Google’s Helpful Content Update (HCU) has dominated SEO and content strategy discussions since 2023. Yet despite the attention, both Google and industry experts offer advice that remains vague and difficult to translate into clear strategies that protect your site.
Google tells us to write “helpful content,” but never defines what that actually means.
In my work, researching site data, I have observed two behaviors. First, HC appears to measure “unhelpful” content on sites in multiple ways. Second, the data shows that content earns unhelpful tags based on how topically aligned it is with other pages across a site.
Content writers market their services to write within HC guidelines. But that’s not possible because writers focus on a single piece of content while HC evaluates every page as part of a larger topical group. Ultimately, you are the guardian of your domain, not the writer.
What’s new about this particular update?
Since August 2021, I have been running daily indexation tests. I did not set out to study HC updates in 2021, but I found that something shifted sharply in December 2022: Google stopped indexing and serving content it previously would have indexed and served.
On December 5, 2022, a standalone HC update rolled out, and the Indexing API that controlled bot behavior altered its bot pattern. This coincided with a steep drop in content indexation rates.
So I began to test various factors. When I topically themed test site content, indexation rates dramatically increased from 63% to 94% for 1st pass and 68% to 88% for JavaScript. [Source]
Google ranks and serves pages as collections (domain)
Data strongly implies that HC isn’t like other updates, which evaluate individual pages in isolation. While reminiscent of the 2011 Panda Update, due to its sitewide impact, the HC update goes far beyond targeting thin spammy content.
Evidence from my test results and observations indicate that Google now evaluates every page as part of a topical group within the domain.
This evaluation compares the topic of new content against all other pages on the site. When HC finds pages with topical distances that are too far or too close to other existing pages, it assigns an “unhelpful” tag that eventually suppresses the visibility of those pages.
Documentation supports this understanding of HC. The April 2024 API documentation leak revealed that Google refers to a “siteFocusScore,” a “siteRadius,” and “siteEmbeddings” pages: “In plain speech, Google creates a topical identity for your website, and measures every page against that identity.” [Source]
Can a site be served in search results without centralized “siteFocusScore”?
Yes. News sites can be and are served. News sites have no singular topical focus, not even under a general news topic.
How could Google measure the topical radius of a domain composed of news articles covering innumerable world news topics?
Given the lack of evident HC impact on news sites, it appears that these sites contain more “helpful” content. But could it be that they are exempted from it?
Did Google accidentally apply HC rules to news sites?
I believe it did because during the summer of 2023, large enterprise news sites that syndicated their original content through self-canonicalized RSS feeds to other publishers began facing indexation issues. Published pages started disappearing from search results within hours of posting. [Source]
Shortly afterward, question boxes popped up in Search Console (SC). This led many to conclude that Google couldn’t categorize or identify news sites based solely on their website data.
Google has never explicitly stated that news sites are exempt from HC. Yet, other evidence, such as a manually enforced Site Reputation Abuse (SRA) penalty, suggests that this exception from the HC classification process is necessary.
Helpful content and topical radius
Based on a topical radius measurement in the API documentation, HC doesn’t assess a site solely on the representation of a single page but rather on the specific collection. Google explained, “The signal is also weighted; sites with lots of unhelpful content may notice a stronger effect.” [Source]
With this information, I concluded that a site’s structure and content determine whether the site will be affected. The following are features that can potentially impact topical radius measurements.
Sites with multiple topics
Domains are vulnerable when they cover multiple topics that humans can connect mentally, but lack a mathematical “topical bridge” in their content that Google can recognize algorithmically.
For example, our brains can easily connect dog beds with dog health insurance or, buying a home to live in, with purchasing a house to flip as an investment.
However, the HC system cannot make these connections without assistance. The topic boundaries it expects are determined mathematically by what already exists on our sites. To bridge these topics for HC, those connections must be present within the content.
Multiple pages targeting the same query
“Old school” SEO methods that create multiple pages targeting an identical query have caused many sites to lose visibility since the September 2023 HCU.
Multiple pages on the same domain targeting the same query now lead to a decline in visibility, even if they aren’t exact duplicates.
This includes pages targeting the following sample phrases:
- Choosing the best local roof replacement contractor
- Tips for finding the best roofing contractor
- Roof replacements and how to find the right roofing contractor
- How to choose the best local roofing contractor, etc.
In such consults, these pages were indexed in Search Console data, but only one page out of the group was served in search results for the query. It is not uncommon for multiple pages on the same “topical node” to disappear and no longer be findable by their keywords.
Even if one page continues to appear in results, having enough similar “unhelpful” content leads to site-wide suppression.
Standard site feature: table of content jump links
Duplicate content (i.e., multiple URLs leading to exact word-for-word content) has never been beneficial. In the age of HC, it is proving to be existential.
Repeatedly, in my private consultations, data for sites using Table of Content (TOC) anchor links in their articles demonstrate sharp drops in impressions, clicks, and rankings.
When I read public social media posts where site owners posted their domains using SEO tools with historical data, the sites that used TOC links saw drops in visibility. These drops were present across the board from large enterprise sites (like Hardbacon) to affiliate bloggers.
All sites I’ve observed using TOC have a declared canonical meta tag. Those to which I had Search Console access revealed that anchor-text URLs suddenly gained impressions and positions, but not in overall clicks.
Just as suddenly, those jump-link URLs disappeared from the data while the site went on to flatline in visibility. Often, data demonstrated that anchor-text URLs outrank their canonical URL for multiple terms.
The only logical explanation is that HC scores every URL with a hashtag (#) as a separate page unrelated to the canonical. Thus, it creates a false and damning narrative of duplicate content on the domain.
Only Google can fix this Achilles’ heel in HC.
When measuring topical radius, judging each # URL separately creates a double whammy of multiple pages answering the same query and replicating each other word for word.
At the latest Google Creator Summit in Washington, DC, Google addressed table of content. According to one participant, in a private setting (no cameras allowed), Google shared that “some Table of Content widgets can confuse their crawlers and even reduce clarity scores.” [Source]
Why is HC enforced algorithmically, and site reputation abuse enforced manually?
There is a reason why Google must enforce Site Reputation Abuse (SRA) as a manual penalty, and manual enforcement supports the contention that news sites are exempt from HC, since many news sites appear to have received penalties.
If HC were used algorithmically on news sites, due to the topical nature of HC’s assessing a collection of pages on the domain, HC could single-handedly decimate the online news industry within 48 hours.
To rein in enterprise news sites, which may have surmised they were granted some form of algorithmic protection and began to move their full power and authority into affiliate offerings, Google was forced to manually enforce the spirit of HC.
By exempting a site type from HC constraints, Google created a situation they had to control, and yet not destroy enterprise news organizations to accomplish their goal.
At the May 2025 private Google Creator Summit, “Google admitted upfront that they know their system has favored larger sites. They didn’t mean to, but that’s how the algorithm evolved.” [Source]
I believe this is exactly how it evolved: Google operates with two sets of rules, depending on whether they consider your site a news site or not.
Did the HCU update hit your site?
What follows is an extreme example of a site repeatedly slammed with HC updates. This site was devastated by the September 2023 HC update and continues to be overwhelmed to this day.
The Search Console data below, gathered over the last 18 months, illustrates clicks, impressions, CTR, and average position. Every page contained TOC links; some had as many as 17 chapters.
Date range: 01/02/2024–05/01/2025
For blogger and affiliate sites that are TOC-dependent on most of their pages, dramatic drops in visibility (impressions) appear to result from their anchor URLs being scored as separate pages, creating a false assessment of duplicate content. These sites may inadvertently cover more topical ground that would not have fared well with HC.
Aggregate views of Search Console data clearly reflect the damage of the helpful content system and, at the same time, also obscure the ability of SEOs to recognize its cause.
Microview: Maximum validation
Below is an example of a single canonical URL compared to a single anchor-text URL of a TOC URL ranking for the same search phrase on the same day.
Given that both URLs ranked in the same position on the same day for the same search phrase, we can reasonably infer that the resulting ranking for an identical position for a similar keyword phrase indicates an exact score match by Google’s ranking and scoring system.
These pages are identical; they are just different URLs with matching content. The following screenshot displays in numbers what duplicate content on the same domain looks like.
The same data reveals that, shortly after this date, impressions and the position of the anchor#1 URL are no longer reflected in the data. Once it disappeared, the canonical URL dropped in impressions and position.
In the past 2 years, I reviewed close to fifty sites without access to Search Console data. In each case, they showed a bloom of visibility (impressions) before a significant reduction in the number of ranking queries that dropped the positions of their canonical pages during and after the updates.
The five steps to recover from an HCU update
Unlike previous updates where we might see evidence of some degree of reversal afterward, HC stands out as one factor that shows no mercy.
In short, there are no easy fixes. Reversing a damaged site from an HC update is not straightforward and requires significant effort, depending on the site size and the time Google needs to reassess the site.
Expect a faster recovery for sites with fewer issues. Some SEOs have anecdotally reported site recovery once the TOC links were removed; however, these are the exception rather than the rule.
In my forensic consultations, I witnessed that most older sites often have one or more unrelated HC issues affecting their decline, such as overusing a keyword phrase or low-level optimization compared to their competitors. These must be identified and remedied.
Additionally, sites with multiple issues may have lost competitive ground to their rivals, or Google may have altered its blended search results and may no longer be suitable for showing that type of content in their results.
Each case is different, and one size does not fit all.
Where to start?
1. Identify the problem pages at the page level
There is a bias in our industry that a larger view containing all data is the best view. Search Console (SC) averaged data has not led me to find what I was looking for, but filtered and refined data has shown me something every time. Thankfully, SC provides a filter for performance data.
If this is your site, you’ll know which pages impact your bottom line, and that’s where you want to start.
You’ll need to dig into the data if you don’t know which pages were hit.
I identify problem pages by unchecking all performance metrics, except impressions, and setting the date range to 16 months to determine the impression bloom and sharp drop dates.
With that data, I filter the results further to see the data for all pages with a “#” in their URLs. This gives me the list of pages I need to review.
Next, I compare a canonical against all its anchor text links to assess the damage. Then I work through the pages one by one. At this point, the safest option is to remove the TOC links.
You may have pages that dropped, but don’t use TOC links. In a spreadsheet, notate a tab for the TOC pages for later review and place the remaining pages in a separate tab for other Issues, taking note if any are off-topic, and then create a separate tab for those URLs.
2. Group overlapping topical node pages
Problem pages are those that answer the same query. Often you can identify them by reviewing URL slugs within sitemaps or by crawling the data of the entire site.
Diagnostically, this process moves us from individual page assessment to assessing query groups.
Often, pages in this group are “Page is Indexed” when examined with the SC Inspect URL. Don’t mistake that for a page being served in the search results. [Source]
Indexed pages not being served in their performance data do not have any impressions or positions. The way to find them in search is to use “quotes” around the URLs. In this case, you get this response.
For topically overlapping pages that show as indexed in GSC, you’ll typically see that Google has chosen to serve only one of them. The rest are not in the findable index, but they still impact performance.
The page Google chose to serve begins a downward path as if losing steam. If you have enough page groups like this, while it appears as a site suppression in full data, the culmination of their weight on the page level provokes suppression.
3.Prioritize revisions
Pages in the Other Issue column have a solution, and the choice about TOC use must be determined. Review the Other Issue list first and determine if any high-performance pages have dropped into that list.
4. Consolidate where you can, and remove what you need to
The easiest pages to solve are those that are way off-topic. Get rid of them!
Another fix is to consolidate portions of the pages whose content isn’t being served in the results into the page that is being served. Then, decide how to handle redundant pages. In my experience, you can delete a page if it isn’t being served and has no performance metrics.
You’ll have to decide if you want to 301 any of these cases. Regardless, ensure your sitemap contains only pages with a 200 response status.
5. Recrawl updated content
My indexation testing confirms that once a site experiences a “ding,” revised content needs a push. Do not wait for Google to “discover” revised content. Instead, resubmit the sitemap(s) and manually request indexation of the home page and any high-value revised pages.
Strategy to future-proof
Some SEOs maintain that the best cure is to move everything to a new domain, but that delays the inevitable. The HC classifier will do to it what it did to the original. It’s only a matter of time, and you’ll be back where you started.
For established brands, moving to a new domain is not an option.
Adopt a topical strategy
Move from a keyword strategy to a topical strategy. If you’ve been doing your own SEO for 10+ years or learned SEO from those trained on this strategy, rethinking this foundational piece will feel like you’re going against the grain.
Changing strategy, however, is the only way to get back into the game now.
Diversify more than just traffic
There are many more places to share insights other than your website. SEO agencies publishing more pages monthly on client sites need to move away from or modify that service.
Be proactive and vigilant
Get accustomed to monitoring your main content in the Search Console on a filtered page level. Don’t rely on analytics and ranking reports to replace diligence.
In Search Console a page may be coming in and out of the data, but you have no idea it’s struggling because your rank tracker caught it at the time it was in SERPS.
Concluding thoughts
There is no such thing as helpful content. There is only unhelpful content. Topical content is now measured beyond the boundaries of an individual piece of content. Whatever is not unhelpful (topically) by default is deemed helpful (topical) content.
My research data shows that unhelpful content in multiple forms contributes to a site’s decline. Helpful content is, at its heart, a topical domain factor. Increasing your understanding of your domain as a collection of pages and a topically cohesive whole, will aid in recovery and minimize future vulnerability.