# How Fit Score, intent tags and sentiment work (https://snitchfeed.com/docs/fit-score) Every new mention a listener finds is analyzed by AI before it reaches your feeds. The analysis adds a short summary, a **Fit Score**, up to three **intent tags**, a **sentiment** score and the detected language. Scoring one mention costs 0.5 credits (see [Credits and usage](/docs/credits)). ## Fit Score [#fit-score] **Fit Score** answers one question: is this the kind of post you actually care about? It is shown as a percentage next to each mention. Hover it in the mention panel to see the in-app definition: a weighted score based on your Brand Context, listener intent and mention content. ### Fit Score bands [#fit-score-bands] | Band | Fit Score | | ---------- | ------------- | | **High** | 70% and above | | **Medium** | 50% to 69% | | **Low** | Below 50% | The **Fit score** filter on feeds, analytics reports and notifications uses these three bands. ### What drives the score [#what-drives-the-score] Fit Score has two parts. 1. **Relevance.** The AI reads the mention (and the author's headline or role on LinkedIn, when available) next to your workspace [Brand Context](/docs/ai-context/brand-context): company description, product summary and positioning, use cases, target audiences and ideal customer profiles. It also reads the listener's [intent signal](/docs/listeners/listener-intent), noisy signals and keywords. A mention that clearly matches the noisy signals scores very low even when it contains your keywords. 2. **Intent weights.** When you save a listener's intent signal, SnitchFeed gives every intent tag a weight from 0 to 5 for that listener, based on your intent signal and noisy signals. A tag you care about gets a high weight; a tag you want to avoid gets 0. When a mention has intent tags and the listener has weights, Fit Score is 60% relevance and 40% weighted intent tags. When a mention has no intent tags, Fit Score equals relevance. ### Where Fit Score is used [#where-fit-score-is-used] * **All mentions** opens with **Fit score** set to Medium and High and sorted by **Fit Score (high to low)**. See [Feeds](/docs/feeds). * [Opportunities](/docs/features/opportunities) lists only High and Medium mentions. * A feed's **Fit score** filter also decides which mentions its [notifications](/docs/notifications/per-feed) send. * The [Dashboard](/docs/features/dashboard) reports the share of Low mentions as **Noise Filtered**. ## Intent tags [#intent-tags] **Intent tags** are AI labels on each mention that describe what the author is doing. A mention gets up to three tags, or none when nothing clearly applies. Do not confuse them with the **intent signal**, which is the text you write on a listener. | Tag | Meaning | | -------------------------- | --------------------------------------------------------------------------------------------------- | | **Buying Intent** | The author is evaluating, researching, budgeting for or searching for a product, service or vendor. | | **Recommendation Request** | The author asks others to recommend tools, vendors or solutions. | | **Comparison** | The author compares products, tools or approaches to make a decision. | | **Seeking Alternative** | The author is unhappy with a current solution and wants to replace it. | | **Content Opportunity** | A question or discussion that could inspire educational, SEO or social content. | | **Pain Point** | The author describes frustration, blockers or inefficiency. | | **Competitor Complaint** | The author criticizes a competing product or company. | | **Competitor Mention** | A competitor is named or discussed, with or without a clear opinion. | | **Feature Request** | The author asks for missing functionality, integrations or improvements. | | **Praise** | The author gives positive feedback or an endorsement. | | **Hiring** | The mention is about hiring, recruiting or job seeking. | | **Brand Mention** | The mention names your own brand, product, company or domain. | | **Promotional** | The author is promoting or marketing a product, offer or themselves. | Use **Include Intents** and **Exclude Intents** in the [feed filters](/docs/feeds#filter-sidebar) to keep or hide tags, or break an [analytics report](/docs/features/analytics) down by **Intent tags**. ## Sentiment [#sentiment] **Sentiment** is the tone of the post, scored from 0 to 1 and grouped into three buckets: | Bucket | Sentiment score | | ------------ | --------------------- | | **Positive** | 0.6 and above | | **Neutral** | 0.4 to just below 0.6 | | **Negative** | Below 0.4 | Sentiment is separate from Fit Score. A competitor complaint can be negative and still be a High Fit Score lead. ## Tune scoring [#tune-scoring] When you change a listener's intent signal or noisy signals, SnitchFeed re-scores that listener's most recent mentions in the background. Each re-scored mention costs 0.5 credits, the same as a new one. * **Complete your Brand Context.** Scoring compares every mention against it. Gaps in your ICPs or use cases make scores less precise. See [Brand Context](/docs/ai-context/brand-context). * **Write a specific intent signal.** Describe the people and situations you want to find in 200 to 500 characters. See [Write an intent signal](/docs/listeners/listener-intent). * **Add noisy signals.** Describe the posts that keep showing up but are not useful, such as job posts or vendor promotions. * **Check weak keywords.** The Dashboard card **Worst Performing Terms** shows the keywords with the highest share of Low mentions over the last two weeks. * **Leave feedback.** **Show more like this** and **Show less like this** on a mention record your feedback with a reason. They do not change that mention's score or archive it. ## Related [#related] * [Feeds: filter, save and review mentions](/docs/feeds): filter by Fit Score, intent tags and sentiment. * [Write an intent signal for better Fit Scores](/docs/listeners/listener-intent): the listener field that calibrates scoring. * [Brand Context: how AI learns your business](/docs/ai-context/brand-context): the workspace profile every score is based on. * [Too few high Fit Score mentions: fixes](/docs/troubleshooting/too-few-high-relevance): what to change when High mentions are rare. * [MCP concepts](/docs/mcp/concepts): the same fields as AI assistants see them.