To configure auto publish and review moderation settings in Reviewbird, go to the Reviewbird Dashboard > Reviews > Moderation.
Auto publish reviews
The auto publish reviews setting allows you to configure what happens by default when a new review is submitted. There are two basic options:
- Publish new reviews automatically
- Publish reviews manually
We recommend using automatic publishing unless you have a particular reason to be wary of automatic publishing.
By default, all reviews will be published automatically but you can configure which reviews qualify.
Auto publish rules
Auto publish rules allow you to configure which reviews will be published automatically. There are two configuration options:
Automatically approve reviews from verified purchasers
When this option is selected, Reviewbird will check whether the review submitter’s email address matches a purchase for the product they are reviewing. If it is determined the reviewer is a verified purchaser, the review will be published automatically.
Automatically publish reviews with at least X stars
If you’d like to manually moderate low-rated reviews, you can use this option to set a star threshold for which reviews get automatically published.
Who can write reviews?
By default, reviews are open to anyone who visits the site. However there are legitimate reasons why you might want to make reviewing available upon invitation only, such as to reduce spam or malicious reviews.
You can either configure reviewing to be Open to everyone or Invite-only:
Moderation settings
In addition to the configuration above, Moderation applies additional checks to catch poor quality, spam, offensive, or other unwanted reviews.
Your Plan determines which options you have access to.
Standard Moderation
Standard Moderation is available to Starter plans and higher.
Standard Moderation features include:
- NSFW content filtering
- PII (Personally Identifiable Information) detection and censoring
NSFW content filtering
NSFW content filtering automatically detects offensive words in most languages. You can configure whether reviews with NSFW content are automatically rejected, flagged, or censored (****)
There are also options for setting a how strict the NSFW filter is, as well as adding custom words to the ban list. This could be useful in a variety of situations, or where you can’t allow customers to mention certain trademarks that might expose you to additional risk.
PII Detection (Personally Identifiable Information)
Reviewbird can scan reviews for common instances of PII (Personally Identifiable Information).
We currently detect the following types of PII:
- Email addresses
- Phone numbers
- Physical addressses
- Credit card numbers
- Social security numbers
- IP addresses
- UUIDs
When PII is detected, you can configure how to handle it. By default, any detected PII will be censored with asterisks (***)
But you can also flag these reviews for manual review, or automatically reject them altogether.
Review Guard
Review Guard is our AI moderation suite. It is available to Plus and Everything plans. Only Everything plan stores are able to customize Review Guard’s default settings.
Review Guard provides two additional types of moderation.
AI Moderation
AI Moderation runs each review through an LLM to determine if the content is legitimate. The types of things we look for are:
- Off topic reviews (wrong product, or not a review)
- Offensive reviews (hate speech, threats, etc)
- SPAM reviews
- Promotional material,
- External links
- Personal attacks (such as making offensive statements about a person)
- Etc
If you have an Everything plan you can customize specifically which types of reviews AI Moderation screens for, or you can disable the AI Moderator entirely.
Media Moderation
Media Moderation screens reviewer photos and videos for offensive or inappropriate content. By default, it will detect and block images or videos with:
- Nudity / strong sexuality
- Gore
- Violence
- Self-harm
- Other offensive content
If you have the Everything plan, you can turn off media moderation or configure specifically how it works:
