What Is Review Schema Markup?
The most misused markup type there is, because inventing a star rating is trivially easy and visibly rewarding. This page sets out what a rating actually claims, the four conditions it has to meet and why we will not implement it otherwise.
The Most Abused Markup There Is
Star ratings are visibly attractive in a result, inventing them requires no skill and nothing checks whether they are real. That combination has made this the most misused type in the vocabulary.
Why the reward is obvious. Stars stand out.
A result carrying a row of stars draws the eye among plain ones. The benefit is visible to anybody comparing results, which is why businesses want it and why suppliers offer it.
Why the abuse is easy. Nothing verifies it.
Asserting a rating requires typing a number. No system holds an independent record of who reviewed your business, so nothing can contradict a figure you supply.
Why it is detectable anyway. Reviews are visible.
A page asserting an average rating with no reviews on it is checkable by anybody who looks, including whoever reviews policy compliance. The absence of the reviews is the evidence.
What this page has to do. Explain rather than forbid.
Simply prohibiting it invites somebody to weigh the odds. The useful version explains what the markup claims, why a false claim is a serious matter, then what a business should do instead.
What this page will not contain. Any method.
Nothing here describes how ratings get fabricated. The abuse is described by what it claims and why that is false, never by how it is accomplished.
What It Actually Asserts
Marking up a rating states that real people gave real ratings for this thing, averaging that figure, in that number. It is a factual claim about the world rather than a presentational choice.
What each part claims. Three things at once.
That reviews exist. That they concern the thing on this page. That the figures shown are what those reviewers actually gave. All three have to be true for the assertion to be true.
Why the count matters as much as the score. It is also a claim.
Stating that a rating comes from a number of reviewers asserts that many people reviewed. Inflating the count while keeping the score accurate is still a false statement.
What it is not. A display setting.
People treat rating markup as a way of showing stars, in the way a stylesheet shows a colour. It is a statement of fact. The visible stars are a consequence of the statement being accepted.
Why that framing does the work. It settles the questions.
Once the markup is understood as a factual claim rather than a formatting option, most of the difficult cases answer themselves. Ratings from elsewhere, testimonials you selected and pages with no reviews all fail for the same reason.
Where the general principle sits. The concept page.
All markup is assertion, which what is schema markup establishes. This type is where the principle has the sharpest consequences.
The Rule
Real reviews, from real customers, about the thing on that page, collected by the business. Four conditions. Failing any one of them makes the markup false.
Real reviews. Written by somebody.
Actual feedback that exists, rather than a figure representing what you believe customers think. This rules out inventing a rating entirely, which is the version everybody recognises as wrong.
From real customers. People who bought.
Feedback from somebody who used the service or product. Not staff, not friends, not the business itself. Block four covers why this catches more cases than people expect.
About the thing on that page. Correctly attached.
Reviews of the business belong on the business, reviews of a product on that product. Applying an overall score to every item is a false claim even where every review is genuine.
Collected by the business. Yours to state.
Gathered through your own process and shown on your own page. Block five covers ratings held by somebody else, which is the condition people breach without meaning to.
Why all four rather than a general principle. Each fails differently.
Somebody can satisfy three and breach the fourth entirely by accident. Setting them out separately is what makes the rule usable rather than merely agreeable.
Self-Serving Reviews
A business writing its own reviews is asserting something untrue. So is marking up testimonials the business selected and edited. That second case is where well meaning people go wrong.
The obvious version. Writing them.
Composing feedback and attributing it to customers. Nobody defends this and it still happens, usually where somebody was asked to make a new site look established.
The version that catches decent businesses. Curated testimonials.
A page of quotes chosen by the business from things customers said, tidied for length, presented as a rating. Every word may be genuine and the average is still not a review score.
Why that fails. Selection is the problem.
A rating claims to represent what reviewers gave. A set chosen by you represents what you decided to show. The difference between those two is the whole of the claim.
What makes it separate from sentiment. Truth of the quote is not the issue.
The individual testimonials can be entirely accurate. What is false is the statement that these constitute the ratings customers gave, which is an assertion about a process rather than about opinions.
What to do with testimonials instead. Show them.
Selected quotes on a page are ordinary marketing and entirely legitimate. The problem arises only when they are converted into an asserted rating.
Reviews From Elsewhere
Ratings collected on a third-party platform generally belong to that platform. Restating them as your site's own is a common error and it is usually made in good faith.
Why businesses do it. The reviews are real.
A business with genuine feedback on a review platform reasonably thinks it can show that score. The reviews exist and the average is accurate, so the objection feels technical.
Why it is still wrong. Whose claim it is.
Those reviewers gave their feedback to that platform, under its terms and its verification. Restating the average as your own asserts that you collected them, which you did not.
The practical consequence. Duplication and conflict.
The platform already asserts those ratings. Your site asserting the same figures produces two claims about one set of reviews, from two parties, one of whom did not gather them.
What is legitimate. Linking and displaying.
Showing your score with a link to the source is ordinary and useful. The difference is between reporting somebody else's figure and asserting it as your own data.
The version stores get wrong. Marketplace ratings.
Scores from a marketplace applied to products on your own site. Product markup covers that case alongside the store-wide rating error.
What Happens When It Is Wrong
False markup can result in the enhancement being removed, else in more serious cases a manual action against the site. Both are documented in published policy rather than being anybody's speculation.
The lesser outcome. The enhancement stops.
Ratings cease appearing in results for that site. Nothing else changes. The thing the markup existed for is gone, usually without any notification that would explain why.
The more serious one. A manual action.
A penalty applied by a person reviewing the site rather than by a process. That affects how the site is treated generally rather than only its enhancements. Reversing it requires fixing the problem and asking for review.
Why the second is worse than it sounds. It is not automatic to undo.
Removing the markup does not restore anything by itself. Somebody has to correct the site and submit it for reconsideration, which takes time nobody planned for.
What triggers a look. Frequently a report.
Competitors notice. A business asserting ratings with no visible reviews is conspicuous to anybody in the same market. Reporting it costs them nothing.
Why we state this without softening it. The alternative misleads.
Describing the risk as unlikely would invite somebody to gamble. Our risky practices material sets out how these actions work in general. Nothing about this type is treated as low risk.
It Also Misleads A Customer
Somebody choosing a business partly on a star rating that nobody gave has been misled. That matters independently of whether anything is ever detected. It is why this is a fixed position rather than a calculation.
Who is actually harmed. Not the search engine.
The discussion usually frames false markup as a rules problem. The person affected is a customer who used the rating to decide, believing other people had recommended you.
Why detection is beside the point. The harm is already done.
A business whose fabricated ratings are never noticed has still influenced every customer who saw them. Escaping the consequence does not undo the misleading.
What that means for the decision. It is not a risk calculation.
Framing it as odds invites weighing the chance of detection against the benefit. If the objection is that customers are being deceived, the odds are irrelevant to whether it should be done.
Where this connects commercially. Trading standards.
Misleading consumers about what other customers have said is not only a search matter. That is a genuine consideration and it belongs to somebody's legal advice rather than to us.
Our position. Fixed.
We do not implement this whatever a client asks. How to get star ratings states that plainly and explains what we do instead.
How To Do It Properly
The markup describes reviews rather than creating them, so the collecting comes first. A business wanting rating markup needs a genuine way to gather and display feedback before any of this applies.
The order of operations. Reviews, then display, then markup.
Collect genuine feedback. Show it on the page. Describe what is shown. Reversing that order is exactly how businesses end up asserting things they do not have.
What collecting properly involves. Asking everybody.
A process that invites feedback from all customers rather than the ones you expect to be pleased. Selecting who gets asked produces a score that misrepresents your customers.
What displaying involves. Showing the poor ones too.
Reviews visible on the page, including the unflattering ones. A page of uniformly excellent feedback is less believable than a mixed set. The mixed set is also true.
What the markup then does. Describes what is there.
At that point the assertion is straightforward, because it restates something a reader can verify by scrolling. That is the whole test.
The uncomfortable implication. Most businesses should not bother.
A business without a review process is not one step from this markup, it is several. Block nine covers the answer that suits most of them better.
Where Reviews Should Live Instead
For most local businesses the profile is the right home for reviews. Those appear in results without any markup at all, which answers the underlying want more directly than this type ever will.
What the profile does. Handles it entirely.
Reviews left there are collected, verified and displayed by the platform. The stars appear beside your listing without you asserting anything, which removes the entire question.
Why it suits local businesses better. That is where people look.
Somebody choosing a local trade or shop looks at the map result and its ratings. Reviews on your own site reach people who already found you, which is a smaller and later audience.
Why it is also easier. No maintenance.
Nothing to keep current, nothing to assert and nothing that can become false. The reviews are held by somebody else and shown by somebody else.
What it costs you. Control.
You cannot select which reviews appear, which is precisely why they are believed. That is the trade and it is the correct one.
Where the detail sits. The profile material.
Our Google Business Profile guides cover collecting reviews and handling poor ones. The full series on markup is on the schema and structured data guide.
We will not
implement invented
ratings.
Whatever a client asks, regardless of whether anybody would notice. Somebody choosing a business partly on a star rating nobody gave has been misled, which is true independently of detection. Collect genuine reviews on your profile instead and the stars appear without any markup at all.
What we will and will not do:
This is a fixed position rather than a preference. It is not negotiable on a project.
Every guide.
One practice.
What structured data is, which types apply to your business, how to test it, where the results features come from and what to do when nothing appears.
