Ad comment glossary
Short definitions of the terms that come up when you moderate and answer the comments under Facebook, Instagram and TikTok ads. Each one links to a guide or post that goes deeper.
Ad comment
An ad comment is a comment left on a paid ad. On Facebook it belongs to the ad rather than the Page timeline, so you read and answer it in Ads Manager, Meta Business Suite or a tool connected through Meta's Graph API.
Dark post
A dark post is an ad created in Ads Manager that is never published to the Page timeline. Its comments exist only on the ad, which is why a Page can look quiet while its ads are busy.
Boosted post
A boosted post is a Page post promoted as an ad. Because it started as a Page post, its comments also show on the Page, unlike the comments on a dark post.
Moderation Assist
Moderation Assist is Meta's Facebook tool, in the Page's Professional dashboard, that hides comments matching criteria such as links, images, keywords, profanity or accounts younger than a week, on your Page and the ads you run. It doesn't go back over comments already posted, and it never replies.
Hide unwanted comments (Instagram)
Hide unwanted comments is an Instagram setting, on by default, that hides comments that may be spam or offensive. You can raise it to Some or More under How others can interact with you, Comments.
Banned words (keyword filter)
A banned-word list hides or deletes comments that contain words you choose, before anything else looks at them. In RYPL, scam matches words that start with it, *scam* matches it anywhere, and "scam" matches only the whole word.
Comment moderation
Comment moderation is deciding what happens to each comment under a post or ad: answer it, hide or delete it, or hand it to a person. Done by hand, rules or AI, it keeps spam and abuse from becoming the first thing a buyer reads.
Spam comment
A spam comment is a link, self-promotion, competitor redirect or bot text posted under a post or ad. In RYPL's Q3 2026 data, under 1% of the comments its AI analyzed were spam, after advertisers' banned-word lists had removed about 5% of all comments.
Scam accusation
A scam accusation is a comment that calls the brand a scam or the product fake. In RYPL's Q3 2026 data, about 1 in 40 comments (3%) did. Real complaints need a public answer; false accusations can be hidden.
Buying intent
Buying intent is a comment that asks about the product (price, availability, specs, how to get it) or says the person wants to buy. In RYPL's Q3 2026 data, 38% of the comments its AI analyzed showed buying intent, and only 2% said outright that they wanted to buy.
AI auto-reply
An AI auto-reply is a public reply written for the specific comment from the brand's own information, rather than a template sent whatever the person asked. Meta's native tools hide comments but don't write public replies; API-connected tools do.
Comment-to-DM automation
Comment-to-DM automation sends a private message when someone comments a trigger word; ManyChat is the best-known example. It answers in the inbox, and any public reply it posts is a line written in advance rather than an answer to what was asked.
Response time
Response time is how long a comment waits for the brand's reply. Edison Research found that 42% of consumers who complain on social media expect a reply within 60 minutes.
Engagement rate ranking
Engagement rate ranking is one of Meta's ad relevance diagnostics. It compares your ad's expected engagement, including clicks, reactions, comments and shares, with ads competing for the same audience.
Quality ranking
Quality ranking is one of Meta's ad relevance diagnostics. It draws on feedback such as people hiding or reporting the ad, and on the post-click experience, compared with ads competing for the same audience.
Comment controls
Comment controls is an Ads Manager setting Meta introduced in November 2024 for select businesses, with a Turn off comments option at the ad level. Many accounts still don't see it, and it depends on the objective and placement.
How to turn off comments on Facebook ads