Key concepts
The building blocks of BuzzGlean - sources (pages), alert rules, keywords, prompts, matches and alert channels.
Sources
A source is a page BuzzGlean watches: a Facebook group, a subreddit, a review site page (Trustpilot, G2, Capterra), an app store listing (App Store, Google Play), or a blog or forum page. Sources belong to your account and can be shared by any number of rules. For plan limits, each page counts once, no matter how many rules use it.
Pages from supported sites start right away. Pages from other sites are reviewed first, and private Facebook groups start once our team has joined them. Each source has a status, such as Active, In review or Can't access.
See Sources.
Alert rules
An alert rule ties everything together. It answers three questions:
| Step | Question | Settings |
|---|---|---|
| Rule | What should we look for? | Prompt, name, keywords, tags, sentiment |
| Sources | Where should we look? | Pages: communities, review sites, app stores, blogs and forums |
| Alert channels | Where should matches go? | One or more channels |
A rule is either active (matching posts and sending alerts) or paused (settings kept, nothing matched or sent).
See Alert rules.
Keywords
Keywords are an optional pre-filter. New posts are first filtered by your keywords, and only posts that contain at least one are checked by AI. Specific phrases (project management tool) keep alerts focused; very common words (help) make them noisy. BuzzGlean warns you when a keyword matches a large share of posts.
For plan limits, keywords are counted as the unique keywords across all your rules.
Prompt
The prompt describes a matching post in plain language, up to 300 characters. AI reads every post that passed the keyword filter and scores how well it fits the prompt.
See Writing good prompts.
Matches and relevance
When a post scores at or above the rule's threshold (0.7 by default), it becomes a match. Each match has:
- a relevance score from 0 to 100%, shown in the feed as a strong (80% and up), partial (50-79%) or weak match
- a short reason explaining why it matched
- up to three tags, such as Buy intent or Complaint
- a sentiment (positive, neutral or negative) when sentiment analysis is on
Posts that contained your keywords but scored below the threshold are filtered out. The dashboard shows how many posts AI filtered for you.
Alert channels
An alert channel is a destination for matches: an email inbox, a Slack channel, a Telegram chat or your own HTTPS endpoint. A rule can deliver to several channels, and one channel can serve several rules.
See Alert channels.
Feed
The feed lists every match from all your rules, newest first, with filters for rule, source, relevance, tags and sentiment.
See Feed.
Credits
Comment credits pay for AI-drafted replies from the comment assistant. Your plan includes a daily allowance that resets every day.