AI Conversation Sentiment Analysis
Track how AI platforms describe your brand over time — and whether that coverage is working for or against you.
Quick start
1. Select a Project from the filter bar
2. Set the Date Range to 30 Days to see a meaningful trend
3. Check which sentiment category is largest on the chart
4. Use Provider Depth to focus on top recommendations if they matter most
5. Review the colour bands to understand the trend direction
Understanding the display
The chart shows sentiment distribution as stacked colour bands, each representing a day. The combined height always equals 100% — the bands show what proportion of AI responses fell into each sentiment category on that day.
Larger Positive (blue) bands signal strong, favourable coverage. Expanding Negative (pink) bands are worth investigating. Mixed (orange) indicates nuanced or balanced mentions, and Unknown (grey) means sentiment could not be determined.
Gaps in the chart are expected — they show days when no AI responses were collected for that project and filter combination.

Sentiment categories
| Category | Colour | Meaning |
| Positive | Blue | AI platforms described the brand favourably |
| Mixed | Orange | Coverage was balanced or contained both positive and negative elements |
| Negative | Pink | AI platforms described the brand unfavourably |
| Unknown | Grey | Sentiment could not be determined from the response |
What should I do next?
| What you see | Action |
| Positive band is growing | Maintain current content and brand messaging — it's working |
| Negative band is growing | Review recent AI responses and identify which topics are generating negative sentiment |
| Mixed band dominates | Investigate whether AI platforms lack clear brand signals — stronger, more consistent content can help |
| Unknown band is large | Check that AI platforms are finding enough brand content to assess — thin coverage produces unknown results |
| No data on most days | Expand the date range or check that AI tracking is active for this project |
| Sudden shift in any band | Add an annotation to mark the date, then review what changed — a campaign, product launch, or news event may be the cause |
Using the filters
The filter bar controls which data appears in the chart. Changes apply immediately.
Date range
| Option | Best for |
| 7 Days | Quick recent check |
| 30 Days | Standard trend analysis (default) |
| 90 Days | Longer-term pattern review |
| 180 Days | Extended trend or campaign analysis |
| Custom | Specific time periods — useful for isolating campaign windows |
Platform
Filter by a specific AI platform to isolate how one service describes the brand. Leaving this set to All combines data across all tracked platforms. Platforms include ChatGPT, Claude, Gemini, and Perplexity, along with their web-browsing variants.
Configuration
A configuration represents a specific set of AI query settings — typically a combination of prompts, topics, and targeting options. Filtering by configuration focuses the chart on one tracked setup.
Phase
Phases represent stages of the buyer journey: Awareness, Consideration, and Decision. Filtering by phase reveals whether sentiment differs depending on how far along the buyer journey the AI prompt is positioned.
Provider depth
| Option | What it measures |
| Mentioned Anywhere | All mentions of the brand in AI responses, regardless of position |
| Top 3 Recommendations | Only mentions where the brand appears in the top three recommendations |
| #1 Recommendation | Only mentions where the brand is the top recommendation |
Narrowing provider depth to top positions gives a clearer picture of how AI platforms rate the brand when it matters most.
Interactive features
• Hover over any point to see the exact date and percentage for each sentiment category
• Click legend items to show or hide individual sentiment bands — useful for focusing on a single category
• Export via the menu icon (⋮) to download the chart as PNG, JPEG, or SVG
Annotations
Annotations mark significant dates on the chart — making it easier to connect sentiment shifts to real-world events such as campaigns, product launches, or news coverage.
Adding an annotation
6. Click anywhere on the chart at the date to mark
7. Enter a label in the pop-up (for example: "Spring campaign launch")
8. Press Enter or click Save
A labelled vertical line appears on the chart at the selected date.
Deleting an annotation
Click the annotation label on the chart. A confirmation prompt appears — confirm to remove it.
Troubleshooting
No data shown
The project may have no AI tracking data for the selected period, or the filters may be too restrictive.
• Expand the date range
• Set Platform and Phase to All
• Verify the project has active AI tracking
Chart shows gaps on many dates
Gaps are normal — they indicate days when no AI responses were collected for the selected filters. A high gap rate across an extended range may indicate that tracking is infrequent or that data collection paused.
All sentiment appears as Unknown
If the Unknown band dominates, AI platforms may not have enough brand content to make a sentiment assessment. Improving brand visibility and publishing clear, authoritative content helps establish consistent sentiment signals.
Annotations not appearing
Annotations are scoped to a project and configuration. Ensure the correct Project is selected. If the annotation was added under a different configuration, it may not appear when filtering by a specific one.
FAQ
Does the chart show absolute mention counts or percentages?
Percentages only. The chart shows the proportion of AI responses in each sentiment category on a given day, not the raw number of mentions. A day with few responses and a day with many responses look the same on the chart.
Why does the chart sometimes look different across platforms?
Each AI platform processes prompts differently, has different training data, and weights brand signals in its own way. Sentiment can vary significantly between platforms, particularly for brands with limited AI coverage.
What does a large Mixed band mean?
Mixed sentiment suggests AI platforms are finding both positive and negative signals about the brand, or that their coverage is nuanced rather than clearly positive or negative. Reviewing the specific AI responses behind this data helps identify what is driving the balance.
Can I compare multiple projects at once?
The chart focuses on one project at a time. To compare across projects, view each separately and use annotations to mark the same key dates for comparison.
How frequently does data update?
Data updates depend on the AI tracking schedule configured for the project. Check account settings for specific timing.
Getting help
9. Review this guide for filter and display explanations
10. Adjust filters to isolate the data in question
11. Contact support for data or configuration issues