> ## Documentation Index
> Fetch the complete documentation index at: https://docs.feedal.io/llms.txt
> Use this file to discover all available pages before exploring further.
# AI insights
> AI-generated summaries, sentiment analysis, topic clustering, and trend detection across your responses.
The **AI** tab analyses your form's responses using a language model and returns structured insights: an overview dashboard, sentiment breakdown, topic analysis, trend detection, and a plain-language summary with recommendations.
Open a form and click **AI** in the top navigation. AI insights require the **Pro plan**.
***
## Generating insights
1. Set the **date range** for the responses you want to analyse (default: last 30 days).
2. Click **Generate insights**.
3. Analysis typically completes in 10–30 seconds depending on response volume.
Each generation is saved to history. Use the **History** dropdown to revisit a previous run without re-generating.
Generating insights consumes API quota. Each generation analyses all responses
in the selected date range. Re-generating with the same date range produces a
fresh analysis.
***
## Overview dashboard
Four stat cards summarise the analysed dataset:
| Card | What it shows |
| ------------------------ | ----------------------------------------------------------------------------- |
| **Total feedback** | Number of responses analysed, with a response rate progress bar |
| **Response rate** | Percentage of visitors who completed the form |
| **Overall satisfaction** | Composite satisfaction score (0–100), with a trending indicator when above 70 |
| **Actionable insights** | Count of key insights identified, broken down by priority |
Three charts appear below the cards:
**Feedback trends** — a line chart with two series (feedback volume and satisfaction score) over time. Use this to correlate changes in your product with changes in sentiment.
**Sentiment distribution** — a donut chart showing the split between positive, neutral, and negative responses.
**Ratings distribution by type** — a stacked bar chart breaking down rating answers across your emoji, star, and scale questions.
***
## Sentiment analysis
The sentiment tab analyses the language in text answers and shows:
**Sentiment trends over time** — three lines (positive, neutral, negative) plotted by date. Useful for detecting whether a product change shifted sentiment.
**Top positive words** — the words most associated with positive responses, ranked by frequency.
**Top negative words** — the words most associated with negative responses, ranked by frequency.
**Sentiment by source** — a stacked bar chart showing the sentiment split for each traffic source (URL, QR, Embedded, Email, Other). Helps you detect whether certain channels attract more critical respondents.
An **AI insight card** summarises the sentiment findings in plain language with percentage breakdowns.
***
## Topic analysis
The topics tab clusters response text into recurring themes and shows:
**Main topics** — a bar chart of the most-mentioned topics, with each bar coloured by sentiment score:
| Colour | Sentiment score |
| ------ | ---------------- |
| Green | ≥ 70 (positive) |
| Amber | 50–69 (neutral) |
| Red | \< 50 (negative) |
**Topic sentiment** — a radar chart comparing positive, negative, and neutral sentiment across the top topics side by side.
**Topic trends over time** — a line chart showing how frequently each topic appears across the date range. Use this to see whether a topic is growing or fading.
**Related terms & keywords** — a tag cloud of keywords grouped under each topic, showing the specific language respondents use.
An **AI insight card** summarises each topic with its mention percentage and positive sentiment score.
***
## Summary & recommendations
The summary tab contains two sections generated entirely in plain language:
**Insights** — a prose summary of the most significant patterns, strengths, and weaknesses identified across all responses.
**Recommendations** — a prioritised list of actionable suggestions based on the insight findings. Each recommendation is tied to a specific pattern in the data.
***
## Trends
The trends tab surfaces changes in key metrics over the analysis period:
**Key metrics change** — a list of metrics (e.g. completion rate, average satisfaction) with their percentage change, direction (↑ or ↓), a description of what drove the change, and a supporting data point.
**Emerging patterns** — newly detected themes or behaviours that have grown in prominence during the period, with an impact statement for each.
***
## Viewing historical runs
Click the **History** dropdown (top right of the AI tab) to see all previous insight generations. Each entry shows:
* When it was generated
* How many responses were included
* The date range that was analysed
Select any entry to view its results. A **Historical data** badge indicates you're viewing a past run rather than a fresh analysis. Click **Back to live** to return to the latest generation.
***
## Next steps
Raw aggregate metrics: visits, responses, completion rate.
Read the individual responses the AI analysed.