Troubleshoot SnitchFeed: missing, low-score or noisy mentions
Fix a listener that finds no mentions, mentions that rarely score high, or a feed full of junk, with a short checklist for each problem.
Last updated
Most problems fall into one of three groups. Pick the one that matches what you see, then work down its checklist in order.
| What you see | Go to |
|---|---|
| Nothing arrives, or far less than expected | No mentions |
| Mentions arrive but few score high, or alerts rarely fire | Few high Fit Score mentions |
| Plenty arrives, but too much is junk | Noisy feed |
No mentions
- The listener is paused. Open Listeners and select it. If Paused is highlighted next to Save, click Active. Posts published while a listener is paused are not collected later.
- You are out of credits. Scans and real-time matching stop when credits run out. Check the Credits tab under Workspace Settings → Plan & Billing, then top up or wait for your plan to renew. See Credits.
- The first scan has not run yet. Reddit and Bluesky match posts as they appear. X, LinkedIn and Hacker News are scanned on a schedule: by default every 4, 12 and 6 hours. Give a new listener until its next scan. See Scans & backfill.
- The feed is hiding them. A listener's own feed shows only Medium and High Fit Score. Add Low to the Fit score filter, switch Sort to Newest, and clear other filters to see everything collected.
- A match setting is too strict. Open the listener's Keywords section and check each term:
- Case sensitive or Whole word is on when it doesn't need to be.
- Must also include needs every word to appear. Move words to Must mention at least one of if any one is enough.
- Skip posts that mention any of these words contains a word that also appears in good posts.
- A platform you expected is turned off.
- X (Twitter) Settings or LinkedIn Settings filters, such as minimum likes or job title, rule out most posts.
- A warning icon sits next to a term. Hover it for the reason, then open the term to fix it.
- Author or Reddit filters limit matches. Only from these authors ignores everyone not listed. Only these subreddits limits Reddit to those communities, and names are case sensitive (
SaaS, notsaas). Also check that Search Posts and Search Comments are on. - The keywords are too rare. Very specific phrases or unusual spellings may have no recent posts. Widen the term or add synonyms. See Keywords.
- A tracked profile shows a red chip. SnitchFeed retries automatically while the listener is active. If it stays red, remove it and paste the profile URL again. See Track profiles.
Few high Fit Score mentions
High means a Fit Score of 70% or more.
- The feed is sorted by date. With Sort set to Newest, older high-scoring mentions sit below newer weak ones. Switch to Fit Score (high to low).
- The intent signal is vague. The intent signal tells scoring what a good post looks like. Name who you want, which situations matter (complaints, comparisons, buying questions) and what to prioritize. Keep it within 200 to 500 characters, then click Update Intent to re-score recent mentions.
- Noisy signals are too broad. Noisy signals lower scores. A broad entry like "anything promotional" can drag down good posts. List only specific patterns you keep seeing.
- Brand Context is thin or out of date. Brand Context tells SnitchFeed what you sell and who buys it. Update it under Workspace Settings → Brand Context.
- Your keywords collect low-intent posts. If most posts are tutorials, news or passing mentions, there is little to score high. Add switching or comparison words under Must mention at least one of.
- The alerting feed only allows High. Alerts follow a feed's filters. Add Medium (50% to 69%) to that feed's Fit score filter and click Update. See Notifications.
- Other filters hide high-scoring mentions. Check Sentiment, Date Posted, listener, platform and intent filters. Clear them for a moment to see what scores high.
Noisy feed
Work top to bottom. The first fixes change every future mention; the last only change what you see.
- Scoring doesn't know what to score down. Tighten the intent signal and list the junk you keep seeing, such as job posts or promotions, in noisy signals. Click Update Intent.
- Unwanted posts share a word. Add it to Skip posts that mention any of these words on the term, for example
hiring,internshiporgiveaway. Matching posts are dropped before scoring. - Terms are too broad. Words like
AI,marketingorfounderpull in unrelated threads. Use specific phrases, require context with Must mention at least one of or Must also include, and turn off platforms a term doesn't need. See Keywords. - A short term matches inside other words. Turn on Whole word. Turn on Case sensitive only when capitalization separates your target from noise.
- A few accounts or subreddits flood the listener. Add them to Never from these authors or Exclude these subreddits. Their existing mentions are archived when you save. If Reddit comments are the main noise, turn off Search Comments.
- The feed shows too much. Set the Fit score filter to High, and use Exclude Intents and Exclude Keywords, then click Update. Alerts follow the feed's filters, so this also means fewer alerts. See Feeds.
Still stuck? Email hello@snitchfeed.com with the listener name.
Next steps
- Intent & Brand Context: write an intent signal that scores well.
- Advanced search: exactly how each keyword field matches.
- Fit Score: what the score bands mean.
Credits
See what each SnitchFeed action costs in credits, from X, LinkedIn and Hacker News scans to AI scoring and replies, and estimate your monthly use.
FAQ
Answers to common SnitchFeed questions: the free trial, supported platforms, finding mentions without tags, compliance, Slack, teammates and cancelling.