General Background About The”Helpful Content Update”
The early announcement by Google about its self-labeled Helpful Content Update highlights a new “site-wide signal” for website rankings in their search engine. Naturally, the SEO industry is eager to discuss the details and potential impacts.
Aside from the announcement itself and additional limited details from John Mu and Danny Sullivan, there are some experienced voices in the SEO industry who are cautiously “reading the tea leaves” on the outcome of this particular update.
Known Quality Issues In Certain SERPs
With the progress of technology and the increase in the amount of money that can be created online, it was inevitable that large-scale content creation through AI efforts, or with large teams of low-cost writers, would eventually have such an impact on search results that Google would need to address them more in-depth.
Anecdotally, I’d personally tried to purchase a stick vacuum and found myself floundering in search results of nearly duplicate information that didn’t actually “help” my need to determine what machine to purchase. The SERP was full of obviously “optimized” sites that didn’t really have anything firsthand to contribute, but just re-packaged the manufacturer’s description a dozen ways.
Certainly, Google has repeatedly given the advice to “create quality content“, and many took that as “create a TON of content, quality be damned.”
“There’s a war going on for our attention in ways many of us are unaware of… Companies are using artificial intelligence and data analytics to observe our behaviors and optimize products, essentially making us addicted to them.”Kevin Payne of Leadhackrs
Reading The Writing On The Wall
There’s Hope This Update DOES Impact Poor Content
I’m excited about the potential of this update, but also a little shocked there hasn’t been more energy behind this already. With so much energy being directed toward AI, it feels like the pendulum has swung too far.Stephen Robertson of Hip Badger
It’s easy to over-index on pleasing the algorithm to get the ranking/traffic. But what the hell good is it if you’ve got near-zero conversions? Hopefully, there is a good amount of crap that gets killed out there; and sure some solid work will inevitably get caught up in this filter.
But, this is a net win for inbound. The take away isn’t really all that new if you think about it. Nail your technicals, internals, and create the level-best search intent-matched content possible. Focus on human-optimization.
What Can We Really Tell About Google From This Update?
The Helpful Content update is a clear signal from Google about how it determines placement in search results. The announcement contains a clear indication of the influence DeepRank has over there.
And here’s our friend DeepRank at work: “Google told us this is using machine learning to evaluate content that is unhelpful. In fact, this is a new machine learning algorithm that is able to identify such content. The algorithms should get better over time between the automated machine learning improvements and Google engineers tweaking and improving the overall algorithms on a regular basis.” And without human initiation.
Second to this “usefulness” is a visitor construction. The machine determines “usefulness” based on its understanding of positive human behavior.Marianne Sweeny
Appropriate Levels Of Cynicism and Caution
I disagree with a lot of the takes I’m reading regarding, what is essentially, specialization. I think people are reading too deeply into the bullet, “Does your site have a primary purpose or focus?”Data Scientist & Python SEO: Eric Heiken
I just read an article from SEER where they comment about how Forbes ranks for tire keywords over car-specific sites and how they will likely be heavily affected by this.
Just because a site is specialized to a single vertical doesn’t make it inherently more authoritative on the topic vs a large organization that has SMEs in various verticals. I believe that bullet is especially referring to sites that try to whack-a-mole random topics based on search volume as opposed to strategically aligning their content to serve a specific audience.
I guess the short version is that I think human-written, audience-targeted content on sites that have a broad focus aren’t going to be affected as heavily as big names in the SEO space are predicting. I think the impacts will stick relatively close to the AI-generated and aggregated content sites.
Wait For The Proof In The Pudding
Looks like this will be a major update, the likes of which we have not had for some time, and it is quite exciting!Technical SEO: Simon Cox
The only aspect that concerns me is the potential recovery time for sites that get hit. Soon there will be a bajillion new articles on how to recover, but ignore all those until someone has some actual experience under their belt and that may take some months!
This is likely to hit those sites that have found quick and cheap ways to gain advertising traffic en masse but not sites built to interact with their target markets. I am hoping that this will be a bit of a beach clean for the web and make it a nicer place to hang out-but suspect we will be hearing a lot of angry noise from the chip stealing gulls for a while.
The Google Useful Content Update aka Remember Panda?
I generally have concerns whenever Google chooses to pre-announce a significant update and spend more time courting widely read search practitioners to write thinly veiled puff pieces as PR water carriers for a monopoly than to invest time and resources in a product that would actually accomplish what the overall PR states. Remember, Google’s greatest engineering feat is their PR machine and subsequent lobbying efforts.
This isn’t just another opinion from a cranky, old SEO expert. Let’s consider some of the details provided in Google’s supposed plan:
1. The penalty will be sitewide, but based on specific pages. You will not be informed which pages, though the rank dampening will vary due to severity, complicating analysis.
2. Helpful is a ML based decision. Garbage in, garbage out on any learned dataset where humans might have made a couple of early errors that are magnified when applied at scale. But what will be considered helpful, ultimately?
3. It’ll take a couple of weeks to roll out in English and then be part of core ranking algorithms where updates can occur based on updated variables.
Now, what’s my problem with all of this?
1. Abusive monopolies gonna be abusive. The easiest thing Google could do here is be transparent. Simply providing a coverage not indexed reason code in GSC would solve a significant amount of headache for specific pieces of content deemed unhelpful. There’s a very good reason why they won’t do it, and it’s not because of those scary, nefarious spammers which they’ll proclaim to be why.
Google has a monopoly on the search market and is purposefully obfuscating their ranking decisioning when they decide to take action—if they were to provide concrete examples on where a webmaster fell out of favor, they’d be forced to explain why. This is where the next point comes into play.
2. Unhelpful is going to be extremely objective. As a joke, my first thought was to recall when Google’s quality rater guidelines specifically told raters to mark as undesirable websites thought to be “helpful affiliates.”
In other words, they specifically targeted well written, expert level content that had a monetization which wasn’t controlled by Google. Much of the industry complaining about the search results, as I currently am on results related to semi-public individuals, note that the primary monetization is Adsense. If Google wanted to incentive only helpful results in search, they could easily enforce quality editorial guidelines in Adsense to achieve this objective.
Alas, this is about money, not the search user. I fully expect to see heartache as smaller publishers and niche affiliates feel the pain and in their analysis of painstakingly removing and re-adding content realize that the ML determined unhelpful URLs might end up being those that convert the best, or satisfy the actual user in ways Google is choosing to ignore.
While perhaps written off as cynical initially, my views on algorithms tend to be prescient in retrospect. The vagueness of the update is purposeful, as it is designed to waste an inordinate amount of time among SEO experts as they jump through yet another series of undefined hoops, forcing their employers to re-evaluate if maybe it might just be simpler to buy more ads.Longtime SEO campaigner: Joe Sinkwitz
Implications of a New Sitewide Ranking Signal
The Helpful Content Update coming out this week is not just an update, but also a new ranking signal. It will be a weighted signal as well. What does “weighted” mean? That means that Google will be taking into account the site as a whole, not just using a page by page analysis. This means if your site has a lot of “unhelpful content” Google will score your ENTIRE site differently than if you have a little bit of poor quality less than helpful content.Kristine Schachinger
This means the impact on a site with a lot of unhelpful content will be much greater than on a site without that issue.
What else do we know? We know from Google that it is not a Core Update. We know it is only applied to English language sites right now and only in Search, not Discover or other Google surfaces. We also know it will focus on shopping, online education materials, entertainment, and tech content — though that does not mean that is the only place it will strike.
We also know that this is a machine learning based ranking signal that will take a few months to rerun, so it might feel like an update going forward, but it is not.
Aside from what Google has told us, we really don’t know much else except that if you are hit by it the effect can be pretty damaging as this is an attempt by Google to surface not just better quality content, but content with more value. This means if you are hit by it and really confused as to why? Check and see if you are writing content that adds more value to the conversation than the same 10 things every other site with this type of content said.
Google did say it could focus on AI content. First, yes AI content is detectable by of all things — Ai, but Google has made allowances for AI generated content in the Webmaster Guidelines, so the mere presence of AI is not the issue, the issue is your AI content just regurgitating what everyone else wrote or is it presenting that content with added value?
A Warning Notice To Affiliate & Ad-Supported Sites
Obviously everything’s speculation at the moment, but I can’t help but think that the timing is interesting. We’ve just started seeing content mills built on GPT-3/GPT-NEO based content start to embarrass Google.
We saw, 6-12 months ago, discussion from a few different quarters about how to build detection algorithms. Yet in April, Mueller was saying that Google couldn’t reliably detect AI generated content-though it was something the web spam team was working on.
This update sounds a lot like it’s the outcome of that work-a Panda update for the age of AI content. If I’m correct, that’s probably a good thing for most SEO experts.James Finlayson
If I had to guess, the biggest sites that are going to be impacted are affiliate and “ad-supported” sites. The specific comment about aggregating reviews without context really reminded me of those “best of product” type pages with AI content and API reviews/links from Amazon.Joshua Jarvis
I’m sure a ton of other sites will be affected too. I’m just thinking that’s the niche that gets swatted hardest. What I’m most excited to see is if location based content gets smacked for having things like “things to do” when it’s a real estate or local service company.
If you’re trying to rank for anything + city, it’s a tried-and-true tactic to throw in stuff about the city, but never really with the intent of the user or the page. It’d be really interesting if these get leveled too.
Remain Calm, Step Away From The Delete Button
If you’re already running for your CMS, about to purge a bunch of content pre-emptively to avoid a “sitewide signal” from tanking your rankings, you need to slow your roll because Google already has sampled the data it needed for the update so anything you do in haste without seeing the actual impact could be cutting off your nose needlessly. Hold on there, Cyrano, perhaps Google will be a more patient Roxanne than you expect!
Keep An Eye On “Roll-Out” Timelines
Some experts have noticed certain verbiage from the release that says Google will be “rolling out a SERIES of improvements.”
It’s good to keep in mind that when Google announces it will “roll out next week” that doesn’t necessarily mean that parts of the update couldn’t already be in deployment or testing. As noted in the tweet above, Google has been at least collecting data for weeks with this update in mind. There also could be some early deployment involved.
Felix from Ranktracker said, “We track 20 million keywords daily, and our rank tracking data shows that we have seen many sites using AI content getting massive ranking drops over the past 24-48hours (August 18th and 19th). It does not seem to be slowing either. Also, many sites that got hit from the May algorithm update are showing signs of recovery on this update.”
Of course, this data point by itself should be treated with the same level of skepticism, as these changes could be due to other factors or algorithm updates. Regardless, you’re going to want to set aside time next week to review the impact and make changes accordingly.
My Kingdom For A…Squirrel?
Some are skeptical of Google’s ability to accurately detect machine-generated content and suggest a squirrel-based defense. This is questionable advice at best, but the underlying message should be to be patient, wait for the outcome, and then execute on best practices.
Providing Some Tentative Action Items
We know quite a lot about this update so far — it’s a weighted sitewide classifier based on how much unhelpful content your domain has, it looks at user generated content, it will roll out in several stages, and it will take months to recover by showing you’re serious about cleaning up your content and providing more value.Jonas Sickler: SEO Manager, Terakeet
I’d say the best thing you can do to protect your domain going forward is to roll up your sleeves and do a complete content audit.
Even if you currently publish great content, you probably have some skeletons in your closet. What was great five years ago is probably crusty and low value now. Go back and comb through everything to make sure older content is in line with your current editorial standards.
This is directly in line with Google’s goals, so most SEOs are excited for the new signal to roll out. If you’re doing things right, it shouldn’t change your SEO strategy.
Ultimately, here’s what SEO experts should recommend for any update: if you aren’t doing anything ‘wrong’ don’t make any knee-jerk changes as these updates level off or self correct… However EVERYONE (agency, in-house, freelance consultant) should use it as a good excuse to gain buy-in. Perform a full content inventory/audit if you aren’t already doing this on a regular cadence.Jake Bohall of Hive Digital
It’s never too late to do a content audit, review content quality, and start implementing best practices, just know that it won’t necessarily impact Wednesday’s scheduled algorithm roll-out.
However, it might be a good time to leverage the excess visibility that this algorithm update “noise” will generate to get buy-in from the c-suite for those optimizations you’ve been trying to get approved!
Jeremy Rivera is the founder of SEO Arcade, a keyword forecasting tool, and a 12+ year veteran of the SEO industry, having worked with Raven Tools, Tapclicks, and hundreds of businesses from enterprise to mom-and-pop shops.