What the helpful content system judges

It is a site-level classifier, not a page-level filter. That is the defining characteristic and the reason HCU demotions behave differently from anything else in search. Google is asking whether the site as a whole exists to help people or to catch searches. A genuinely excellent article on a site that mostly produces search-shaped filler is held down alongside the filler.

Sites hit by the August 2022 and September 2023 releases share a recognisable profile. Not every item applies to every site, but three or more together is the pattern.

Signal	What it looks like	Measurement (how to check your site)
Content produced to a schedule, not a need	Publishing volume driven by a content calendar and keyword list	Count pages published against pages earning impressions
No first-hand experience	Articles written by researching the current top ten	Ask what the page contains that is not already ranking
Topical sprawl	Coverage extending well beyond the site's actual subject	Group content by topic and look for unrelated clusters
Anonymous or generic authorship	Posts under Admin, a brand name, or invented author personas	Check whether any author can be verified externally
Answers that leave the reader searching again	Long pages that never quite resolve the question	Read your top losers as a reader, not as the publisher
Large archive of zero-impression pages	Old posts nobody finds and nobody would miss	Export 16 months of Search Console and sort by impressions

The last row is the most actionable place to start. On most demoted sites I audit, 40 to 70 percent of indexed pages earn effectively no impressions. That ratio is itself a signal, and reducing it is the core of the work.

Why September 2023 was different

The September 2023 release produced the most severe and longest-lasting demotions of the decade. Many sites lost most of their organic traffic within days and stayed down for over a year. Three things made it worse than August 2022.

• It was applied harder and more broadly. Why: Sites that survived the 2022 release with minor losses were caught fully in 2023.
• The next reassessment was unusually far away. Why: The March 2024 core update was the first meaningful re-evaluation, roughly six months later, and it was also the update that merged the system into core.
• Recovery advice at the time was wrong. Why: The common recommendation was to add more and better content. Adding to a site judged for producing too much of the wrong thing usually made the ratio worse, not better.

Sites that recovered did the opposite of the prevailing advice. They cut, consolidated, narrowed their topical focus, and added visible authorship and first-hand evidence to what remained. The full method is in the content pruning strategy.

The recovery method

Establish the real scope first

• Confirm the drop date against confirmed updates. Why: A site-wide drop in September 2023 or at a later core update fits this pattern. A drop elsewhere probably does not. Use the date checker.
• Check whether every template fell together. Why: Site-level classifiers hit everything. If only one page type dropped, you are looking at a different system and a much smaller job.
• Rule out click displacement. Why: Flat impressions with falling clicks is not a demotion at all. See impressions stable, clicks falling.

Change the composition of the site

• Inventory every indexed URL against 16 months of impressions. Why: You cannot fix a ratio you have not measured. This produces the list everything else works from.
• Consolidate overlapping pages into fewer, deeper ones. Why: Merging preserves links and history that deletion destroys, and depth is what the system rewards.
• Remove what has no value in any form. Why: Pages with no impressions, no links and no purpose are pure drag. Redirect where a relevant target exists, otherwise let them go.
• Narrow the topical range. Why: A site that covers one subject thoroughly reads as a resource. One that covers everything reads as a content operation.

Make the remaining content demonstrably first-hand

• Add named authors with verifiable credentials. Why: Anonymous publishing is one of the strongest negative signals in the rater guidelines. Score your gaps with the E-E-A-T rubric.
• Add something to each surviving page that exists nowhere else. Why: Original data, your own screenshots, a named case, a tested result. This is the clearest marker of a people-first page and it was decisive in the 2026 core updates.
• Fix the production process, not just the archive. Why: Sites that pruned and then resumed publishing the same way were demoted again at a later core update.

Then wait for a core update

Since March 2024 there is no separate helpful content refresh. Reassessment happens during core updates, which have averaged 82 days apart over the last two years with a range of 7 to 147 days. Watch the live tracker for the next rollout and measure only after Google marks it complete. Timelines by update type are in how long recovery takes.

Full recovery frequently takes two core update cycles rather than one, because the first pass reduces the drag and the second registers the improved composition. Planning for one cycle and getting two is the most common source of disappointment in this work, so I plan for two from the start.

What does not work

• Publishing more content. Why: The system judged the site for producing content that did not help. More of it deepens the problem.
• Rewriting everything for readability. Why: Polish is not the missing ingredient. Original substance is, and generic rewrites often remove the specifics that were working.
• Adding author boxes without real people behind them. Why: Invented author personas with stock photos are a trust failure, not a fix, and they are increasingly easy to detect.
• Technical SEO work alone. Why: Core Web Vitals and schema do not address a content composition judgement. Worth doing, but not the cause.
• Waiting without changing anything. Why: Core updates reassess the site as it currently exists. An unchanged site gets the same verdict repeatedly.

Prefer to work through it yourself first? The step by step recovery guide covers the same method in full detail, free.

Helpful content recovery FAQ

Is the helpful content update still running?

Not as a standalone update. Google merged the helpful content system into the core ranking systems in March 2024. The judgement still operates, but it is now applied and reassessed as part of core updates.

How long does HCU recovery take?

Implementation is usually four to eight weeks because the work spans the whole site. Reassessment then waits for a core update, and full recovery often takes two core cycles, so four to eight months is realistic.

Why did my site not recover after I improved my content?

Usually because the improvement was page-level while the judgement is site-level. Improving fifty pages on a site of five hundred barely moves the ratio. The other common reason is that the reassessment has not happened yet.

Should I delete content to recover from HCU?

Consolidate first, delete second. Merging thin overlapping pages into fewer strong ones preserves links and history. Delete only pages with no impressions, no links and no purpose.

Can a site fully recover from a September 2023 HCU demotion?

Yes, and several have, but the ones that did changed what the site fundamentally is rather than editing the pages that fell. Sites unwilling to reduce output or narrow focus generally did not recover.

Is HCU a penalty?

No. It is algorithmic, with no Search Console notice and no reconsideration request. See manual action versus algorithmic penalty for the distinction.