Viewing and Analyzing Conversation Results

After reading this article, you will be able to view detailed response data for any conversation, interpret the charts and breakdowns the platform provides, export response data, and understand how Oomiji's AI analyzes open-ended responses.

This article covers what happens after your conversation has collected responses. If you are looking for how to create a conversation, see Creating and Configuring Conversations. If you are looking for the dashboard-level overview of all your conversations, see Navigating the Insights Dashboard.

Opening the Observations Page

The Observations page is where you view responses for a specific conversation. To reach it, navigate to Insights, find the conversation you want to see, click the three-dot menu next to its name, and select Observations.

Note: The Analyze option does not appear in this menu because this conversation does not contain open-ended questions. See Analyzing Open-Ended Responses with AI below for when Analyze becomes available.

What You See on the Observations Page

Response Overview

The Observations page opens with the conversation name and status at the top, followed by a breadcrumb link back to the main Conversations list. Below that, five summary metrics appear in a row: Success rate (the percentage of started conversations that were completed), Conversation started (total number of respondents who began the conversation), Conversation completed (total who finished), Questions answered (total individual question responses across all respondents), and Active days (how long the conversation has been live).

A conversation with no responses yet will show 0% Success rate and zeroes across all four counts. This is normal for a newly activated conversation or one you have just cleared with Remove Responses.

Below the metrics, three expandable sections appear: Time Chart, Completions, and New Segment. Time Chart and Completions work the same way they do on the Insights dashboard. New Segment lets you create a segment of respondents directly from this conversation's results, which you can then use for targeted follow-up in other sections of the platform. For more on segments, see Building and Managing Segments.

The upper right corner of the page contains a Segments dropdown (for viewing or applying saved segments), a three-dot menu, and a + Create button.

Viewing Closed-Ended Question Results

Below the summary metrics and the Time Chart / Completions / New Segment row, an Analysis section appears. On the left side, a vertical list shows every question in the conversation by number and question text. Clicking a question loads its results on the right side.

For closed-ended questions, results display as a bar chart showing the count and percentage of respondents who selected each answer option. Three dropdown controls above the chart let you change the visualization: the chart type (Bar is the default), the sort order (Descending is the default), and the answer filter (Top 20 answers is the default). A Charts dropdown in the upper right of the analysis area lets you toggle between Charts view (the visual bar chart) and List view (a text-based list of responses).

For example, a question asking respondents to indicate their age group might show bars for "55 to 69" at 50%, "35 to 54" at 27%, "70 and over" at 20%, and "21 to 34" at 3%. The longest bar represents the most common answer. Look for large gaps between the top answer and the rest (strong consensus) versus evenly distributed bars (divided audience or a question that may need to be more specific).

Viewing Open-Ended Question Results

Open-ended responses are where your customers tell you things you did not think to ask about. Before AI analysis runs, these responses appear as raw text. Reading through them individually is valuable for small response sets, but becomes impractical at scale. This is where the AI analysis described later in this article becomes essential.

Filtering and Segmenting Results

The New Segment section on the Observations page lets you create a segment of respondents directly from this conversation's results. Click New Segment to expand the panel.

The segment builder reads: "If a conversation matches [Any/All] of the following conditions." Select your criteria from the Select Criteria dropdown, then set the condition using the Select Condition dropdown. Click View to preview which respondents match, or Reset to clear the criteria and start over. Click Save in the upper right of the panel to save the segment for use elsewhere in the platform.

The Segments dropdown in the upper right corner of the page lets you apply a previously saved segment to filter the results displayed below, so you only see responses from contacts who match that segment's criteria. The + Create button next to it opens the same segment builder.

The ability to segment results directly from the Observations page is one of the most powerful features of the Insights section. Rather than looking at all responses as one group, you can isolate respondents who gave a specific answer to a closed-ended question and then examine what that subgroup said in their open-ended responses. 

For example, if a Radiogroup question asks about satisfaction level, you can filter to only the respondents who selected "Very Dissatisfied" and then read their open-ended explanations. This turns broad survey data into targeted insight. For more on how segments work across the platform, see Building and Managing Segments.

Exporting Response Data

You can export all collected responses for a conversation as a CSV file. From the Insights dashboard, click the three-dot menu next to the conversation and select Export Data.

Exported data is useful for offline analysis, for sharing results with team members who do not have Oomiji access, or for importing response data into another tool.

Analyzing Open-Ended Responses with AI

When your conversation includes open-ended questions, Oomiji's AI can automatically analyze and categorize the freeform text responses. Instead of reading hundreds or thousands of individual answers manually, the AI identifies themes, patterns, and categories across the full set of responses, producing a detailed report.

This is the platform's core differentiator. Most survey tools can count how many people selected Option A versus Option B. Oomiji can also tell you why they chose what they chose, by analyzing the language they used in their own words. Combined with closed-ended segmentation, this means you can identify what each audience segment is saying, not just how they voted.

How AI Analysis Triggers

❗AI auto-categorization does not run immediately.❗ It triggers automatically within the first 7 days after a conversation is activated, provided the conversation has received at least 50 responses containing open-ended answers. The system checks for this threshold during the 7-day window and runs the analysis when the conditions are met.

There are two possible outcomes:

  • If 50+ responses are collected within 7 days: The AI analysis runs automatically. After it completes, a manual Refresh button appears. You can click Refresh to re-run the analysis once every 24 hours as new responses continue to come in.
  • If fewer than 50 responses are collected within 7 days: The AI analysis does not trigger and will not run for that conversation. The manual Refresh button will not appear. This is a hard cutoff; there is currently no way to manually initiate AI analysis on a conversation that missed this window.

This means timing matters. If you are launching a conversation and want AI analysis, plan your distribution to reach the 50-response threshold quickly. A conversation sent to a large list in a single distribution is more likely to trigger analysis than one sent in small batches over several weeks.

Running the AI Report

Once AI analysis has triggered, you can generate a detailed AI report. The workflow involves providing the AI with guidance about the context of your data before it processes the responses.

The Analyze option only appears in the three-dot menu for conversations that include open-ended questions and have collected sufficient responses. If you do not see Analyze in the menu, your conversation either does not contain an open-ended question type or has not yet reached the response threshold.

To run the AI analysis, go to the Insights dashboard, click the three-dot menu next to the conversation, and select Analyze.

A dialog box appears with the conversation name in the header and a text field with the prompt: "Please briefly describe what this conversation is about, including its industry and main purpose, in 500 characters or less." This context helps the AI produce more relevant categorizations. For example, if your conversation collected feedback from wine consumers, you might write: "This is a conversation with customers of a wine retailer and is intended to determine their purchasing preferences, interests, and needs to help the company identify high-value customer segments."

Be specific about your industry and your goal. The AI uses this context to interpret the language in your open-ended responses. A generic description produces generic categorizations.

Click Analyze to submit. The AI processes the responses and generates the report. Click Cancel to close the dialog without running the analysis.

What the AI Analyzes

The AI targets open-ended question responses specifically. It does not analyze closed-ended responses (those are already structured and counted automatically). The analysis works at the individual question level, meaning each open-ended question in your conversation is analyzed separately.

The AI identifies themes and categories in the freeform language. For example, if 200 respondents answered "What could we improve?" the AI might categorize their responses into themes like "pricing concerns," "customer service experience," "product quality," and "delivery speed," with the percentage of respondents whose answers fell into each category. This turns unstructured text into data you can segment and act on.

Pairing Closed-Ended and Open-Ended Analysis

The most effective way to use AI analysis is in combination with closed-ended segmentation. A conversation that asks "Which of the following best describes your satisfaction level?" (Radiogroup) followed by "What is the reason for your answer?" (Open-Ended) gives you both structure and depth.

You can segment respondents by their closed-ended answer (for example, all respondents who selected "Very Dissatisfied") and then view what that specific group said in the open-ended question. The AI analysis of the open-ended responses tells you the themes within that segment. This is how you go from knowing that 30% of respondents are dissatisfied to understanding what they are dissatisfied about, without reading every individual response.

For more on creating and using segments, see Building and Managing Segments.

Managing Response Data

Two additional actions are available from the three-dot menu on the Insights dashboard that affect response data:

  • Remove Responses permanently deletes all collected responses from the conversation. 

The Remove Responses action cannot be undone. Once removed, the response data is gone and cannot be recovered. This is primarily used to clear test data before launching a conversation to your real audience.

  • Deactivate stops the conversation from accepting new responses. All existing response data is preserved. The conversation remains on your dashboard and can be reactivated at any time. Use this when you have finished collecting responses and want to close the conversation without deleting anything.

What's Next