Most experts handle testimonials with two mistakes at once: they expect review schema on their own site to earn Google stars, which it will not for a service or a person, and they park their best praise on surfaces machines barely read. This guide gives you the honest rules: what Review and AggregateRating markup actually does, why Google killed self-serving stars in 2019, where testimonials must live for AI assistants to count them, and how to structure each one so it works as evidence.
A testimonial is a witness statement about you. The question that decides its value is not how glowing it is, but whether the machines that now vet you ever get to hear the witness.
What review markup actually is, and what it is not
Structured data gives machines a labelled version of what a page says. The Review type marks a single evaluation, with a named author, a rating and a body of text, and AggregateRating summarizes many of them into an average and a count. Labelling praise this way is genuinely useful: it removes ambiguity about who said what about whom, which is exactly the kind of clean attribution this whole discipline is built on.
What the markup is not is a lever that makes Google decorate your listing with stars. That confusion costs service professionals real time and, when agencies sell it as a deliverable, real money. The display of review stars in search results, what Google calls a review snippet, is governed by a short list of eligibility rules, and a consultant, advisor, coach or agency marking up praise about themselves fails those rules twice over. Knowing exactly where the line sits is the difference between structured trust and decorative code.
Does review schema work for a service business or a personal brand?
Here is the straight answer, from Google's own review snippet documentation. Review rich results are supported for a fixed set of content types: books, courses, events, movies, products, recipes, software applications and local businesses, with a critical catch on that last one. Person is not on the list at all, so reviews about you as an individual cannot produce a review snippet no matter how they are marked up. Generic services are not on the list either.
The second barrier is the self-serving rule. Since a 2019 policy change, Google ignores reviews for LocalBusiness and Organization types when the entity being reviewed controls the reviews, meaning praise about your business published on your own site, or piped in through a third-party widget you embed. The stars you still see on service sites in the wild are either legacy screenshots, markup Google is silently ignoring, or borderline implementations that risk a manual action for review spam.
Review markup about yourself, on your own site, earns you nothing in Google and nothing extra in AI answers. The legitimate uses are narrower: mark up reviews of eligible things you sell, a course, a book, a product, on their own pages, and keep testimonials about you as clean, attributed, crawlable text.
One genuine exception deserves a flag. If you sell a course or a book, those are eligible types, and reviews of that specific product, collected and displayed on its page with correct markup, can earn review snippets. An advisor cannot get stars for "working with me," but the same advisor's flagship course can carry legitimate stars. Route your review-collection energy accordingly.
Why testimonials matter more to AI, not less
If stars are mostly off the table, why is this worth a strategist's attention? Because the machines that increasingly decide who gets hired do not need stars. AI assistants read text, and third-party praise is among the strongest text evidence that exists about a person. When an engine weighs whether to recommend you, it is effectively asking whether independent sources corroborate your claims, the same corroboration logic that runs through every rung of the trust ladder. A named client describing a specific outcome is corroboration in its purest form.
The catch is the word independent. A machine deciding how much weight to give praise looks at where the praise lives. Ten anonymous quotes on your own homepage are claims you make about yourself, weighted accordingly, close to zero. The same ten quotes as Google Business Profile reviews, Clutch entries or a case study on the client's own domain become independent witness statements, each one crawlable, attributed and hard for you to have manufactured. Same words, different surface, different verdict.
Where should testimonials live? The placement table
| Surface | What machines do with it | Verdict |
|---|---|---|
| Google Business Profile | Feeds Maps and local results, shows native stars, and gets read into AI answers about local providers | First priority for any client-facing professional |
| Independent review platforms (Trustpilot, Clutch, G2) | Treated as third-party corroboration; frequently crawled and cited by assistants in vendor answers | High value, pick the one your buyers actually check |
| Client's own site or case study | The strongest independence signal: someone else's domain vouching for your name with specifics | Rare and worth negotiating for in every engagement |
| LinkedIn recommendations | Attributed to real professional identities on a heavily referenced platform | Useful supporting layer, not a foundation |
| Your own website | Read as self-published claims; no stars for services, modest evidentiary weight | Keep for human conversion, mirror the originals that live elsewhere |
| Images and script-only sliders | Screenshots and JavaScript-only carousels are barely readable to crawlers at all | Where testimonials go to be invisible; convert to real HTML text |
The strategy the table implies: collect praise onto independent surfaces first, then quote it on your own site with a link back to the original. Your homepage testimonial block becomes a mirror of verifiable statements rather than a gallery of unverifiable ones, and every quote gains the weight of its original location.
How to structure a testimonial machines can actually use
- Full attribution. Name, role and company, with permission. "Sarah M." is decoration; a named director at a findable company is evidence an engine can cross-reference.
- A problem and an outcome in the text. "Great to work with" carries no information. "Rebuilt our intake process and enquiries doubled in a quarter" gives a machine a fact pattern it can attach to your name for exactly the queries you want to win.
- Crawlable HTML. Real text on a real page. Not an image of a tweet, not a widget that renders only in the browser.
- A link to your canonical identity. Wherever the praise lives, it should point at the same name, and ideally the same domain, as everything else about you, so the reference lands in the right file.
- Specific beats voluminous. Five detailed, attributed accounts across two independent platforms outweigh fifty anonymous blurbs. Machines are counting witnesses, not adjectives.
Treated this way, testimonials stop being a website section and become entries in your permanent evidence file, one shelf of the broader proof portfolio that a serious expert assembles deliberately. They also compound with your professional record: for a doctor, lawyer or advisor whose category is heavily trust-gated, the review layer sits directly alongside credentials, as we mapped in PEO for doctors, lawyers and experts.
Building the collection system, so the evidence never dries up
Almost nobody has a testimonial problem; they have a collection problem. The praise exists, spoken on calls, buried in thank-you emails, mentioned in renewal conversations, and it evaporates because no system catches it. The fix is a standing routine, not a heroic annual scramble.
Ask at the moment of the win, not at the end of the engagement. The strongest testimonials are captured within days of a result landing, while the specifics are fresh and the gratitude is real. Build the ask into your delivery checklist: when a milestone hits, the request goes out that week. And make the ask directional rather than blank. "Could you write a few words about working with me" produces adjectives. Three prompts produce evidence: what problem were you facing, what changed, what would you tell someone considering this work. The client writes in their own voice, but the structure guarantees a problem and an outcome end up in the text.
Route each captured piece deliberately, using the placement table above. The client comfortable with a public Google or Clutch review goes there first, because independent surfaces outrank your own. The client whose company forbids public reviews can often still approve a named quote in a case study, and the client who cannot be named at all can sometimes let their results be described anonymously, which is worth little to machines but still helps human buyers on your site. One ask, three graceful landing spots, no wasted praise.
Finally, keep a ledger: a simple spreadsheet of every testimonial, where it lives, who gave it, the date, and the permission granted. Reviews age. A 2019 quote about a service you no longer offer is clutter, and a platform profile that went quiet three years ago reads as decline. A quarterly half-hour pass through the ledger, refreshing, retiring and re-requesting, keeps the trust layer current, which matters to buyers and machines alike.
The honesty rule, because machines keep receipts
A closing boundary. Do not fabricate reviews, do not launder them through fake profiles, and do not buy them. Beyond the legal exposure, US and UK regulators both treat fake reviews as actionable deception, there is a mechanical reason: review fraud leaves patterns, platforms invest heavily in detecting it, and a purge or a penalty becomes part of the public record that engines read about you forever after. The same discipline that makes PEO work, real signals honestly earned, applies double to the trust layer, and the full argument for why shortcuts backfire is in the limits, risks and ethics of PEO.
Earned honestly and placed deliberately, though, reviews are among the highest-leverage assets in this whole discipline: they are the one signal category where your happiest clients do the writing for you. If you want your existing praise audited, restructured and moved to the surfaces machines actually read, that is a standard part of our engagements.
Questions
Does review schema on my own site get me star ratings in Google? +
Do testimonials still matter for AI visibility if they cannot earn stars? +
What makes a testimonial machine-readable? +
See what AI says about you today.
Start with a reading. We show you the words the engines return about your name, then map the fastest signal to move.
Get named →