of one page being weighed down by another has been around at least since the first Panda release. I'm not writing about anything new here, as evidenced by the many resources below. But I’m constantly surprised by the amount of dead weight most websites continue to carry around, and hope that this post motivates folks to finally get rid of some of it. Your choices are many: 404, 301, rel ="canonical", noindex, disallow... Some of the resources below will help you decide which solutions to use for your unique situation.
Pruning for SEO resources:
- How To Do a Content Audit
- Pruning eCommerce Sites
- What, Why, & How of Pruning for SEO Mozinar
- Long Click and the Quality of Search Success
- Dwell Time: The Most Important Metric You're Not Measuring
- Site Quality: Google's Major Ranking Signal Explained
- eCommerce Content Audit Toolkit
- Clean Your Site's Cruft Before it Causes Ranking Problems Whiteboard Friday
- Google Search Quality Rater Guidelines PDF
- Cut the Cruft - My Presentation at MozTalk Denver, 2016
So many good tips here but for me it was the way you clarified the quality and relevancy. I know that a basic content audit and pruning will help but did not think of it that much with recent clients until now. The iceburg example was what really made that stick. Also the ability to create new content to put on top of the old is not always the best when some of that older content depending on the quality can just be weighing down the good stuff.
Great read, resources, and examples thanks for the contribution to the Moz community and helping make all that visit better marketers.
Thank you Tim. This topic is not without some controversy. It helps to hear that my perspective on it was clearly communicated.
Hi Everett,
Thanks for writing this post, it's really detailed and focused on the right aspects to improve sitewide quality and their impact on rankings.
The idea is right, do the right things to improve how your website looks to search engines, and then hope for the best. As per our experience, identifying and doing the right thing is most important aspect of any website SEO, it not only help in ranking higher but also helps to reach target audience, and it's a positive impact conversion rate optimization.
Thanks again !
Regards,
Vijay
Thanks for posting. I'd like to discuss a few points here.
You explained Quality Rank (QR) in details here but not so much the "Relevance Rank" (RR). At what point does a low-quality website can outrank a high-quality one due to it's relevancy. For example, will a website with 100% QR / 40% RR rank higher than 60% QR / 80% RR?
Another important thing is understanding the value of quality indicators. Obviously backlinks are the top factor for SE, but what would you recommend focusing on? Of course it's cool to have all of the items on the list checked, but can you optimize the process and kick your QR just enough to rank higher if you have a decent RR?
Hello Igor,
I agree. Relevancy is certainly important. Like I said, we have been optimizing for relevancy since the beginning of SEO. I'd recommend focusing on wherever you have the most problems and opportunity, which will be different for each site.
Not exactly relevant, but due to the way ice floats on water, removing a chunk from the bottom would cause that iceberg to sink further. :)
What metrics do you suggest looking at to determine cruft-iness on eCommerce sites? Bounce rate & duplicate content are the ones we usually point to - has anything else stuck out for you in the past?
DOH! Thanks Alex. I once said my truck turns like a tank and someone pointed out that tanks can turn 360 degrees on a dime. So now my truck turns like a boat, a more apt analogy. Maybe I should make the "cruft" part rock or metal. ;-)
Great question. Here are some things to look at:
There are many others, but generally if the page has no external links, is substantially duplicate or thin, has poor engagement metrics, and little or no revenue or traffic from organic search it is a candidate for removal from the index UNTIL the page gets improved.
Maybe you could use a "Sandbags on a hot air balloon" analogy?
Luckily, we learn more from comments than from the Post.
We should always encourage the people who can teach us from their comments more than a lenghty Blog posts. :-)
Great post Everett
We have an events website and at any one time, due to the nature of events 'expiring' the number of old events is always growing
Would you recommend adding a no index tag to any past events as they are very seldom searched for but we'd like to leave the content live
As we include rich metadata perhaps google is intelligent enough to know these events are in the past anyway so we shouldn't risk noindexing 100000s of past eventpages?
We've found that the more past events we have, the better the upcoming events rank as they all interlink. Would no indexing the past pages affect this optimisation?
Automatically NoIndexing them after a certain period of time seems like a logical approach if you need them on the site for visitors, but not for search engines. You may also want to look into the "Unavailable After" tag, which looks something like this:
<META NAME="GOOGLEBOT" CONTENT="unavailable_after: 25-Aug-2007 15:00:00 EST">
Nice post as usual Everett.
I recently started analysing logfiles and for a lot of our sites it was quite a shock to see the frequency with which Googlebot is visiting the pages. On some of our bigger sites, 85% of the pages are crawled less than once a month. Pruning the lower quality pages would certainly increase the crawl frequency on more interesting parts of our site. Working in publishing however it's not easy to convince my colleagues to prune articles.
Apart from the content audit you mentioned, checking the log files can also reveal valuable information on "useless" or duplicate pages. On one site we discovered that old urls from previous versions of the site were still returning 200 status. These url's couldn't be found by crawling the site, but Googlebot remembered them & kept on crawling them. Part of them where "empty" pages, the others where duplicates of existing pages. We eliminated these pages (too soon however to see if there is an impact on traffic)
Dirk,
It sounds like you are digging in with all of the tools at your disposal to help those clients. They're lucky to have you working on their site. With Screaming Frog's recent announcement that they're in the log file analysis game we'll probably see even more SEOs doing the same.
Paul mentioned in his comment some other potential reasons pruning could work, other than the QualityRank explanation I gave in the post, which would align with your ideas above about how improving crawl frequency for better pages could help.
I still think there is an overall quality score for the site as a whole, based on a variety of on-page, off-page and user-generated metrics, and it would stand to reason that reducing the amount of very low quality pages on the site would benefit the remaining pages in that way.
Thanks for sharing Everett, very good post.
I really enjoyed this one. We've been doing this A LOT recently with our clients and are seeing some small improvements pretty quickly. Definitely feel as time goes on we'll reap even bigger benefits from this strategy. Cheers!
Great Post!!
Another clear example of that quality is always better than cantidas ...
As simple as removing the less important pages of the search engines of Google ...
Everett - Thanks for the contribution that you have made.
Question: Though pruning can help improve quality, will it not affect backlinks which is still one of the most influential factors in Google's page ranking algorithm?
Prakash,
Good question.
Read a little deeper into how we suggest pruning, which starts with a content audit to determine whether these pages have any external links or traffic - in which case we would either improve or redirect them most of the time.
Very interesting. I do this kind of pruning continuously, reviewing old posts, converting them to evergreen articles, deleting or redirecting the ones with less traffic to other posts that rank better, etc... and the results are quite good.
Regards,
Czd
It seems or possible focused was e-commerece sites are ideal for the purposes of Pruning. However read it and i believe this process can work with all type of business or sites. Those have much luggage which they carrying on along with their regular. By remove that it will help them to gain more visiablity for sure.
I have yet to learn a lot about this I am just keep reading on this and will continue till I get my hands on it :).
Thanks for sharing.
Nice post man. The only issue I take with this is that there's too much emphasis placed on the removal of webpages and a direct benefit derived form it, when in reality it's more nuanced.
If you're removing pages that are receiving little or no traffic, than you aren't going to see a traffic loss. You may improve indexing by removing excess pages, but I doubt that's what is causing traffic improvement.
More than likely, the effects are indirectly caused by
Paul,
Thanks for your input. Better crawlability is definitely one of the main reasons to prune your site. I'm not sure improving the existing content is the reason though. This is because we've seen this work at scale with just the pruning part, not the improving part. In other words, we removed low quality pages without improving others in some cases and noticed similar effects.
I respect your opinion (always have) but we've done this many times now and I strongly believe the main reason for the lifts is due to the removing of excess pages. I could be wrong, but that is what I think.
I can be wrong. That's okay :)
Have you tried to measure indexing changes? I'm curious about how much of a factor that is. I'm more concerned with the why than anything else.
I suppose that it could be that some sort of engagement factor is actually being considered by Google (I don't want to open a can of worms on that debate) and that poor engaging content is negatively impacting the overall site performance. It could also be that NLP is a larger part of the algorithm than I would have expected and by removing extraneous pages, there's improvements to semantic closeness.
Interesting in any case.
Do you mean measuring the amount of indexed pages? Yes, and - on sites that need to be pruned - we'll typically see rankings and traffic improve as the amount of indexed pages in GSC goes down.
I'm not sure we could isolate all of the variables to reach the level of certainty we'd all like in this industry. We just do what we find works and try to retroactively figure out what's going on in many cases, which is what I've done here.
The can of worms I want to open is the one that follows advice from Googlers telling SEOs that Rel= "Canonical" on massive sites is the way to solve the cruft problem, and that a Noindex tag is superfluous. They've actually warned AGAINST pruning, which boggles my mind considering how much money they probably spend crawling, indexing and ranking all the cruft on the open web.
Rel "Canonical" to another page doesn't remove a URL from the index, though the page to which it canonicalizes will be the one to show up. You can test this by searching with highly specific queries that surface non-canonical pages, such as these: https://t.co/iHpheuzYoW . Still indexed.
Noindex,Follow still allows the pages to be crawled, but log file analysis will probably show you that these pages are now being crawled much less frequently. More importantly, as explained above, they are no longer "indexed" by Google, which seems to have an effect in-and-of itself.
Telling Google "This URL is equivalent to this other URL" is not the same as saying "This URL should not be in your index".
I also want to make sure it is clear that I'm not talking about pruning "instead of" improving. Improving the page is the better option in most cases. This is a matter of scale and prioritization.
As for the effect being due to closer semantics, that is one of the reasons I pointed out the difference between quality and relevancy. You may need to prune highly "relevant" pages in terms of the semantic closeness to your main topics, but I suspect you'll still see the lift by removing relevant, but low-quality, URLs from the index.
Thanks again for joining the discussion!
Thanks. Great article! Re: "Telling Google "This URL is equivalent to this other URL" is not the same as saying "This URL should not be in your index"." are you saying that the "cruft" page should be noindex and NOT canonical?
Risabb,
In many cases, yes. Yet, what Google tells people is that Rel="Canonical" is enough.
I'm not 100% sure that's how icebergs work, but I dig the post and am going to try it out on my blogs as soon as I feel like I've satisfied the consistent posting schedule requirement.
The "republish when it can be improved" part is crucial for anyone who just skimmed and missed it.
Great post Everett!
Maintaining consistent quality across my client's websites has always been a priority in the creation phase, but this was a good reminder to continually go back and reevaluate/improve on old content as well. Looks like i'll be doing some content audits when I get back into the office. Thanks for the good information!
Hello! Thank you for sharing, based on my research and readings most content with the word count of 1,800 above are high ranking sites in google. I mean most of them are in the first page of google... which one is true the longer the content the better of course your content need to be relevant... or the time that your site will be viewed by the user?
Hello Everett,
Yes, Quality is evergreen and it is always has the upper hand over the quantity and it is proved many times. Most of the well performing blogs/websites does have the quality over quantity. Nice informative post, and thanks for sharing :) Have a great day!
Hi Everett,
Great post and I personally have removed a bunch of deadweight pages on sites before that have made my rankings shoot up.
Likewise, if you have a post that is on page 2, I find that sending some internal links to it can sometimes be enough to get it to page 1.
We think that Google has a real challenge on it's hands to determine the most relevant results. For example, it would rank a property portal above a real estate company when someone is searching for "real estate company." Just because the portal is a behemoth containing many real estate companies, this does not make it more relevant, since it is not actually a "real estate company." Google needs to figure this out and apply fairness and relevance to it's results. There are many many examples like this across ALL sectors.
Thank you so much Everett for sharing such a valuable blog. It's really helpful and i joined the Moz Blog today that's why i'm very impressed with this blog.
Absolutely awesome. Thanks. I'm all about semantic SEO and covering topics within a websites theme. This ties into that perfectly.
Yea, quality always matters!! Great post Everett, thanks for sharing an awesome content :)
thanks for sharing. Great post.
so you don't want faceted navigation URLs indexed? I always thought this was ok since Amazon does it. Seems like it would allow Google to give the user the most relevant result if they searched "red running shoes size 10". Would love if you had a source for this somewhere. Great post!
I think part of the trouble there is when you get into things like "red running shoes size 10" vs "size 10 running shoes red" and "shoes: running: size 10: red"
Shouldn't all of those searches bring the user to a pre-filtered list of red, size 10 running shoes? I suppose you could have a category for red, and a category for size 10, but that's not really the best user experience. Users love the faceted navigation, so why would we want to make a user get to the overall running shoe category and then filter when we could have a faceted URL that is pre-filtered for them?
DCRader, I would make those static category pages instead of facets if they get a lot of search volume. Given that many people search for shoes by color, that could potentially be one to leave indexed and optimized. Maybe facets and category pages are the same thing on some platforms. Then it just becomes a matter of scale. Having 100 canonical category landing pages indexed may be necessary for your business, but if you have 10,000 of them indexed it may be time to adjust your strategy.
Thanks for your reply. I see what you're saying. The hard part is mixing a faceted design with category landing pages. Got me thinking on this. I'm off to hunting for some examples now. Thanks.
Great post. All the quality indicators mentioned are worth considering to improve the overall quality of stuff that people like us serve to our users. Thanks for making it so elaborative and practically useful.
Absolutely lovely content. I will try to follow all the steps in orded to improve my website
please tell me the issue in my website: www.cineforest.com traffic daily decreasing
Nice post!
Hi Everett..thank you very much for share,really I like your post.
Hi Everett ,
Great!! thanks for sharing I have a question about ranking my website. which is the better google page speed for ranking my website kindly suggest me.
Amazing Share! Look forward to putting in these strategies to test.
the quality is very important, excellent post
wow, very informative and helpful