Feed Is Noisy

Too many irrelevant or low-value matches—tune intent, exclusions, keywords, and thresholds before disabling listeners.

Noise usually means keywords are catching more than you meant, or AI scoring has not been told what to punish. Fix in this order.

1. Tighten Intent and Negative intent

Intent — Be explicit about the situations you want (buying signals, complaints, comparisons) so borderline posts score lower.
Negative intent — List recurring junk: jobs, promos, homework, unrelated industries, influencer spam, etc. Update when new noise patterns appear.

2. Add or expand exclude keywords

Exclude keywords are hard filters: if the phrase appears (exact match, per-row case rules), the post is dropped. Add strings you never want—often template sellers, recruiting boilerplate, or affiliate phrases. See Create listener (keywords section).

3. Narrow tracked keywords

Broad tokens (AI, marketing, founder) pull in huge unrelated threads. Prefer specific phrases aligned with pain and evaluation—see Keywords to track.

4. Matching options

Turn on whole word for short keywords that appear inside unrelated words.
Use case sensitive only when it genuinely reduces false positives without starving real mentions.

5. Authors and subreddits

Exclude authors who flood the listener with off-topic posts.
On Reddit, exclude subreddits that produce chronic noise for this listener.

6. Feed and notification thresholds

Sort by Most relevant while tuning so high-signal items surface first.
Raise notification relevance so Slack or Discord only pings on stronger matches (Per-listener notifications, Relevance threshold).

Not Getting Mentions — opposite problem: nothing arrives.
Too few high-relevance mentions — posts arrive but rarely score high enough to act on.