How to get Google reviews (and reply to them) without breaking the rules

97% of consumers read Google reviews before choosing a local business. And the fastest way to lose yours isn't asking badly: it's rewarding them, filtering who gets asked, or marking them up as if they were yours. Google removes all three.

Why Google reviews matter (and how far that goes)

BrightLocal's February 2026 survey sets the context: 97% of consumers read reviews of local businesses, 41% read them "always" when searching (29% in 2025), and 85% are more likely to use a business after reading positive reviews.

Relevance, distance and prominence

Google explains the local pack with three factors: relevance, distance and prominence. Reviews feed the third one, which, according to the official documentation, "is based on information like the number of websites that link to your business and the number of reviews you have". More reviews and positive ratings mean better local ranking. There are no published weights: anyone selling you an exact formula is guessing.

What a review can't fix

A profile with the wrong category, an inconsistent address or no service page isn't saved by two hundred reviews: they are an entry point, not a substitute for the profile work. If your profile is suspended, no review campaign brings it back; with multiple locations, each storefront competes with its own history.

It's not just the stars

The same study shows 92% look at the rating, 31% only use businesses rated 4.5 or higher (17% in 2025), 44% care that a review is from the last month and 37% that the owner replied. What decides is the recent average, not the historical one.

The rules that can cost you your profile

Google treats fake or incentivized reviews as "fake engagement": it removes the reviews and also restricts profiles.

Incentives: discounts, gifts, raffles

Offering payment, a discount, free product or a raffle entry in exchange for posting, changing or removing a review is prohibited, and so is the "in kind" version. If the benefit depends on the review existing, it's an incentive.

Filtering who gets asked (review gating)

Discouraging negative reviews or asking only customers you believe are happy is prohibited. The classic funnel, "did you enjoy it? yes, then Google; no, then an internal form", is exactly what the policy points at.

Counter pressure and staff quotas

Pushing for the review while the customer is on site, setting staff a number of reviews per shift, or asking them to request specific content isn't allowed either. The old advice to "ask before they walk out the door" no longer fits.

What happens if you get caught

Sudden spikes, new accounts or identical texts trigger automated review: the profile stops receiving reviews for a period, existing ones are hidden and a public notice about removed fake reviews appears. There is an email notice and an appeal.

The legal context, without the panic

In the United States, the FTC rule on consumer reviews (16 CFR Part 465) has been in force since 21 October 2024 and allows civil penalties of around $53,000 per violation, the figure adjusted in 2025. It's US law, not a Google rule.

How to ask for Google reviews: the system that works

Asking for reviews isn't prohibited. Rewarding, filtering and pressuring are.

1. The right link and QR code

Your profile generates a direct link to the review form and a QR code. Use that link, not "search for my business on Google": every extra step loses people. Put it in the site footer, the email signature, the invoice, the closing WhatsApp message and at the counter.

2. The moment

When the job closes or the order is delivered, with the experience fresh. Asking before the customer has seen the result turns a review into a promise and usually produces generic text.

3. The message

Ask for an honest opinion, not five stars. Don't dictate the text or suggest phrases. Three lines: what was done, when, and the link. Sending can be automated; choosing recipients can't.

4. Who to ask

Every customer of the period, not just the ones you think are happy. That's the line between a system and a violation: if sending depends on your prediction, you're filtering.

5. Consistency

A continuous flow at every close, not a one-week campaign. With 44% checking whether the latest review is from the last month, frequency beats accumulated volume. Google publishes no threshold, so be suspicious of "you need twenty a month".

Reviews on your own site: the markup mistake

The most expensive mistake is also the quietest: nothing looks broken, it just doesn't deliver what the owner thinks they bought.

Self-serving reviews earn no stars

Google's documentation is blunt: if the entity being reviewed controls the reviews about itself, those pages are not eligible for review snippets with stars. It makes no difference whether the markup is hand-written or a third-party widget.

AggregateRating on LocalBusiness

The typical case: a business marks up its own listing with 4.9 and 187 ratings.

{
  "@type": "LocalBusiness",
  "name": "Example Garage",
  "aggregateRating": { "@type": "AggregateRating",
    "ratingValue": "4.9", "reviewCount": "187" }
}

Syntactically valid, not eligible for stars, because the business controls the ratings. And markup that exaggerates what's visible can end in a manual action.

You can't aggregate reviews from other sites

Review or rating data from other websites doesn't go into your markup: not from your Google profile to your site, not from a marketplace where you sell.

What actually helps

Showing real reviews helps conversion and helps AI assistants read them, without markup promising stars you won't get. If you use a review app, check which JSON-LD it outputs: many publish an AggregateRating about the business and the owner assumes it's optimized.

Replying to Google reviews: the part almost everyone gets wrong

The reply depends only on you, and it's where the worst reflexes show.

What Google says (and doesn't)

Google states that replying shows you value customer feedback and that "positive reviews and useful responses can help your business stand out". It doesn't present it as a guaranteed factor: it isn't a ranking trick.

Negative reviews, in three steps

Acknowledge the problem without excuses, state the action taken and offer a private channel to resolve it. No customer data in the public reply.

Positive reviews: no need to thank each one

The guidance recommends replying when you have new or relevant information and notes that you don't need to publicly thank every reviewer. A good counterpoint to "reply to 100% within 24 hours": that rule doesn't exist.

Fake reviews, spam and extortion

Report them from the profile; there's a dedicated channel for review-based extortion attempts. Don't reply with accusations and don't pay. Log dates and keep your own screenshots.

The four most common mistakes

The same copy-pasted reply everywhere, arguing in public, offering a discount to change the review (another incentive) and replying with customer data.

Reviews and AI search

The same study adds a new layer: 82% read AI-generated review summaries and 23% rely on them alone. That summary is built from real, recent and varied reviews.

What makes a review quotable

Concrete detail: service, city, timeframe, result. "Very professional" adds nothing to a summary; "they fitted the boiler in two days, on a Sunday" does. You can't ask for that text, but you can make it more likely.

What you don't control

Which reviews the AI summarises and in what order. You do control that they exist, that they are recent and that services and neighbourhoods vary, the same logic behind GEO and citations in AI assistants.

How to measure it

Six monthly metrics: new reviews per month, average rating for the last quarter, reply rate, median response time, service and city mentions, and profile actions.

What doesn't work as a single KPI

The rating on its own. With 31% requiring 4.5 or higher, going from 4.9 to 4.7 isn't a crisis. What moves the needle is steady volume and replies with substance.

An eight-point checklist

1. Direct link and QR published where the customer closes the service.
2. A request at every close, to every customer, without asking for stars or dictating text.
3. Zero incentives, zero raffles, zero recipient filters.
4. Check which JSON-LD your site outputs and remove the self-serving AggregateRating.
5. Reply templates for negatives and for fakes, with no personal data.
6. Reporting the reviews that break the policy.
7. A monthly panel: new reviews, quarterly rating, reply rate.
8. Re-read Google's policy before any acquisition campaign.

Reviews aren't bought or manufactured: they're requested from everyone, at the right moment, and answered with useful information. That's the only version that also survives a policy review.