Dashboard Overview
The main dashboard (your workspace home page) shows today’s operational numbers at a glance:- Conversations today — How many conversations started today.
- New contacts — Contacts captured today.
- Satisfaction — Today’s average CSAT rating, with a health indicator (good / warning / poor).
- Needs attention — Conversations that are escalated, past an SLA target, or otherwise waiting on you.
- Identified rate — Share of today’s conversations tied to a known contact.
The Trends Page
The Analytics page (in the sidebar) is a tabbed reporting surface with two tabs: Trends and Agents. A shared filter row above both tabs controls the date range for everything below it.Date range and export
Pick a preset (7 days, 30 days, 90 days) or Custom to choose exact start/end dates. The range is clamped to your plan’s analytics retention window — see Analytics by Plan below — and a note appears under the charts when your requested range was shortened. An Export CSV button downloads the currently visible trend data.Volume
A single Daily volume chart plots conversations, escalations, new contacts, and blocked spam per day.Resolution
Two charts show who’s closing your conversations and how much of it is AI:- Resolutions by actor — conversations closed by AI, by a human agent, and by auto-close, per day.
- AI resolution rate — the share of closed conversations resolved by AI with no human involved (AI-resolved closes divided by total closes that day).
Quality
Three small charts: Resolution rate (share of closed conversations resolved), Satisfaction (CSAT) (average rating per day, on a −100% to +100% scale), and Feedback (thumbs up vs. thumbs down per day).Cost
A Token usage chart shows AI tokens consumed per day. ReplyBubble does not currently convert this to a dollar figure in the dashboard — use your AI provider’s own billing dashboard for actual spend, or export the CSV and apply your provider’s per-token rate. Daily numbers come from a nightly rollup that runs at 02:00 UTC, so today’s data on the chart is a live partial point covering only conversations, new contacts, and token usage — the rest of today’s metrics (resolution rate, CSAT, agent closes, etc.) appear after tonight’s rollup.Agents
The Agents tab is a leaderboard of your team’s inbox activity over the selected date range, sorted by conversations closed. Columns:
Spam-flagged conversations are excluded from every column on this leaderboard, matching the rest of the analytics surface.
Automatic Topic Detection
When visitors ask questions your bot can’t handle, those questions don’t just disappear into a log. ReplyBubble groups them for you.- Automatic clustering — Unanswered questions are sorted into topic groups automatically. Instead of reading through hundreds of individual messages, you see patterns like “shipping questions” or “pricing confusion.”
- Spot knowledge gaps at a glance — Topic clusters are ranked by frequency, so you can see which gaps are costing you the most conversations.
- Draft new answers with AI — Click into any topic cluster and ReplyBubble will draft a knowledge base answer for you using AI. Review, edit, and publish in one click.
- Review before publishing — Nothing goes live without your approval. You always have the final say on what your bot tells visitors.
Insights Workflow (Pro)
The Insights page gives you a deeper view of topic clusters with actionable tools. Overview metrics:- KB miss topics — questions your knowledge base couldn’t answer
- Success topics — questions answered well
- Failure topics — questions where the bot struggled
- Average resolution time
- Click Run Analysis to generate fresh topic clusters. Repeated clicks within 60 seconds of each other are debounced to prevent duplicate runs.
- Review the topic list. Each topic shows sample questions, conversation count, and status.
- Filter by status: New, Resolved, or Ignored.
- Click into a topic to see representative conversations and a suggested knowledge base answer.
- Click Publish to KB to add the drafted answer directly to your knowledge base.
- Mark topics as resolved or ignored once you’ve addressed them.
Success and Failure Patterns
Beyond individual topics, the Insights page identifies broader patterns:- Success patterns — The types of questions your bot handles well, with conversation counts. Use these to understand your strengths.
- Failure patterns — The types of questions where your bot struggles. These are your highest-priority knowledge base improvements.
Contact Intelligence
Every person who interacts with your chatbot gets a contact profile.- AI-generated summary — After a conversation tied to an outcome-producing form (contact, ticket, booking, or feedback) wraps up, ReplyBubble can attach a structured AI summary to the contact. What it contains depends on the outcome: sales inquiries get a qualification summary (service interest, timeline, budget, buying signals); support tickets get an issue summary with severity and sentiment; bookings get an intake summary (reason, urgency, notes); general conversations get an intent-and-topics summary with sentiment. Not every contact has one — only those linked to a qualifying conversation outcome.
- Sentiment tracking — Where the summary type includes it (support tickets and general conversation summaries), a sentiment badge shows on the contact’s timeline.
- Single timeline view — Everything about a contact lives in one place: conversations, form submissions, and any notes you’ve added, in reverse-chronological order. No jumping between tabs.
Analytics by Plan
Free plans include all the analytics features described above, limited to a 30-day lookback window. Pro extends that to a full year and includes REST API access for pulling metrics programmatically. Business adds unlimited history. See Billing & Plans for full plan details.
If you request a Trends date range that starts earlier than your plan allows, ReplyBubble silently shortens it to fit and shows a note under the charts explaining the adjustment — your request never returns more history than your plan includes, even via the API.
Tips for Getting the Most Out of Analytics
- Check the Insights page weekly. Its topic clusters (default-labeled “Unanswered Questions”) are the fastest way to improve your bot. Even adding two or three new answers a week makes a noticeable difference in resolution rates.
- Use topic clusters to prioritize. Don’t try to fill every gap at once. Start with the topics that come up most often — they’ll have the biggest impact.
- Watch the AI resolution rate trend, not just resolution rate. A rising resolution rate driven by auto-close rather than genuine AI resolutions can hide a knowledge base problem — compare the “Resolutions by actor” chart alongside it.
- Use the Token usage chart to plan model changes. If usage is climbing faster than conversation volume, try switching to a lighter model for simple questions and compare the trend afterward.