Clicks whisper.
Prompts yell.
Clickstream tells you where users went. It can't tell you what they were trying to do. Every prompt is signal. Unresolved intents write your roadmap.
Works on top of the Product Agent.
The dashboard: intent, resolution, and sentiment from every agent conversation.
The why behind the what.
Know what users want
Every prompt is stated intent — demand in your users' own words, classified and ranked. Not inferred from clicks.
Know if you delivered
Resolution, sentiment, and gaps for every conversation. See where the product wins today, and what to build next.
Know what it costs
Token spend by model, by action, down to the user. Control the economics of every answer.
No black box: every metric drills down to the actual conversations behind it.
"Tapping into intent data changed how we prioritize. We know exactly what users want in every conversation — and instantly close the loop with agentic experiences."
Understand agentic experience usage.
Usage, tokens, and cost for every action the agent takes — the same analytics you run on features, now on agentic workflows.
Read the demand. Build what's next.
Analyze conversations
Read what users actually ask the Agent, classified by User Intent and broken down by segment or use case.
Catch frustration early
User Sentiment flags the conversations that turned negative, so you see where users struggle before they give up.
Grow agent adoption
Know exactly who's using the agent and who isn't, with Audiences and Segments by plan, account, or activity.
See if it sticks
Active-user and retention trends show whether the agent keeps users coming back, or catch the early signs it doesn't.
Find your next feature
Agent Gap turns every request the product couldn't serve yet into ranked demand for net-new capability — split into what to build and what to document.
Increase user success
See which conversations resolved, replay the ones that didn't with Session Replay, and fix what's breaking.
Track account adoption
Segment agent usage by account to spot adoption gaps before they widen.
Prove the faster path
Same job, two ways to do it. Compare time-to-outcome when users ask the agent against the traditional UI path, measured end to end on your own product. It's the modernization case, made with your own data.
Catch and report issues
Unresolved and Escalated conversations get flagged automatically, and Session Replay shows what went wrong.
Every token, accounted for.
Every answer your agent gives has a price. Conversational Analytics shows you exactly where the spend goes — and gives you the levers to control it.
Cost by model
See token consumption and spend per model across every conversation. Route simple work to cheaper models in Studio, and watch the same job cost less.
Called most vs. costs most
Two rankings of your actions and tools: the ones users trigger constantly, and the ones that burn the budget. They're rarely the same list — and that's where the optimization lives.
Down to the user
Cost per user and per account, next to the value they got. Spot the outliers before the invoice does, and know exactly which accounts your agent spend is serving.
Ready to see Conversational Analytics in action?
Book a live walkthrough and see real conversations turned into intent, resolution, and gap signals.
Get a demoWhat teams do with intent data.
Decide what to build next
Action Gaps and Feature Requests rank real demand for capabilities that don't exist yet — new value users are literally asking for, so the roadmap follows evidence, not opinion.
Spot friction before it spreads
A rising Reported Issue line or falling Resolution Rate is an early warning you can act on this week.
Increase revenue
Find the high-intent use cases users succeed at and double down — new experiences, new workflows, new value that drives adoption and expansion.
Conversational Analytics provides the signal. Your team decides what to do with it.
Pair intent with behavior.
Conversational Analytics tells you what users want and why. Product Analytics tells you whether they got there — sign-ups, retention, feature adoption, and full session replay. Together they give full visibility into product adoption.
Explore Product AnalyticsCommon questions about Conversational Analytics
What does Conversational Analytics show?
Conversational Analytics reads every conversation users have with the Product Agent and labels it:
- Intent: Information Request, Action Request, Reported Issue.
- Outcome: Resolved, Unresolved or Escalated.
- Sentiment: Great, Good or Negative.
- Gaps: an action the agent didn't have, or knowledge it couldn't find.
It also tracks topics, action calls, tokens and estimated cost, and every metric drills down to the conversations behind it.
How do I find out what users are actually trying to do in my product?
Read what they ask for. Clicks show where users went; a prompt states what they wanted, in their own words. Foldspace classifies each conversation by intent (Information Request, Action Request, Reported Issue), tracks the mix over time, and splits it by segment or account. Across Foldspace deployments, 73 to 90% of user requests ask the agent to do something, not to answer a question.
How can user prompts inform the product roadmap?
When the agent can't deliver, the conversation gets a gap tag. An Agent Action Gap means users asked for something the agent can't do yet, so it goes on the build list. An Agent Knowledge Gap means the answer was missing or stale, so it goes on the writing list. Both lists are ranked and open to the requests behind them, so intent feeds the roadmap in users' own words.
How do I know if the agent delivered what users asked for?
Each conversation gets one outcome: Resolved, Unresolved or Escalated to a person. Resolution Rate tracks the resolved share over time, and handoffs show how often the agent passed a user to your team. Sentiment scores each conversation Great, Good or Negative, and a rising Negative share often moves before Resolution Rate does. When the numbers don't explain a conversation, session replay shows what the user did on screen.
What does each agent action cost?
Every action call records its estimated cost and tokens: input, output, cached and reasoning. Break the total down by action, model or provider to rank the actions called most against the ones that cost most. They're rarely the same list. On Enterprise, Quota Control caps the tokens each agent, user and conversation can consume, with optional warnings in the chat before anyone is blocked.
Does Foldspace only see what happens in the agent?
No. It sees both halves of the session: user-led and agent-led. Conversational Analytics reads what users ask the agent. Product Analytics, from the same SDK, captures what they do on their own: page views, sessions and the events you send with foldspace.track(). Both land in one dashboard, and session replay shows the conversation on the same timeline as the clicks.
What does Conversational Analytics need to run?
The Product Agent, live in your product. Conversational Analytics is available now and reads the conversations users have with the agent, so conversations and agent events flow in once the agent is embedded, with nothing else to configure. Call foldspace.identify() to tie conversations to named users and accounts you can segment, not anonymous sessions.
Stop guessing. Start reading intent.
Your users are telling you exactly what to build next and where to win. This is how you hear it.