Every interaction with your AI agent is a measurable signal — but if you don't measure it in the moment, it's lost the second the conversation ends. Pulse, Delto's native module, configures and schedules surveys inside the same platform (when the conversation closes, when a skill ends, or after a chat with a human agent), with formats like NPS, stars, thumbs up/down or open text. It also reads “between the lines”: tone, overall satisfaction and tags, even when the customer doesn't complete a survey. Results flow into Analytics and Insights (chat with your data in natural language) and export to your other systems.
Every interaction with a customer is an opportunity to understand what works, what doesn't and to what degree. In banking, where an unresolved or poorly resolved query directly hits trust, that information is too valuable to let slip. The problem is that, if you don't measure it in the moment, it's lost the second the conversation ends: you don't know whether your customer was satisfied, where they hit friction, or what needs went unresolved. You can't improve what you don't measure. Measuring every interaction is what lets you go from operating blind to making decisions on real evidence, catching friction early and improving your agent continuously. Surveys that are already built in Knowing what your customers think shouldn't depend on an external tool or on integrations you wire up yourself. That's why at Delto we offer Pulse , the native module to configure, customize and schedule surveys for your end customers directly inside the platform. Pulse lives in the same place where agents are designed and deployed, and connects directly to every conversation. So you measure your customer at the moments that matter, without adding friction to their experience and without depending on integrations to get started. And although it all originates and lives inside the platform, it isn't isolated: responses can be exported and pushed automatically to the system you already use for other channels, centralizing all your feedback in one place. What your customer says through the conversational agent sits alongside the rest of the voice of the customer you already have from other digital channels. The platform ships with three survey types ready to use: At the end of the conversation with the virtual agent, to understand how it went for the customer and whether they resolved what they came for. At the end of each skill , usually shorter and often a single question, to know specifically whether that query was resolved. At the end of the chat with a human agent , to measure the quality of assisted support. You can also configure any additional surveys you need, tailored to what you want to measure and the exact moment you want to do it. Measuring beyond what the customer says Not every question is after the same thing, so you can design them in the format that best captures each one: an NPS to measure loyalty, a star rating or a thumbs up/down for a quick, frictionless read, single or multiple choice to narrow options, or a short open answer when you want to hear your customer in their own words. But the ability to measure isn't limited to what the customer answers explicitly. Precisely because feedback is monitored natively in every conversation, the platform also reads between the lines: you can measure the customer's tone throughout the interaction, their overall satisfaction , and apply tags that fit your operation to classify and understand each case. So you get feedback even when the customer doesn't complete a survey. From data to conclusions Capturing interaction feedback is only half the process. With the responses and the data in Analytics , you can see the full table of results and track key metrics on a dashboard. And you can go a step further with Insights , our feature to chat with an agent and ask questions of your own data in natural language, for more complex analysis or to reach conclusions that would otherwise take hours. Real situation With Pulse Without measurement A customer abandons a task halfway through You detect the exact skill where it happened and fix it You only see that completed tasks dropped, without knowing why A change in the bot worsens the experience You notice it in that skill's CSAT within days You find out weeks later, from the rise in complaints You want to know what customers think of the new loan flow You filter that skill's feedback and get the answer instantly You build a survey on the side and wait to gather responses A customer is upset but doesn't complete the survey Tone of voice and tags register the signal anyway That case is lost without a trace Leadership asks for a satisfaction report You pull it from the dashboard or ask Insights in natural language You cross-reference spreadsheets and loose data by hand The result is each customer's feedback measured and analyzed in every conversation , centralized alongside the rest of your channels and ready to become decisions. In a business where trust is everything, listening to your customer can't be a separate process or an end-of-month snapshot: it has to be part of every conversation. That's exactly what Pulse does and what we do at Delto: we turn every interaction into a measurable, actionable signal, so improving the experience stops being a hunch and becomes a continuous, data-backed practice. Want to see Pulse working on your own channels? Talk to our team .
What is Delto's Pulse? It's Delto's native module for measuring conversational feedback: it configures, customizes and schedules surveys for your end customers inside the same platform where agents are designed and deployed, connected to every conversation.
At which moments can surveys be measured? It ships with three default types: at the end of the conversation with the virtual agent, at the end of each skill (short, often a single question), and at the end of a chat with a human agent. You can also create any additional surveys you need.
How does Pulse measure when the customer doesn't complete the survey? Because feedback is monitored natively in every conversation, Pulse measures the customer's tone, their overall satisfaction and applies tags — capturing the signal even when there's no explicit answer.
Which question formats does it support? NPS, star rating, thumbs up/down, single or multiple choice, and short open text — each format to best capture what you want to measure.