Fixing Ranking Losses from Google Reviews Updates

How the reviews system works now

The system began in 2021 as the product reviews update, aimed at shopping reviews. Google later widened it to cover reviews of almost anything people research before choosing: services, businesses, destinations, games, films and more. After several separately announced reviews updates, Google said in 2023 that it would stop announcing each one because the system now improves on a regular basis. In practice that means review pages can move at any time, not only on announced dates, and a loss may line up with a core update rather than a named reviews update.

Google describes the system as working mainly page by page. However, it has also said that on sites where a large share of content is reviews, the whole site may be assessed. That matters for affiliate sites and comparison sites: a mass of weak reviews can hold back the strong ones.

Two things fall outside this system and are easy to confuse with it:

• Customer reviews left by buyers on your product pages or Google Business Profile are not what the reviews system assesses. It looks at editorial review content, written to help someone decide.
• Review stars in results come from structured data and have their own rules. Losing stars is a markup issue, covered in the review snippets guide, not a sign of a reviews update.

Confirming the reviews system hit you

A reviews loss has a recognisable shape: review, comparison and best-of pages fall together, while information pages, service pages and the homepage on the same site hold steady. To check:

1. In Search Console, open Performance and add a Page filter using Custom (regex). A pattern such as /(reviews?|best|vs|compare)[-/] catches most review URLs; adjust it to your own URL structure.
2. Compare equal periods before and after the drop, and note the change in clicks and average position for the filtered group.
3. Switch the filter to Doesn't match regex with the same pattern and compare the rest of the site.
4. If the review group fell sharply and the rest barely moved, the reviews system is the likely cause. If both fell by similar amounts, treat it as a broader core update issue.

The Algorithm Impact monitor helps with the timing side: it lines up your ranking movements with tracked updates so you can see whether the fall coincided with an announced core update or happened in a quieter period, which is common for review content now that reviews improvements are no longer announced one by one.

What Google says a useful review contains

Google publishes guidance for review writers. Paraphrased, it asks reviews to:

• evaluate the thing from the point of view of someone who has used it, and show real knowledge of it;
• provide evidence of that use, such as original photos, video, audio or measurements;
• share quantitative measurements where they help, such as battery life, noise levels or timings;
• explain what sets it apart from competitors, and cover comparable options;
• discuss benefits and drawbacks, based on the writer's own findings;
• describe how a product has changed from earlier versions where relevant;
• identify the factors that matter when deciding, and how the product performs on them;
• for ranked lists, include enough useful detail on each entry to stand on its own.

Here is how that looks in practice for a fictional review site, Kettle and Crumb, reviewing a coffee grinder:

The fix, step by step

A testing methodology page you can adapt

A methodology page is one of the most useful additions a review site can make, because it answers "how do you know?" once for every review. Keep it factual and specific to what you actually do. A structure that works:

1. How we choose what to review. Reader requests, best sellers in a category, new releases.
2. How we get products. Bought at retail price, loaned by the maker, or supplied free, and whether makers see reviews before publication (they should not have approval).
3. How we test. The tests you run for each category, the equipment you use to measure, and the minimum testing period.
4. Who tests. The reviewers, with links to their author pages.
5. How we make money. Affiliate commission, advertising, and a statement that commission does not decide ratings.
6. How we update reviews. When you retest, and how you mark changes.

Then add editorial review markup so search engines can read the reviewer and the pros and cons. Google supports pros and cons markup for editorial product reviews. This example is for the fictional Kettle and Crumb review:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Example Burr Grinder 200",
  "review": {
    "@type": "Review",
    "name": "Example Burr Grinder 200 review: tested for six weeks",
    "author": {
      "@type": "Person",
      "name": "Sam Example",
      "url": "https://example.co.uk/authors/sam-example"
    },
    "datePublished": "2026-09-01",
    "reviewRating": {
      "@type": "Rating",
      "ratingValue": 4,
      "bestRating": 5
    },
    "positiveNotes": {
      "@type": "ItemList",
      "itemListElement": [
        { "@type": "ListItem", "position": 1, "name": "Consistent grind at espresso settings" },
        { "@type": "ListItem", "position": 2, "name": "Low retention in the chute" }
      ]
    },
    "negativeNotes": {
      "@type": "ItemList",
      "itemListElement": [
        { "@type": "ListItem", "position": 1, "name": "Static cling at the finest settings" },
        { "@type": "ListItem", "position": 2, "name": "Louder than similar grinders" }
      ]
    }
  }
}
</script>

Best-of lists and comparison pages

Ranked lists are where review sites most often fall short. A list of ten products, each with two paragraphs rewritten from the product description and the same "check price" button, gives the reader nothing they could not get from a shop. To rebuild a list:

• Cut it to the products you have actually tested. A list of five tested items is more useful than fifteen untested ones.
• Say how each entry was chosen, and link to its full review where one exists.
• Give each entry a different "best for" reason backed by a finding, such as the quietest, the best under a price point, or the easiest to clean.
• Add a comparison table with the measurements you took, not specifications copied from the box.
• Explain what changed when you update the list, for example a product replaced because a newer model tested better.

Comparison pages ("X vs Y") should reach a conclusion. A page that lists both products' features side by side and then says "it depends" is not helping. Say which you would buy for which situation and why.

Platform notes

WordPress

Review plugins often add rating boxes and schema automatically. Check they are not adding a second, conflicting Review block alongside your SEO plugin's output, and that affiliate link plugins apply rel="sponsored" to cloaked links such as /go/ redirects.

Shopify and other shops

A shop's own blog reviewing products it sells is still review content, and readers know the shop has an interest. Be open about that, include genuine drawbacks, and compare with products you do not stock where it helps the reader.

Custom sites

Store test measurements as structured data in your content system rather than free text. That makes comparison tables consistent across reviews and lets you update a figure once when you retest.