> ## Documentation Index > Fetch the complete documentation index at: https://docs.salestools.io/llms.txt > Use this file to discover all available pages before exploring further. # Prospect Cohorts: AI Clusters, Intent Signals, and Actions > Explore AI-clustered prospect groups across five views, read intent signals, refine by trait, and push the right cohort into an agent or sequence. Cohorts is your audience intelligence map. Sally groups every prospect you've gathered into named clusters based on what they actually share — region, industry, tech stack, hiring signals, buying role, and more — then scores each group so you can see at a glance which audiences are ready to act on now and which to keep nurturing. Instead of manually segmenting lists, you arrive at a board of ready-made, signal-rich audiences and go straight to deciding what to do with them. ## What you can do Each cohort block is sized by prospect count and coloured by intent level, so the biggest, warmest audiences jump out immediately. Switch between Treemap, Heatmap, Geo, Semantic, and List to examine the same cohorts from different analytical angles. Uncheck trait chips and buying-role buckets to narrow a cohort to exactly the slice you want before pushing it to an agent or sequence. Add prospects to My Data, hand them to an AI agent, launch a sequence, export to CSV, or open the full audience in the prospector. ## Reading intent colours and block size Every cohort block on the board communicates two things at a glance: how many prospects it holds and how ready they are to buy. Cohort colours follow a warm-to-cool spectrum that maps directly to purchase intent: | Colour | Intent level | What it means | | -------------- | ------------ | ------------------------------------------------------------- | | **Warm clay** | Act now | High buying signals — prioritise these cohorts first | | **Mid-warm** | Engage | Moderate intent — worth active outreach and nurturing | | **Cool slate** | Nurture | Lower near-term intent — keep warm with lower-cadence touches | Within each colour band, **darker shades signal higher priority** — a dark clay block outranks a light clay block. **Block size** shows the raw prospect count: bigger blocks hold more people. Start with the largest, darkest warm-clay blocks for the fastest path to pipeline. The summary line beneath the board title shows your totals: total cohorts, total prospects, and how many are **ready to act on**. A legend on the right side of the board reminds you what colours and shading mean at any time. ## Exploring cohorts across five views The default view. Every cohort appears as a block sized by prospect count and coloured by intent. The treemap gives you the fastest full-board overview — scan for the largest, warmest blocks and click any one to open its detail drawer. **Best for:** Getting an instant lay of the land across all cohorts. A facet grid that cross-references **Industry** against **Region** by default. Each cell is coloured by the combined intent of prospects at that intersection. Dense, warm cells show where your audience concentrates geographically and by vertical. **Best for:** Spotting which industry-region combinations have the most engaged prospects. A world map with a marker for each country where prospects are located. Marker size reflects audience volume in that country. Click a marker to see which cohorts have members there. **Best for:** Planning territory coverage or a city trip and visualising geographic reach. A proximity plot where cohorts that are genuinely similar to each other appear close together. Distance between bubbles reflects real audience similarity, not just shared labels — useful for spotting adjacent audiences you might have missed. **Best for:** Finding cohorts that are close enough to merge into one play, or discovering unexpected audience neighbours. When proximity data isn't available for a cohort, the Semantic view falls back to a priority-by-untapped plot. The axes still carry meaning, but distance between bubbles doesn't indicate similarity in that case. A sortable table showing every cohort with columns for intent, prospects, ICP fit, reachable, freshness, and channel. Sort by any column to rank cohorts however matters most to your current play. **Best for:** Comparing cohorts side-by-side on specific metrics before deciding where to focus. ## How to open and read a cohort Use the view toggle in the top right to switch between Treemap, Heatmap, Geo, Semantic, and List. Then click any block, cell, marker, bubble, or row to open its detail drawer on the right. The drawer works identically regardless of which view you opened it from. The top of the drawer shows a **radar fingerprint** — a spider chart that compares this cohort's profile across key dimensions against your overall audience. A cohort that skews strongly in one direction is more specialised; a balanced shape signals a broad, general-purpose audience. Below the radar, four tiles give you an at-a-glance scorecard: | Tile | What it measures | | ------------- | ----------------------------------------------------------- | | **ICP fit** | How closely this cohort matches your ideal customer profile | | **Reachable** | What percentage of prospects have workable contact data | | **Freshness** | How recently the data in this cohort was verified | | **Untapped** | How many prospects haven't been contacted by your team yet | Sally writes a plain-language **description** explaining who these people are and what makes them cohere as an audience. Below it, a **campaign angle** suggests a specific message frame or value proposition that fits this cohort's shared characteristics. A **best channel** recommendation tells you whether to prioritise email, LinkedIn, or phone for this audience. The **Why they cluster** section lists the shared traits Sally used to group these prospects — Region, Industry, Tech stack, Hiring tech, Hiring roles, Specialties, Size, and Titles. These chips are also what you edit when you refine the cohort in the next step. ## How to refine a cohort before acting A cohort starts with all its matching prospects included. Narrow it to exactly the right slice before you push it anywhere. Under **Why they cluster**, every shared-trait chip starts checked. Click a chip to **exclude** that value — the prospect count below updates live to drop anyone matching it. For example, uncheck a region chip to remove prospects in that geography from your intended action. The prospect grid splits the cohort's people into four buying-role buckets — **Decision Makers**, **Champions**, **Influencers**, and **End Users** — each with a live count. Uncheck a role to remove those people from the current set. As you exclude values, any trait chip that no longer has matching prospects is greyed out automatically. Re-include the exclusion that caused it to make that chip clickable again. The **Reset (N excluded)** control at the top of the section shows how many items are currently filtered out. A **Reset** control appears next to "Why they cluster" once you've excluded anything. Click it to re-check every chip and return to the full cohort. Closing and reopening the cohort drawer also resets all exclusions. ## How to act on a cohort All actions in the drawer footer apply to the **current filtered set** — the whole cohort when nothing is excluded, or your refined slice when chips and roles are unchecked. Click **Add to My Data** to import the cohort's prospects into your workspace. A confirmation dialog shows exactly how many people will be added. You can attach tags before confirming. Contacts land in **My Contacts** and their employers in **My Accounts**. Prospects without a recoverable contact record are reported as skipped in the confirmation summary. If the entire cohort is skipped, run a **Recompute** so the cohort's contacts can be freshly resolved. Click **Add to AI agent** to take this audience into the agent builder, where you can configure an outbound play tailored to the cohort's characteristics. Click the **rocket icon** to launch a sequence directly from the agent builder, or the **download icon** to export the cohort's prospects as a CSV file. Click **View all N prospects** to open the full audience in the prospector, where you can work the complete list with all standard prospecting tools. ## How to refresh your cohorts Click **Refresh** to re-fetch the current cohorts without rebuilding them. Use this after a recompute has finished to pull in the updated results. Click **Recompute** when you want Sally to rebuild every cohort from your latest prospect data. The AI re-embeds, re-clusters, re-labels, and re-scores your entire audience from the ground up. A recompute runs in the background and typically takes **10–20 minutes**, depending on how many prospects you have. You'll receive a confirmation when it starts. Come back and click **Refresh** once it's done to see your updated cohorts. Avoid navigating away mid-recompute — the job continues running server-side, but the in-progress indicator only shows on this page. ## Troubleshooting Cohorts haven't been built for your workspace yet. Click **Trigger first refresh** on that screen to start the pipeline. It usually takes 10–20 minutes — hit **Refresh** afterward to see your cohorts. The footer actions always target the current filtered set. Before clicking **Add to My Data**, confirm that your exclusions are still applied by checking the **Reset (N excluded)** indicator. If the count in the confirmation dialog matches the full cohort, no chips or roles are being excluded. Skipped prospects don't yet have an importable contact record. Run a **Recompute** so those contacts can be resolved in the next cohort build, then try adding again. Another exclusion has left that chip's value with no matching prospects. Re-include the exclusion that caused it — re-check its chip — and the greyed-out chips become interactive again. ## Related Work the full list of prospects inside any cohort using the standard prospector. Where cohort prospects land as contacts and accounts after you add them to My Data. Hand a cohort to an AI agent and build a full outbound play around it. Find warm relationship paths into the accounts that appear in your cohorts.