Listener Intent

SnitchFeed scores matches with AI using your intent signal (what you want to find) and noisy signals (what to score down). The clearer your instructions, the less noise—fewer job posts, promos, and off-topic threads—and the more actionable your feed and alerts.

This page uses the same running example as Keywords to track: you sell a Notion competitor and want people complaining about Notion or asking for alternatives—so you can join the thread while Fit Score stays high.

Intent signal and noisy signals

In the listener dialog you’ll see two fields that work together:

Intent signal

Intent signal is a short, plain-language brief of the posts you want this listener to surface. It:

• steers Fit Score so keyword hits that match your goal rank higher, and
• powers keyword suggestions when you click generate (the model needs a strong Intent signal to be useful).

Be sure to be as descriptive as possible. The product enforces 150–500 characters (after trimming), matching Noisy signals.

Example (Notion competitor)

I want posts from startup founders, ops leads, and marketers who are frustrated with Notion or shopping for a replacement: complaints about sync, mobile, permissions, billing, or enterprise limits; threads asking what to use instead of Notion; and comparisons with other docs or wiki tools. Prioritize switching intent and real workflow pain—not influencer aesthetic drops or template giveaways unless someone is clearly evaluating tools.

Write in first person or imperative (“I want…”, “Find…”), name who (roles), what situations (complaints, comparisons, migrations), and what to prioritize vs skip. Vague intent signals (“anything about productivity”) produce vague scores and weak keyword suggestions.

Noisy signals

Filling out Noisy signals is required for every listener. It uses the same character limits as intent signal (150–500 characters after trimming). Describe what should score with a low Fit Score: threads that technically match your keywords but are not worth your time.

Quick patterns many teams list here: job listings, affiliate or promo spam, student homework, generic tutorials when you only want buyer conversations, or industries that noise up a broad keyword.

Common noisy signal examples (use whatever fits your listener):

• Job posts — open roles, “we’re hiring,” and company recruiting threads that match your keywords but are not buyer conversations.
• Hiring posts — hiring managers or founders posting roles rather than discussing the product category as users.
• People looking for jobs — resume shares, “DM me for referrals,” job-search updates, and networking-for-employment posts.
• Students — coursework, class projects, homework help, and academic discussion that overlaps with your keywords.
• Certification brags — “I passed X” / “finished Y certification” posts with no evaluation, purchase, or workflow pain angle.

Example (Notion competitor)

Job listings and recruiting (including at Notion or unrelated roles); Notion template sellers, marketplace spam, and affiliate link dumps; pure fan posts with no complaint or purchase angle; student homework and shallow tutorials; giveaways and growth hacks with no evaluation; crypto, NFT, or unrelated hustle threads; generic motivational posts with no mention of tools, switching, or problems with Notion.

List categories of junk you keep seeing in the feed so SnitchFeed learns to score them down—even when keywords still match.

Per-listener fields

Intent signal and Noisy signals each require 150–500 characters (after trimming). They are listener-specific: you set them in each listener’s create/edit dialog (Create listener). Different listeners can target different competitors or motions while sharing the same workspace Brand Context.

Together, Intent signal and noisy signals narrow the gap between “matched keywords” and “matches worth replying to.” When you change either field, SnitchFeed may re-score recent mentions so the feed reflects the new guidance.

For workspace-wide company facts (who you are, what you sell), keep Brand Context accurate—Intent signal still tells the model what to listen for.