As of April 1, 2026, Google has instituted a new evaluation of review responses, enabling a rejection of review replies that do not abide by their content policies. This review reply status detail is now included in the GBP API as a ReviewReplyState, which will show as PENDING, ACCEPTED, or REJECTED depending on Google’s moderation.
You can find a list of recommendations from Google on replying to reviews here and their overall content policies here.
That said, subsequent to this change, there was an industry report released that identified over 10,000 replies rejected from a sample set and found six main patterns in why review replies would get removed. These are outlined below.
Duplicated Common AI-Phrases:
67% of all rejected replies in the dataset contained at least one obvious AI stock phrase, the most common of which are included below:
While the above may be concerning, you can take solace in the fact that the responses being AI-generated is not the cause for their removal. Instead, having the same phrases repeatedly pushed by AI looks like spam and demonstrates a lack of attention to responses. This is why the replies in this scenario were removed.
This change may feel extra demanding for multi-location businesses since managing responses for 500 locations can surface key points that get repeated. We understand that thanking someone for their kind words can be the easiest way to show your appreciation. However, Google's AI is now smarter than your current AI templates, but they can be updated to keep up.
Including variables like reviewer and location names is a good first step in customizing review replies past basic “thank you” templates. Ensuring your AI-review prompts are well-crafted to provide varied and customized responses is even better.
Profanity in Names, Business Names, and Menu Items:
The considerations for removals in this category are if a word or name in one language is profanity in another language, the review mentions an item from your business that has explicit content, such as a “Pornstar Martini”, or if your auto-response includes a user’s fake name that contains profanity or explicit language.
Exact Duplicate and Minimum-Content Replies:
Since exact duplication at scale is a textbook spam signal, responses that are the same across tens or hundreds of reviews will most likely get rejected. Even something simple like using the variable of the reviewer’s name in the reply would help shake off the spam signal, although we suggest customizing your prompts or replies to have more unique values than just the reviewer’s name mentioned.
Contact Details in the Reply Body:
Google sees including your support email in a public reply as commercial or off-platform solicitation, and it gets blocked. This applies to any “contact us” or “call us” language, included in a reply.
You may still point reviewers to your support lines, however, it’s recommended to direct them to where you already list this information on your listing or website, using phrases like, “Please reach out to us using the official contact info on our profile”. Although it may feel like it adds an unnecessary step at first, it is in fact directing people to engage more directly with your listing information while still following Google’s anti-spam guidelines.
Hashtags in Reply Text:
Google sees hashtags as SEO manipulation. Forcing additional business names and location keywords into a reply is read by Google’s moderation filter as manipulation, whether or not that is the intent.
To boost visibility to specific elements of your business that you may be trying to capture with hashtags, we instead recommend building out your structured and free-form services as well as uploading additional brand imagery with SEO-optimized file names. These data points can capture additions that may be unsuccessfully attempted through hashtags.
Bulk Scheduling and Reply Velocity:
When one account submits a large volume of replies all at once or within a short span of time and they all follow the same structure, Google sees that pattern as consistent with automated abuse.
That said, Google doesn't judge a reply in a vacuum. It tracks how fast you're posting, how many replies you've dropped recently, and if your messages all sound the same. It's hard to spot a bot just by reading one comment, but when you combine copy-pasted text with en-masse posting schedules, the automation pattern stands out.
While this reply status was only introduced in April of 2026, Google processed these reply status fields retroactively back to at least 2020. If you thought all your replies prior to this new status were untouched by this new level of moderation, that sadly may not be the case.
For example, you could be telling your clients or managers that you have a 100% response rate, while Google is quietly eliminating 10% of your work behind the scenes. This retroactive rejection also means that businesses who inexplicably lost a portion of their review replies see a decrease in their visibility and trust signals without a clear path to resolution.
What makes this new process especially difficult is that Google provides no notification when your reply is rejected. The reply appears to you to have been sent successfully while it may be stamped with a “REJECTED” state internally, and not shown publicly. The one sure fire way to see these statuses is accessing API-level visibility on the ReviewReplyState field.
The good news is that once you actually have visibility into the statuses, even if Google rejects your response, you are able to edit it – or start from scratch – and re-submit your reply.
Looking at the industry as a whole, this development is yet another confirmation that the “set it and forget it” approach is not only ineffective, but it can actually work directly against you.
Reputation management has been so focused on AI-automated review management that it has become too easy to just set up your agents and templates and let them run continuously. This Google update, however, proves that while they do want you to reply to your reviews, each review represents a unique customer experience and your reply should reflect that; review replies should be customized and authentic, not stock-applied.
Because of this analysis we were able to establish an evaluation to identify phrases that might trigger rejections so we can provide ongoing monitoring and make adjustments to stay ahead.
Despite finding low numbers for our client’s review response removals, we have established direct Google review response status transparency in the platform. When a reply to a Google review is submitted, we will be checking after 20 minutes, an hour, and then overnight to evaluate Google’s acceptance which will be reflected in the platform. With this new visibility you can confirm the transmission of the reply, then separately confirm Google’s moderation result as soon as that information becomes available.
On this front we also provide comprehensive reporting, allowing your team to see how many replies were submitted, published, pending, or rejected, then compare outcomes across individual replies, templates, and AI automations to quickly identify where changes are required. It's our hope to make "Reply Scorecards" available in the new Lia interface within the next two weeks.
