Google’s Review Snippet Guideline (July 24, 2026): What Michigan Service Businesses Must Fix Before Yelp’s Reviews Reach ChatGPT

On July 24, 2026, Google added a single sentence to its review snippet structured data documentation: don’t include fake or undisclosed incentivized reviews on your page or in your markup. One day earlier, on July 23, Axios reported that Yelp had licensed roughly 330 million reviews and more than 8 million business listings to OpenAI for use inside ChatGPT.

Those two items ran in different sections of the trade press and belong in the same conversation. Reviews just became a bigger input to how customers find Michigan businesses, and simultaneously a bigger liability in the code on your own site. If your service pages carry star ratings and your front desk hands out gift cards for feedback, you have work to do this week.

What exactly did Google change in the review snippet guidelines?

Google added one guideline to the review snippet (Review and AggregateRating) documentation, reported by Barry Schwartz at Search Engine Land on July 24, 2026. The new line prohibits fake or undisclosed incentivized reviews both on the page and inside the structured data.

Google gave two examples of what it means. The first is reviews not based on a genuine experience of the product or service. The second is the one that will catch honest businesses off guard: reviews written in exchange for a benefit such as money, discounts, vouchers, or free products that don’t clearly and prominently disclose the incentivization.

Read that second example twice. Incentivized reviews are not banned. Undisclosed incentivized reviews are. A Sterling Heights flooring retailer that gives a $25 gift card for a review is compliant if the incentive is disclosed clearly and prominently next to the review, and non-compliant if it isn’t. The guideline sits in the eligibility section governing whether your page can show star ratings at all, so the enforcement mechanism is a manual action against your structured data.

Why does this matter more now that ChatGPT has Yelp’s review data?

Because review data now feeds more systems than it did two weeks ago, and those systems are less forgiving than a human shopper. Yelp shares rose 8% on the announcement, and because the deal is non-exclusive, Yelp remains free to license the same corpus to other AI companies.

Here is the mechanism worth understanding, because it predicts a lot of cases nobody has written about yet. Answer engines don’t read your star rating as a picture. They read it as a number attached to an entity, cross-referenced against the same entity’s numbers on third-party platforms. When a generative engine assembles an answer to “best furnace repair in Livonia,” it is reconciling your site’s AggregateRating against Yelp, Google Business Profile, and whatever else it has licensed. Consistency across those sources reads as corroboration. A 4.9 on your own site against a 3.8 on Yelp reads as an unreliable source, and the engine resolves the conflict by citing the competitor whose numbers agree with each other.

Does a structured data manual action hurt your rankings?

No. A structured data manual action causes Google to ignore the markup on the affected pages. The pages themselves keep ranking where they ranked. This is the detail most people get backwards, and getting it backwards leads to the wrong diagnosis.

The loss is entirely in click-through rate, which is why it goes unnoticed for weeks. Picture a Livonia HVAC contractor holding position three for “furnace repair Livonia” with 2,000 monthly impressions. With stars, that listing might earn 6%. Without them, against two competitors that still show ratings, 3.5% is a reasonable expectation. That is roughly 50 lost sessions a month on one query, and it hurts most in December, when a furnace call is worth several hundred dollars and metro demand peaks.

Nothing in Search Console’s performance report flags this. Average position is unchanged. Only clicks fall.

How do you tell why your stars disappeared?

Match the symptom to the cause before you rewrite anything. Four patterns account for nearly all star loss, and each has a different fix.

• A manual action notice in Search Console under “Structured data issue.” This is unambiguous. Something in your markup violates a guideline, now including the incentivization rule. Fix, then file a reconsideration request.
• Stars gone, no manual action, valid items still reported in the Review Snippets enhancement report. Your markup is fine and Google is choosing not to display it. Rich results have always been discretionary, and display rates fluctuate. Don’t rewrite working schema in response to this.
• Valid items dropping to zero in the enhancement report. A technical break, usually a template change, a plugin update, or JavaScript-rendered markup Google no longer sees. Run the Rich Results Test on a live URL, not on pasted code.
• Stars intact but AI answers stopped naming you. Not a markup issue. This is a corroboration gap between your on-site rating and your third-party profiles, and it is the failure mode the Yelp licensing deal makes more expensive.

How should a Michigan service business audit its review program this week?

Seven steps, all executable by a team of three without an enterprise stack.

1. Inventory every page emitting Review or AggregateRating markup. Crawl the site and filter for those schema types. Most multi-location businesses find markup on templates nobody remembers building, including location pages spun up years ago.
2. Write down every way you currently ask for reviews. Post-service texts, invoice inserts, front-desk cards, referral programs, contest entries, employee bonuses tied to review counts. Include the ones that predate the current marketing team.
3. Flag anything that trades a benefit for a review. Gift cards, service discounts, free filters, drawing entries, waived fees. A benefit contingent on leaving a review is incentivization even when you never asked for a positive one.
4. Add clear and prominent disclosure, or stop the incentive. Prominent means adjacent to the review, in body-copy size, readable without expanding anything. A footnote in your privacy policy does not qualify.
5. Remove incentivized reviews you cannot retroactively disclose. If you can’t identify which of your 340 reviews came from the gift-card era, the safer move is pulling AggregateRating from that page until the pool is clean.
6. Reconcile your on-site rating with Google Business Profile and Yelp. A gap larger than about 0.4 stars is worth investigating, because it usually means your site is averaging a filtered subset rather than everything you’ve received.
7. Re-test and set a recurring check. Run the Rich Results Test on three representative URLs, then calendar a quarterly re-check. Plugin updates break schema silently.

Multi-location operators should expect step one to run long. A dental group with offices in Royal Oak, Troy, and Dearborn typically has three generations of location-page templates live at once, the same structural problem we work through in our local SEO engagements with Detroit-area service businesses.

What dates belong on your calendar?

• July 23, 2026 — Yelp and OpenAI licensing deal reported by Axios; Yelp review data begins reaching ChatGPT users.
• July 24, 2026 — Google’s new review snippet guideline published in the Search Central documentation.
• October 21, 2024 — the FTC’s Rule on the Use of Consumer Reviews and Testimonials took effect; it is already enforceable against you today.
• December 2025 — the FTC sent warning letters to ten companies over possible violations of that rule, per its own announcement.
• September through November 2026 — the window to fix this before Michigan heating season, when furnace, roofing, and restoration searches spike and a missing star rating costs the most.

The FTC angle is not decoration. Civil penalties under that rule run up to $53,088 per violation, which dwarfs anything a lost rich result costs. Google’s guideline is the cheap early warning.

Is this guideline actually a big deal, or just Google restating the obvious?

Search Engine Land called the addition “somewhat obvious,” and on the fake-review half, that’s fair. Fabricated reviews were never permitted. But the disclosure half is a real shift, and it lands on the wrong people.

Google has converted an advertising-law disclosure obligation into a technical eligibility requirement for a search feature, and put the burden on the site publishing the markup rather than the platform hosting the reviews. Operations running actual fake-review schemes mostly aren’t marking up schema on their own domains. The businesses exposed here are ordinary ones: the Ann Arbor orthodontist running a $50-off referral promotion, the Grand Rapids landscaper entering reviewers in a spring drawing. They aren’t committing fraud. They are non-compliant on a disclosure technicality nobody told them about, and enforcement arrives as a manual action rather than a warning. If you run any review incentive, treat the audit above as urgent and pair it with the kind of technical audit that catches schema drift before it costs traffic.

Frequently asked questions

Can I still offer a discount in exchange for a review?

Yes, under Google’s guideline, provided the incentive is disclosed clearly and prominently wherever the review appears and in the structured data. The safer framing is offering the benefit for feedback regardless of sentiment, never contingent on a positive rating. Note that Google’s rules and the FTC’s Consumer Reviews Rule are separate obligations, and satisfying one does not automatically satisfy the other.

How long does a structured data manual action take to clear?

Google does not publish a service level for reconsideration requests, and reported turnaround varies widely. Plan for weeks rather than days. Fix the underlying issue across every affected template before filing, because a partial fix produces a rejection and restarts the clock. Document what you changed in the request itself.

Does this affect my Google Business Profile reviews?

The new guideline governs review snippet structured data on your own website, not the reviews on your Business Profile. Google Business Profile has long had its own prohibition on incentivized reviews, enforced separately through review removal. Both matter, and the practical answer is to run one compliant solicitation process rather than maintaining two sets of rules.

Should I remove AggregateRating markup to be safe?

Not as a default. Star ratings still earn meaningful click-through, and removing valid markup surrenders that for nothing. Remove it only where you cannot verify the review pool, such as pages averaging reviews collected during an undisclosed incentive program. Restore it once you have a clean, disclosed set to average.