Independent review sites hit in 2023 and 2024: have you seen real recoveries in the 2026 core updates, and what changed on those that came back?

1 reply

• 

  adminAdmin

  Original poster 24 hours ago

  The short answer is yes, some independent review sites hit by the September 2023 helpful content update have come back. However, the best-documented rebounds landed in the June 2025 core update, not in the March 2026 or May 2026 core updates, and most of them were partial. I have not found a credible public dataset showing a broad 2026 recovery for that cohort. The 2026 analyses that do exist point the other way for sites that write about products rather than make or sell them.

  The bigger qualification is about the second half of your question. When dozens of sites recover on the same day, the most likely cause is a change on Google’s side, not a fix they all happened to ship. That makes “what changed on the sites that came back” a much noisier question than it looks, and it is where most recovery playbooks go wrong. The useful decision is not “what should I copy?” but “is my site recovering for reasons I control, and is the domain worth more investment?”

  What the public record shows, update by update

  Here is the evidence I would weigh, roughly in date order. The evidence type matters: a Google statement tells you what is possible, a practitioner dataset tells you what happened to a group, and a single case study tells you what one team did while it happened.

  When	What was reported	Evidence type	What it means for a hit review site
  Sept 2023	Helpful content update hits many independent review sites. HouseFresh reported Google traffic falling from about 4,000 visitors a day to about 200.	Case study	The damage behaved like a sitewide assessment, not a page-by-page one.
  Mar 2024	The helpful content system is folded into Google’s core ranking systems.	Google documentation	There is no separate “HCU” left to lift. Recovery happens, if at all, through core systems.
  May 2024	John Mueller: “sites can grow again after being affected by the ‘HCU'”, adding that it “can take a lot of work, time, and perhaps update cycles”.	Google representative	Possible, not quick, and not a technical fix.
  Aug 2024	Google says it aims to surface “small or independent sites that are creating useful, original content”.	Google representative	Stated intent, not evidence of outcome.
  Jun to Jul 2025	Glenn Gabe, tracking 400 sites hit hard by the HCU, saw a number surge from 8 and 9 July, mid-rollout. “Many aren’t back to where they were before September 2023”, some were full or near-full recoveries, and some “dropped even more”.	Practitioner dataset	The strongest evidence of a cohort rebound. The shared start date points to a Google-side adjustment.
  Oct 2025	HouseFresh reports visibility above pre-update levels. Its editor’s summary: “Hard work and a lot of luck”.	Case study	One site, many simultaneous changes, and an honest admission that attribution is unclear.
  Mar 2026	Amsive’s analysis of SISTRIX data across 2,000+ domains found the winners list was “almost entirely ‘the company that owns the thing'”. Independent publishers were not analysed as a recovery group.	Large visibility dataset	A headwind for sites whose role is writing about products.
  May 2026	Aleyda Solis found intent, market fit and source type drove the biggest shifts. Transactional marketplaces gained and reference aggregators fell. Gabe noted sites hit in January 2026 for commodity content had not recovered, and some fell further.	Practitioner datasets	Being a summarising layer on top of someone else’s information is the risky position.

  Read together, the pattern is: a real but partial rebound for part of the cohort in mid-2025, then 2026 updates that rewarded first-party sources and destinations where users complete a task. I would not describe 2026 as a recovery year for independent reviewers based on what is public. If you have seen HCU-hit review sites climb in March or May 2026, that is worth sharing with dates and a peer comparison, because the published analyses have not isolated that group.

  Why “what did the winners change” is the wrong first question

  Think of a harbour at low tide. When the tide comes in, every boat rises, including the ones whose owners did nothing. If you only interview owners of floating boats about the repairs they made, you will get a long list of repairs and no idea which one mattered, or whether any did.

  The technical version: core systems apply site-level assessments, and Google’s own documentation says it “could take several months for our systems to learn and confirm that the site as a whole” has improved. When Google re-weights one component of those systems, every site sharing a similar profile moves together. Gabe’s observation that the 2025 surges all started on 8 and 9 July fits that explanation much better than hundreds of sites independently fixing the same thing in the same week.

  There are three distinct explanations for any rebound, and they call for different decisions:

  1. Google re-weighted something. A cohort moves on the same day, whether or not those sites changed. Your changes may be irrelevant to the lift, and the lift can reverse.
  2. Your changes were recognised. You made sitewide changes months earlier, the next core update moved your site, and comparable sites did not move. This is the only pattern that tells you what to scale.
  3. The results page changed around you. Competitors were removed or demoted. HouseFresh credited site reputation abuse enforcement with levelling the product review space, and Google’s August 2026 policy update changed how abusive sections are handled (in the EEA they may now be separated so they rank independently, rather than receiving manual actions).

  Survivorship bias sits on top of all three. Sites that made the same changes and stayed down rarely write case studies, so the published “what we changed” lists are drawn only from the boats that floated.

  Important distinction:

  A recovery is a traffic pattern. A fix is a cause. You can have either one without the other, and most public recovery stories only prove the first.

  Changes that keep showing up in credible recoveries, and how much weight they carry

  With that caveat in place, some changes appear consistently in Google’s guidance and in the few documented cases. I would treat them as necessary conditions to rule out, not as a recipe.

  Change	Why it plausibly matters	Evidence strength
  Visible proof of first-hand testing: how many products were tested, the results, the method, original photos	Google’s “How” guidance says readers trust reviews more when they understand “the number of products that were tested, what the test results were, and how the tests were conducted”. The reviews guidance asks for quantitative measurements and evidence of your own experience.	Documented guidance. Not proof that it drives recovery.
  Brand demand and presence off Google (YouTube, newsletter, community)	HouseFresh grew a YouTube channel, including a Linus Tech Tips collaboration. Branded demand is hard to fake, and it keeps the business alive while you wait.	One case study. Could be correlation.
  Design, UX and content changes rather than technical fixes	HouseFresh: “neither of us are technical people so most changes were design, UX and content-based.” Mueller said recovery needs more than technical tweaks.	Case study plus Google representative.
  Rebuilding or removing commodity pages, such as spec rewrites and untested “best X” roundups	Core systems assess the site as a whole, so a large weak section can weigh on strong pages. Gabe’s 2026 notes show commodity content sites still falling.	My reasoning plus observation. The weakest link in this table.
  Becoming a primary source rather than a derivative layer	Solis frames it as: “are you where users complete the task, or a layer summarising it?” Original test data is primary information. Paraphrased specs and marketplace ratings are a summary of someone else’s work.	Interpretation of a 2026 dataset. Worth taking seriously.

  Watch for this:

  Mass-deleting content because a recovered site pruned heavily can remove pages that earn links, brand searches or affiliate revenue. Prune on evidence (no clicks, no links, no unique test data, no revenue), not to hit a target page count.

  How I would tell whether my own site is actually recovering

  1. Rebuild a baseline from before September 2023. Search Console keeps 16 months of performance data, so unless you exported earlier data, you will need your own archives, analytics, or a third-party visibility index to compare against the pre-hit period.
  2. Split brand from non-brand. Use the branded queries filter. If growth is mostly people searching your name after finding you on YouTube, your audience recovered, not your rankings for “best air purifier for allergies”. Both are good news, but they lead to different decisions.
  3. Segment by page type. Separate hands-on single reviews, roundups, comparisons and informational guides. If single reviews move and roundups stay flat, you know where the site-level problem still bites.
  4. Annotate every core update and your own major changes with custom annotations, then check whether any lift started on a specific day mid-rollout.
  5. Build a peer cohort. Track 10 to 30 comparable independent review sites in the same vertical in a visibility tool. Same-day movement across the cohort points to Google. Movement only on your site, after your changes, supports your changes as the cause.
  6. Read clicks, impressions and position separately. When Google stopped supporting &num=100 in September 2025, many sites saw desktop impressions fall and average position improve, while clicks were affected far less, according to Brodie Clark’s analysis. Do not read that period as a new hit or a recovery. Rankings coming back also does not guarantee clicks coming back where AI Overviews answer the query, and Google’s Search generative AI performance reports show impressions only. As of 31 August 2026 they are available to all eligible sites worldwide.

  Pattern you see	Likely explanation	What I would do
  Non-brand lift on a mid-rollout date, peers moved too, you changed little	Google-side re-weighting	Bank it, but do not credit your changes. Keep improving, because it can reverse.
  Lift only on your site, months after sitewide changes	Your changes were likely recognised	Document exactly what changed and extend it to the sections that have not moved.
  Branded clicks up, non-brand flat	Audience growth, not ranking recovery	Keep building off-Google channels. Do not plan revenue around roundups returning.
  Peers up, you flat or down	A site-specific problem remains	Audit the page types that did not move, starting with commodity sections.
  Positions up, clicks flat	SERP layout or AI answers absorbing demand	Check which query types still send clicks and shift effort there.

  What to test if you are still down

  A normal SEO split test struggles here, because the problem is a site-level assessment. You cannot put half the site in a control group when the signal is about the whole domain. The honest alternative is a staged, documented change measured against a peer cohort across at least two core updates. That gives you strong correlation, not proof.

  Hypothetical example:

  A kitchen appliance review site has 600 URLs: 80 hands-on reviews with original photos and measurements, 220 “best X” roundups built largely from spec sheets, and 300 deal and news posts. It lost most of its non-brand traffic in September 2023 and has been flat since. The test: consolidate the weakest roundups into a smaller set built only from products the team has actually tested, publish the testing method on every review, and leave the deal posts untouched as an internal comparison. The primary metric is non-brand clicks to the review and roundup sections, compared with the untouched section and a 15-site peer cohort. If the tested sections gain relative to both across two core updates, extend the approach. If nothing moves, the constraint is probably source type rather than page quality.

  Two results would change my advice. If the whole peer cohort rises and falls together regardless of what anyone ships, spend less time on Google-facing edits and more on channels you control. If your rebuilt sections move while comparable sites stay flat, you have found your own evidence, which is worth more than any recovery thread.

  The practical position

  If this were my site, I would stop reading recovery stories as recipes. I would track two things: non-brand clicks to hands-on review pages against a peer cohort across the next two core updates, and the share of revenue that does not depend on Google. If the first stays flat while the second grows, the domain is a brand with Google as a side channel, and that is a legitimate business. If the first moves only on the pages with genuine testing and original data, that tells you exactly what to scale and what to stop publishing.

  Need help with this on your own site?

  Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

  See SEO services