Technical SEO
Review and AggregateRating Schema Done Right: Rules, Risks, and What Actually Shows
Self-serving review markup is the fastest way to lose rich result eligibility. What Review and AggregateRating schema can legitimately do for local service sites, and how to implement it without risk.
Star ratings are the most requested and most frequently mishandled piece of structured data on local service websites. The appeal is obvious: a row of stars next to your listing looks like social proof before anyone clicks. The problem is that this is the one area of structured data with explicit rules about who may rate whom, and violating them costs rich result eligibility across the whole site rather than just on one page.
This guide is about the boundary. It covers what Review and AggregateRating markup are actually for, the self-serving review restriction and why it exists, which entity types can legitimately carry ratings, how to represent third-party reviews without misrepresenting them, and the alternatives that give you credibility signals when star markup is not available to you.
It assumes you already have working business markup. For the foundation, see our guide to schema markup for local service websites, and for the layering approach see structured data beyond LocalBusiness. If your existing markup validates but produces nothing, work through the schema diagnostic checklist first.
What the two types actually describe
Review describes a single evaluation: one author, one rating value, one body of text, about one specific thing. AggregateRating summarizes many of those evaluations into an average with a count. They are related but not interchangeable, and the most common implementation error is publishing an AggregateRating with no underlying reviews that exist anywhere a visitor can see.
Both types attach to a subject. That subject matters more than the rating fields, because it determines whether the markup is legitimate at all. A rating attached to a specific product, a specific service, a course, a recipe, or a book behaves differently from a rating attached to the organization publishing the page.
Every rating also needs a scale. Include bestRating and worstRating explicitly rather than assuming a five-point scale is obvious, and make sure ratingValue is a number in that range. Averages should be reported at the precision you actually calculate, not rounded up to something friendlier.
The self-serving review rule, and why it is not negotiable
A self-serving review is a review about a business or organization that the business itself collected and published on its own site, then marked up to produce stars in search results. Guidelines are explicit that ratings for a LocalBusiness or Organization entity gathered and displayed by that same entity are not eligible for rich results. This is not a gray area or a matter of enforcement luck.
The reason is straightforward: nothing prevents a business from writing five perfect testimonials about itself. Ratings only carry information when the rater is independent of the rated. Allowing self-collected business ratings to draw stars would make the feature meaningless within a week, so the restriction protects the signal rather than punishing marketers.
The practical consequence is that a testimonials page on your own site should not carry AggregateRating markup about your business. You can absolutely display those testimonials, and you should, because they persuade humans and provide extractable evidence for AI systems. You just should not dress them up as machine-verified ratings.
What violating it costs
The downside is asymmetric. At best the markup is ignored and you gain nothing. At worst the site receives a structured data manual action, which removes rich result eligibility for all types, including the FAQ and breadcrumb markup you implemented correctly, until the issue is fixed and reviewed. Trading sitewide eligibility for stars on one page is a bad bet.
Where ratings are legitimate
Ratings attached to specific offerings you sell, where a genuine review collection process exists and the reviews are visible on the page, are on much firmer ground than ratings about your company as a whole. If you run a productized service with real customer feedback displayed per offering, that is a defensible implementation. A single sitewide company average is not.
Third-party reviews: representing without misrepresenting
Most local service businesses hold their real review equity on third-party platforms. The temptation is to sum those counts and average those scores into one impressive AggregateRating on the homepage. Resist it, for two reasons. First, aggregating across platforms with different scales, moderation standards, and time windows produces a number that does not correspond to anything verifiable. Second, and more importantly, republishing platform ratings as your own markup is precisely the pattern the self-serving rule addresses.
The honest approach is to display third-party reviews as third-party reviews. Attribute each one to its platform and author in the visible content, link to the source where terms permit, and state counts and averages per platform rather than blended. Machines and humans can both verify a claim like that, and verifiability is the entire point.
There is a real upside to doing this cleanly. AI systems assembling a recommendation weight corroborated reputation heavily, and a page that clearly states where reviews live, how many there are, and what the average is per source gives them extractable, checkable facts. That is a better outcome than stars you were never eligible for.
Implementation details that separate correct from careless
When you do have a legitimate case for review markup, the details determine whether it survives scrutiny.
Visibility parity
Every review you mark up must be visible on the page, with the same author, the same rating, and substantively the same text. Marking up ten reviews while showing three is a mismatch, and mismatches are what reviewers look for when assessing intent.
Real authors, real dates
Include an author name as given by the reviewer and a datePublished that reflects when the review was actually written. Undated reviews age invisibly, which is a problem for a business whose crew and quality may have changed substantially since a review from four years ago.
Counts that match reality
reviewCount and ratingCount are not decorative. If the number in your markup exceeds what a visitor can find, you have created a discrepancy that undermines every other claim on the page. Update these programmatically from the same source that renders the visible content, never by hand.
Do not attach ratings to articles or pages
A rating belongs to a thing being evaluated, not to a blog post about a topic or a service area page. Attaching ratings to article-type content is a common plugin behavior and a reliable way to have markup discarded.
What to do instead when stars are off the table
Losing eligibility for a visual feature does not mean losing the credibility benefit. Reviews influence local search outcomes primarily through your business profile, not through markup on your own site, and they influence AI recommendations through corroborated reputation across independent sources.
So the higher-leverage work is operational rather than technical: a steady request process that produces recent reviews at a consistent rate, responses to every review including the difficult ones, and specificity in the reviews themselves, because a review that names the service and the city is far more useful as evidence than one that says "great job." Our guide to review velocity and reputation management covers that system in depth.
On the page itself, present reputation as extractable facts. State how many reviews you hold on each platform, the average on each, the date range, and let visitors verify. Pair that with the answer-first structure described in page structure that wins AI extractions, and you get most of the trust benefit without any of the markup risk.
A short audit for review markup already on your site
If your site inherited review markup from a theme, a plugin, or a previous agency, run this before assuming it is helping.
Five questions
Is there an AggregateRating attached to your LocalBusiness or Organization entity? Were those ratings collected and published by you? Do all marked-up reviews appear visibly on the same page? Do counts in the markup match what a visitor can find? Are ratings attached to any article or informational page? A yes to the first two, or to the last, means removal rather than adjustment.
How to remove safely
Strip the rating nodes while keeping the visible testimonials and the surrounding business markup intact. Nothing about the page's usefulness depends on the rating fields. Then check for a manual action notice before assuming you are clear, and request review if one exists.
Frequently Asked
Questions & answers
Can I add star rating markup to testimonials on my own website?
Not for rich results. Ratings about your own business that you collected and published are self-serving and ineligible. You can display the testimonials, but adding AggregateRating about your organization risks having markup ignored or triggering a manual action.
What is the difference between Review and AggregateRating?
Review describes one evaluation by one author about one subject. AggregateRating summarizes many evaluations into an average and a count. AggregateRating should never appear without underlying reviews that genuinely exist and are visible.
Can I mark up reviews that came from a third-party platform?
You can display and attribute them, but republishing platform ratings as your own structured data generally falls under the same self-serving restriction. Attribute each review to its platform and author in visible content and state per-platform counts rather than a blended average.
What happens if I use review markup incorrectly?
The best case is that the markup is ignored. The worst case is a structured data manual action that removes rich result eligibility for every type on the site, including correctly implemented breadcrumb and FAQ markup, until it is fixed and reviewed.
Do reviews still affect local rankings without markup?
Yes. Review quantity, recency, rating, and response behavior on your business profile influence local visibility regardless of on-site markup, and corroborated reputation across independent sources influences whether AI assistants recommend you.
Should I include bestRating and worstRating?
Yes. State the scale explicitly rather than assuming a five-point default, and make sure the reported average falls within it at the precision you actually calculate rather than rounded upward.
Can ratings be attached to individual services?
Ratings for specific offerings with a genuine review collection process and visible reviews on the page are on firmer ground than a company-wide average. Do not create per-service ratings by splitting one general company score across your offerings.
Put this into practice
Check your market for exclusive leads
See whether your service area and category are still open for exclusive representation.
Check availability