Who your review automation decides not to ask
The flow looks harmless when somebody demonstrates it. A customer finishes a job. A message goes out asking how it went, one to five stars. Four or five and they get a link to your Google listing. One or two and they get a form that comes to you privately, with a kind sentence about wanting to put things right.
That fork arrived as a feature, with a name, in a product bought to save time on asking for reviews; nobody in your business designed it. It is called review gating, and the interesting thing about it is how differently the law and the platforms treat it.
I am not a lawyer, none of this is advice about your obligations, and the rules below are United States federal ones plus the contracts of seven platforms. What follows is what the regulation says, what it pointedly does not say, and what the products actually ship.
Start with the answer that surprises people
The Federal Trade Commission's Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465, took effect on 21 October 2024. It is the rule behind every headline about banning fake reviews, and it carries civil penalties for knowing violations.
It does not ban asking only your happy customers. The Commission says so itself, in a question and answer written for businesses:
"Can my business ask for reviews only from customers whom we think are happy with our services?
The rule does not contain a specific prohibition against such conduct. But this practice could violate the FTC Act. See, e.g., Endorsement Guides 16 C.F.R 255.2(d) and (e)(11)."
FTC, Consumer Reviews and Testimonials Rule: questions and answers
A great deal of published advice says the opposite, so the answer bears repeating: the rule leaves this conduct alone. An older instrument reaches it, and reaches it more gently than most people writing about it admit.
What does apply, and how hard it bites
The Endorsement Guides have carried an example about this since their July 2023 revision, more than a year before the rule took effect. Example 11 describes gating in the regulation's own words:
"A marketer contacts recent online, mail-order, and in-store purchasers of its products and asks them to provide feedback to the marketer. The marketer then invites purchasers who give very positive feedback to post online reviews of the products on third-party websites. Less pleased and unhappy purchasers are simply thanked for their feedback. Such a practice may be an unfair or deceptive practice if it results in the posted reviews being substantially more positive than if the marketer had not engaged in the practice. If, in the alternative, the marketer had simply invited all recent purchasers to provide feedback on third-party websites, the solicitation would not have been unfair or deceptive, even if it had expressed its hope for positive reviews."
16 CFR 255.2(e)(11)
Two qualifiers hold that sentence up. May be, not is. And if it results in the posted reviews being substantially more positive than they otherwise would have been. That is a test about effects, applied afterwards, on facts.
The Commission put the same hedge in its own voice when it proposed the rule: "whether any given instance of review-gating is deceptive can be highly fact specific."
And the Guides are guides. They do not carry the civil-penalty authority the rule carries. When the FTC sent warning letters about the rule on 22 December 2025, the figure it named was "up to $53,088 per violation" for knowing violations of the rule, and that is a different instrument from Example 11.
So anyone telling you the FTC has banned review gating is compressing three documents into one. The accurate version is quieter and, I think, more useful.
The consequence that is inside the rule, and that nobody quotes
There is a real effect of gating within Part 465, and it lives in a place people do not look: the paragraph that protects ordinary businesses from being liable for a review somebody else wrote.
Section 465.2(d)(1) exempts reviews "that resulted from a business making generalized solicitations to purchasers to post reviews or testimonials about their experiences". If you email every customer and one of them turns out to be your cousin, or an employee, or writes something false, that exemption is what keeps the review outside sections 465.2(b) and (c).
The rule's preamble then defines what makes a solicitation generalized:
"By 'generalized solicitations,' the Commission means to exempt from Sec. 465.2(b) and (c) solicitations sent to large groups of customers, such as those who purchased a particular item or who became customers during a given time period, where specific customers are not chosen based on the likelihood that they will express a particular sentiment. In contrast, solicitations made only to customers whom the business believes to be happy customers would not be 'generalized solicitations' and would therefore be subject to Sec. 465.2(b) and (c)."
Federal Register, 89 FR 68034, 22 August 2024
Gating leaves section 465.2 itself intact and removes the shelter that a send-to-everybody campaign has. The filter your tool ships with is, in the Commission's own construction, the thing that stops your campaign counting as the safe kind.
The platforms are far less relaxed
While the federal position is an effects test with qualifiers, the places the reviews actually go say the plain thing. I read seven platform policies on 8 September 2026.
| Platform | Asking allowed | Choosing whom to ask |
|---|---|---|
| Google Business Profile | Yes, explicitly | Prohibited, by name |
| Trustpilot | Encouraged | Prohibited, including by timing |
| G2 | Encouraged | Prohibited, mechanism named |
| Capterra | Yes | Prohibited as sentiment targeting |
| Apple App Store | Only through Apple's own prompt | Custom prompts disallowed |
| Amazon | Yes, if unbiased and un-incentivised | Addressed indirectly |
| Booking.com | Not addressed | Not stated |
Google's prohibited content page puts both halves in one place. Under things merchants may not do:
"Discourage or prohibit negative reviews, or selectively solicit positive reviews from customers"
And, a few lines down, under things merchants may do:
"Solicit or encourage the posting of content that does represent a genuine experience, without offering incentives to do so or attempting to influence the rating or the contents of the review."
Google, prohibited and restricted content
G2 names the mechanism rather than the outcome:
"Biased Review Collection: collecting reviews by any method that segments out potentially negative reviews or exclusively targets positive reviews..."
G2 community guidelines
Trustpilot goes furthest of anybody, and its third bullet is the one to sit with:
"Be selective with invitations, by picking and choosing which customers you invite. This includes
Inviting only those people who you know have had a positive experience
Setting review invitations to go out to customers at a certain stage in the customer journey that will only be reached by people who have had a good experience
Pre-screening people who leave a review by directing customers who have had a good experience to leave a review and those with a bad experience to contact you directly"
Trustpilot guidelines for businesses, version 7.2, June 2026
The middle bullet catches something none of the US sources I read treats as gating at all: sending the request at a point in the journey only satisfied customers reach. Asking after a successful delivery rather than after any delivery is a filter made of timing, and it has the same effect as a filter made of stars.
Apple removed the problem structurally instead of forbidding it. Developers must use Apple's own review prompt, and "we will disallow custom review prompts", which means there is no place in the flow to put a fork.
What your tool actually ships
The fork is a documented product feature with a settings page.
GatherUp describes its flows in a help article dated 23 July 2025. The first one:
"Ratings & Review Flow (formally Review Mode) is a 2-step process designed to create a positive and negative flow based on initial rating that leads to 3rd party review collection."
The vendor's typo for "formerly" is in the original. Note what the third option is called by contrast: the flow with no branch is "the quickest way to more public 3rd party reviews without capturing an initial rating".
Other vendors sell the same mechanism under names like a survey-to-review loop: ask the happy respondents to publish, intercept the unhappy ones privately before anything becomes public.
Judge.me sells a softer version for shops: a rating threshold that puts a "Contact us" screen in front of a low rating before it is submitted, with configurable settings of one, two or three stars and under.
Every one of those is a supported feature, documented, in products bought by ordinary businesses to save an hour a week.
Which of your automations makes a decision you would not have made by hand? That is one of the things a process audit is for: going through what runs, what it decides, and what each piece is actually worth.
The line the industry drew for itself
This is where an owner can be caught out while doing everything the vendor recommends.
Several reputation vendors publicly disclaim gating, and the clearest of them is BrightLocal, which explains, in a help article whose URL is literally the question, that it does not gate:
"No, we don't use review-gating in Review Generation for Reputation Manager (RM)."
Then, a few lines further down the same page, under "How do we deal with reviews?":
"When we contact your clients or customers and ask them to rate your service, we funnel them into either a 'Positive' or 'Negative' feedback flow."
"The key thing to note here is that they have the option to leave a review for your business in both flows. If the customer reports a negative experience, we do not remove the option to leave a review."
BrightLocal, does the Review Generation process use review gating
That is a coherent position, honestly stated, and it shows exactly where the industry has drawn its line: sentiment branching is fine, withholding the link is not. Both flows end at the same review sites, so nobody is filtered out.
Now hold that against Trustpilot's bullet about journey timing. A funnel that sorts by sentiment and shows both groups the same links satisfies the industry definition completely and still fails Trustpilot's, because Trustpilot is policing the composition of who gets asked rather than the mechanics of the link.
Two defensible standards draw two different lines, and one automation cannot satisfy both. This is a version of a problem this blog keeps running into: when two systems disagree about the same field, somebody has to decide which one governs, and if nobody does, the tool decides by default.
What you are allowed to do, which is more than people think
An article that only listed prohibitions would misstate the position badly, so here is the other half, from the same rule.
Section 465.7(b), on suppression, is narrower than its reputation. It bans materially misrepresenting that displayed reviews "represent most or all the reviews submitted" while suppressing by rating or sentiment. And it says explicitly that a review is not considered suppressed on those grounds where the criteria are "applied equally to all reviews submitted without regard to sentiment". The rule's own examples: the review contains trade secrets or confidential commercial information; defamatory, harassing, abusive, obscene, vulgar or sexually explicit content; another person's personal information or likeness; or discriminatory content. Or it is clearly false or misleading, the seller reasonably believes it is fake, or it is wholly unrelated to what you sell.
The FTC's guidance adds three more permissions that people are commonly told they have lost. Organising reviews is not suppressing them, and defaulting to five-star-first is not a violation. Using particularly positive reviews in marketing materials is outside this provision, though it can still be deceptive under the FTC Act. And contacting a customer who left a bad review to fix the problem is fine: the answer to whether the rule prohibits it is "No."
Incentives are allowed too, so long as nothing expresses or implies that the review has to be positive. The line the Commission draws is about conditioning, not about generosity, and it gives worked examples of phrasing that crosses it.
The setting worth finding before anyone asks
Whether or not any of this amounts to breaking a rule, the practical question is whether anyone in your business knows what the automation is doing on your behalf.
Does your flow ask everybody, or only some people? That is one screen in whatever tool sends the request, and the answer is usually a default somebody accepted during setup.
Is there a fork keyed to a rating? If there is, find out what happens on the unhappy branch. Both branches reaching the same review links is the industry's own standard, and it is a five-minute check.
When does the request go out? Timing is a filter even when nobody thinks of it as one, and it is the one Trustpilot names and nobody else does.
Where did the rule you are following come from? Several vendors state the US federal position more strictly than the FTC does, and at least one platform states it more strictly than any vendor. Following your tool's blog post is not the same as following the policy of the site your reviews live on.
The uncomfortable part of all this is that gating works, in the narrow sense that it produces a better-looking rating. That is why the Endorsement Guides test asks whether the practice made the reviews "substantially more positive". A filter that changed nothing would not have been worth installing.
If you would rather know what your automations decide before somebody else notices, listing what runs, what it chooses, and where each choice came from is part of the process audit. $299, three business days, and a written map you keep whether or not you hire us.
Nothing here is legal advice, and all of the law described is United States federal law.
Related: why automations fail without telling anybody is the same shape of problem with a different symptom, and keeping a register of what is automated is how you find the settings screens in the first place.