Insights

Insights turns every campaign interaction -- email replies and call transcripts -- into actionable intelligence. When a prospect responds to one of your emails or joins a call, the system automatically analyzes the interaction, aggregates the results across your campaigns, and turns what it learns into concrete suggestions that improve your AI agents, refine your personas, and shape your company strategy.

How It Works

The insights system operates in four stages:

  1. Analysis -- every incoming reply is classified by AI across 9 dimensions. Every call transcript is analyzed with additional call-specific dimensions like buying signals and competitive mentions.
  2. Aggregation -- individual analyses are rolled up into quantitative breakdowns and AI-synthesized themes at multiple scopes (company, product, agent, campaign).
  3. Agentic synthesis -- an AI agent uses tools to search, verify, and cluster patterns across all your email and call data into qualitative intelligence reports.
  4. Suggestions -- synthesized intelligence drives concrete, actionable suggestions for your AI agent configurations, persona targeting, and company strategy.

The Insights Dashboard

Go to Insights in your dashboard sidebar to see your intelligence. At the top, a scope selector lets you choose what level of data to view. The selector shows how many replies and calls have been analyzed at each scope.

The dashboard has two main tabs: Replies and Market Intelligence. The Replies tab shows individual reply analyses in a sortable table, and the Market Intelligence tab shows the AI-synthesized qualitative report. A resizable analytics sidebar on the right shows quantitative charts and call data.

Scope Selector — Switch between company-wide, product, agent, or campaign-level views. Reply counts update in real time as new analyses come in.

Scope Levels

Insights aggregate at four levels:

  • Company -- everything across all campaigns, agents, and products. Your overall market reception at a glance.
  • Product -- filter by a specific product to see how prospects respond when you pitch it. Useful for comparing reception across your product line.
  • Agent -- filter by a specific AI Sales Rep to evaluate which agent's tone and approach gets the best results.
  • Campaign -- drill into a single campaign to see exactly what's working and what isn't for that outreach effort.

Automatic Reply Analysis

When a prospect replies to any campaign email, the system analyzes their response across 9 dimensions:

9 Dimensions Extracted Per Reply
SentimentPositive
IntentMeeting Request
Interest level5/5
ObjectionsNone
QuestionsPricing tiers?
TopicsAPI, integrations
What workedTechnical specificity
What didn't
SummaryStrong interest, wants demo
Recent Analyses
positive

Prospect expressed strong interest in the API integration capabilities and asked about enterprise pricing tiers.

InterestedInterest: 4/5initial
Feb 18, 2026
negative

Replied with "We already use a competitor for this." Mentioned they signed a 2-year contract last quarter.

Not InterestedInterest: 1/5initial
Feb 17, 2026
mixed

Interested in the concept but raised concerns about data privacy compliance for EU customers. Asked for a SOC 2 report.

QuestionInterest: 3/5follow-up
Feb 17, 2026
positive

Forwarded the email to their VP of Engineering and asked to set up a demo call next week.

Meeting RequestInterest: 5/5initial
Feb 16, 2026

Reply Analysis — Every reply is automatically classified across 9 dimensions. Recent analyses appear in a live feed with sentiment badges, intent labels, and interest scores.

DimensionWhat it captures
SentimentPositive, negative, neutral, or mixed
IntentInterested, not interested, objection, question, meeting request, referral, out of office, bounce, unsubscribe, or other
Interest level1-5 score of how engaged the prospect is
Objections raisedSpecific pushback extracted from the reply
Questions askedWhat information the prospect is seeking
Topics mentionedKey themes and subjects in the reply
What workedWhat about your original email resonated
What didn't workWhat fell flat or triggered a negative response
SummaryAI-generated 1-2 sentence summary of the reply

This happens automatically in the background. You don't need to configure anything -- just connect your email and start a campaign.

Call Transcript Analysis

When a prospect books a call and it takes place on Google Meet, the system fetches the transcript and analyzes it with AI. Call analysis includes all 9 dimensions from reply analysis plus additional call-specific dimensions that only surface in live conversation:

Call-Specific DimensionWhat it captures
Buying signalsIndicators of purchase intent -- budget mentions, timeline discussions, stakeholder involvement
Competitive mentionsWhich competitors came up, what was said about them, and how your offering compared
Prospect talk ratioHow much the prospect spoke vs. listened -- a strong indicator of engagement
Key momentsCritical turning points in the conversation with timestamps: objections overcome, interest spikes, concerns raised
Next stepsWhat was agreed upon: follow-up meetings, demos, trials, or introductions to decision-makers
Pricing discussedWhether pricing came up and how the prospect reacted

Call transcripts are linked to campaign recipients through booking tracking, so the system knows which campaign and agent the call relates to. Call analyses merge with reply analyses in the aggregation -- both count toward the scope totals and feed into qualitative synthesis.

For more on how calls are connected, see the Calendar & Calls documentation.

Quantitative Insights

A compact summary strip at the top of the dashboard shows key metrics for the current scope: total replies, average interest level, and positive rate with a sentiment breakdown bar.

Total Replies

142

Interest Level

3.4/5

Positive Rate

42%

Top Intent

Interested

Sentiment Distribution

How recipients feel about your outreach

Positive: 42
Negative: 28
Neutral: 18
Mixed: 12

Reply Intent Breakdown

What recipients are saying

Interested
34
Question
22
Objection
18
Meeting Request
12
Not Interested
8
Referral
4
Out of Office
2

Quantitative Insights — Stats cards, sentiment donut chart, and intent bar chart update in real time as new replies are analyzed. All data is scoped to your current selection.

Analytics Sidebar

The right side of the dashboard includes detailed quantitative analytics in a resizable panel:

  • Sentiment distribution -- a donut chart showing the breakdown of positive, negative, neutral, and mixed replies. Hover over segments to see exact counts.
  • Intent breakdown -- a horizontal bar chart showing the volume of each reply type: interested, not interested, objection, question, meeting request, etc. Bounces are shown separately.
  • Reply timing -- a distribution chart showing when replies arrive relative to when emails were sent, helping you understand response patterns
  • Booking timing -- a distribution chart showing when call bookings happen relative to outreach, helping you gauge time-to-meeting
  • Recent call analyses -- the latest analyzed calls with buying signals, competitive mentions, and next steps

AI Market Intelligence

Beyond the quantitative charts, the system generates qualitative, AI-powered analysis of your reply and call data. Click Generate Insights (or Regenerate if insights already exist) to produce a full AI synthesis.

The synthesis is produced by an agentic AI that uses tools in a loop -- searching across your data with vector embeddings, verifying patterns, and pulling real quotes -- to produce a thorough, evidence-based report.

AI Market Intelligence

Synthesized from 142 replies — last updated Feb 18, 2026New replies available

Overall reception is moderately positive (42% positive sentiment), with the strongest engagement coming from mid-market SaaS companies. The primary barrier to conversion is existing competitor contracts -- prospects are interested but locked in. Messaging that references specific company achievements significantly outperforms generic value propositions. Compliance concerns (SOC 2, GDPR) are a growing theme and should be addressed proactively in initial outreach.

AI Market Intelligence — Click through the tabs to explore the executive summary, objection and question themes, messaging patterns, and market signals. The amber indicator shows when new data is available for regeneration.

The qualitative analysis includes:

Executive Summary

A 2-4 sentence, evidence-based summary of the most important takeaways from your data. This is designed to be actionable -- not just a restatement of the numbers.

Objection Themes

Raw objections from individual replies and calls are clustered into semantic themes. Each theme includes:

  • Theme name and description
  • Frequency (high, medium, low)
  • Reply count -- how many replies and calls contributed to this theme
  • Example quotes -- real excerpts from prospect replies and call transcripts

Question Themes

Similar to objection themes, but focused on what information prospects are asking for. Useful for identifying gaps in your messaging or knowledge base.

What Resonates

Messaging patterns that generate positive engagement. Each includes a pattern description, an explanation of why it works, and a confidence level.

What Falls Flat

The inverse -- patterns that trigger negative responses or disengagement. Each includes the pattern, an explanation, and a confidence level.

Market Signals

Non-obvious cross-cutting patterns that emerge from the data: differences between initial vs. follow-up responses, emerging trends, correlations, and other insights you wouldn't spot manually.

Call-Specific Insights

Patterns that are visible only in call conversations -- not in email replies. These include buying signal trends, common competitive mentions across calls, how prospect engagement (talk ratio) correlates with deal progression, and key moments that recur across multiple calls.

Email vs. Call Divergences

Where email sentiment or objections differ meaningfully from what prospects say on calls. For example, prospects may raise pricing objections frequently in email but rarely on calls -- suggesting pricing is used as an email brush-off rather than a genuine concern. These divergences help you tailor your approach differently for each channel.

Data Source Management

The insights dashboard shows all the individual reply analyses and call analyses that feed into the qualitative synthesis. Each data source can be individually excluded from or included in the next regeneration cycle using a toggle.

This is useful when:

  • A reply is spam or irrelevant and would skew the analysis
  • You want to see how insights change when certain data points are removed
  • A call transcript was from an internal meeting rather than a prospect conversation

Excluded items are still stored -- they're just not included in the next synthesis run.

Staleness Detection

The system tracks whether your qualitative insights are up to date. When 5 or more new replies or call analyses have been recorded since the last synthesis, you'll see a "New data available" indicator. Click Regenerate to produce fresh insights that incorporate the latest data.

Recent Analyses

The Replies tab shows the latest classified interactions in a sortable table, split into two sub-tabs: Replies (human responses) and Bounces (bounced/out-of-office). Each entry displays:

  • The sentiment badge (color-coded)
  • An AI-generated summary of the reply or call
  • Badges for intent, interest level, and whether the reply was to an initial email or follow-up
  • Reply time relative to the original email (color-coded: green for under 4 hours, amber for under 24 hours)
  • A "Booked" indicator if the reply led to a meeting
  • The date it was analyzed

Click any row to open a detail panel showing the full reply text, sender information, extracted objections, questions, what worked, what didn't, and other metadata.

Both email replies and call analyses appear in this feed, giving you a complete picture of incoming interactions without leaving the insights page.

Reanalyze All Replies

If you need to re-run the AI classification on all existing replies (for example, after the analysis model has been improved), click "Reanalyze All Replies" at the top of the page. This queues every reply in the current scope for re-analysis.

The Insights-to-Action Loop

This is what makes insights more than just a dashboard. Aggregated intelligence from both email replies and call transcripts doesn't just sit in charts -- it drives concrete suggestions across three parts of your system:

1. AI Agent Field Suggestions

Market intelligence is analyzed and turned into specific, field-level suggestions for your AI Sales Reps. The system can suggest improvements to:

  • Outreach email instructions -- how the agent writes cold emails
  • Reply instructions -- how the agent handles inbound replies
  • Follow-up sequence instructions -- per-step follow-up strategy
  • Objection responses -- pre-approved responses to prospect pushback
  • Never-say rules -- forbidden phrases and topics

Each suggestion includes reasoning, intelligence highlights (the data points that triggered it), an impact level, and a before/after diff. You accept or dismiss each suggestion individually. Learn more in the AI Sales Reps documentation.

2. Persona Suggestions

Intelligence drives suggestions to improve your persona targeting:

  • Refine existing personas -- adjust descriptions, pain points, or characteristics based on what's working
  • Improve search queries -- tighten targeting based on which prospect profiles convert
  • Discover new personas -- identify entirely new audience segments from patterns in your reply data
  • Deprecate underperformers -- flag personas that consistently produce low engagement

Learn more in the Persona Management documentation.

3. Company Strategy Suggestions

At the highest level, intelligence is synthesized into strategic directions for your business:

  • Positioning adjustments based on how the market perceives you
  • Product fit signals indicating where your product resonates most
  • Messaging refinements based on what language converts
  • Geographic opportunities revealed by regional response patterns
  • Feature priorities surfaced from prospect questions and objections

These appear on your Company strategy page. Learn more in the Company documentation.

The Compounding Effect

The effect compounds over time. Campaign 1 generates replies and calls. Those become insights. Insights generate suggestions that improve your agents, personas, and strategy. Campaign 2 runs with better targeting, better emails, and better reply handling. Its data compounds on top. The longer you use it, the smarter everything gets.

Next Steps

  • Learn about AI Sales Reps to understand how agents use insights and receive suggestions
  • Set up Campaigns to start collecting reply data
  • Read about Company to see company strategy suggestions and persona intelligence
  • See Calendar & Calls to understand how call transcripts feed into insights