> For the complete documentation index, see [llms.txt](https://boundaryai.gitbook.io/boundaryai-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://boundaryai.gitbook.io/boundaryai-docs/feedback-groups/feedback-groups.md). # Feedback Groups A **Feedback Group** is a space that centralises every piece of feedback for a specific project, programme, or initiative. It is the organisational backbone of your work in BAI Analytics: a single container for every channel, every dataset, every analysis, and every report that belongs to one effort. When you put feedback inside a Feedback Group, you stop thinking about individual surveys, sources and uploads as isolated artefacts. You start thinking about a project: *"what is the customer experience in Europe this quarter?"*, *"how is the second-year cohort feeling about the new curriculum?"*, *"what do partners think of our onboarding?"* And BAI Analytics keeps every input and every insight in one place so you can answer those questions over time. *** ### What a Feedback Group contains A Feedback Group can hold any combination of the five feedback channels BAI Analytics supports. You do not have to use all of them. Most groups start with one or two and grow over time. | Channel | What it is | | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Surveys** | Custom-built questionnaires sent to respondents via a share link, QR code, or embed. | | **Uploads** | Existing data imported as spreadsheets (`.xlsx`, `.xls`, `.csv`), documents and interview transcripts (`.docx`, `.pdf`, `.json`), or audio recordings, transcribed automatically. | | **Connectors** | Tickets and conversations pulled in automatically from third-party tools (Zendesk, Freshdesk, Intercom, Jira, HubSpot, and others). | | **Web scraping** | Reviews and mentions collected on a schedule from public sources (Google Reviews, Trustpilot, TripAdvisor, Reddit, X, Google News, and others). | | **AI Visibility** | AI-assistant answers about your brand (Claude, ChatGPT, Gemini, Perplexity, and others), collected with the agent wizard. See [AI Visibility](/boundaryai-docs/bringing-in-your-feedback/ai-visibility.md). | Whichever channels you use, BAI Analytics normalises everything into the same analytical model (themes, sentiment, segmentation, Custom Monitoring) so a Trustpilot review and an Excel column of NPS comments end up in the same dashboards. #### Supported upload formats | Format | How it's used | | --------------------------------------- | --------------------------------------------------------------------------------------------------- | | `.xlsx`, `.xls`, `.csv` | Spreadsheet upload: you map columns to questions. | | `.docx`, `.pdf` | Document upload, including interview transcripts: the app detects speakers so you can assign roles. | | `.json` | JSON interview transcripts: speaker roles are read from the file. | | `.mp3`, `.wav`, and other audio formats | Call recordings and voice memos: transcribed automatically, then analysed like any other upload. | #### Real examples * *Management Faculty: Course Evaluation 2026* * *Product X: Customer Feedback (Europe)* * *Campus Life Insights: First-Year Cohort* * *Partner Engagement: Q2 2026* * *Beta Program: In-Product Feedback* *** ### Creating a Feedback Group 1. Click **New Feedback Group** from the Feedback Groups page (or the dashboard). 2. Enter a clear, descriptive name (2 – 255 characters; must be unique within your organisation). 3. Click **Create**. You land directly inside the new group's page, ready to add a survey, upload a file, connect a platform, or set up a scraper. A Feedback Group starts empty by default, no auto-generated survey, no placeholder content. You decide what belongs in it. {% hint style="success" %} **Naming tip.** Use a name that will still make sense in a year. *"Q2 NPS"* is fine for this quarter, but *"Customer NPS, Quarterly"* keeps working as you add the next cycle. {% endhint %} *** ### The Feedback Group page Every Feedback Group has its own page, laid out around the **project** rather than around individual surveys, so you can read the state of a programme at a glance. #### Header The header shows the group name, a back-link to the Feedback Groups list, and a rename pencil. (Renaming is disabled for the default *Individual Surveys* group.) #### Quick Overview A summary band at the top of the page surfaces: * **Total interactions analysed** across every channel in the group. * **Sentiment breakdown**: positive, slightly positive, neutral, slightly negative, negative. * **Sentiment trend**: a chart of how sentiment has moved over time. * **Last interaction**: the timestamp of the most recent piece of feedback. Time-range filters let you narrow the view to the last week, month, three months, six months, twelve months, or all time. #### Source tabs Below the overview, the page is split into tabs for each source type: * **All Surveys**: combined view across every channel, useful when you do not care which channel a response came from. * **Surveys**: manually-built surveys. * **Uploads**: Excel, CSV, document, and JSON imports. * **Connectors**: connected ticketing platforms. * **Web Scraping**: running and historic scrapers. Each tab shows what is already in the group, plus a quick-start CTA to add a new source of that type. #### Reports widget A reports panel on the page shows how many reports and report schedules are scoped to this group, with shortcuts to **New Report** and **See** **Reports**. Reports are described in their own section below. *** ### Group-level analytics A Feedback Group is more than a folder. The genuinely useful thing about putting surveys, uploads, and tickets in the same group is that BAI Analytics rolls them up into cross-channel analytics. #### Cross-source overview The Quick Overview at the top of the page aggregates *every* response across *every* channel. A scraped Trustpilot review counts the same as an open-ended NPS comment from a survey, and both feed into the same sentiment trend. #### Grouped Themes (previously called Super-Themes) **Grouped Themes** are themes that span multiple sources inside the group. Where a single-source analysis might surface *"price"* and *"shipping speed"*, a Grouped Theme groups those threads across surveys, uploads, and tickets so you can see what is *actually* dominant at the project level, not just what is loud in one channel. For each Grouped Theme you get: * **Aggregate mention count** across the whole group. * **Sentiment**: a single signed score that summarises whether the theme is praised or criticised on average. * **Source surveys**: which inputs contributed. * **Last computed** timestamp and a manual **Recompute** option to re-cluster after a major data change. Grouped Themes use AI to cluster similar findings across channels, so wording differences (*"too expensive,"* *"price is too high,"* *"not worth the cost"*) collapse into a single row. #### Ask the data (BETA) Open a grouped theme and use **Ask the data** to ask a plain-language question about it. The panel describes itself like this: > Ask any question about this grouped theme, answers are grounded in the actual comments. The input placeholder is: > What do customers say about pricing? Click **Ask** to run the question. Under **Try:**, the panel suggests prompts such as: * What are the main pain points? * Are there any contradictions? * What do customers want fixed first? The panel also shows this disclaimer: > BETA feature: for the most accurate insights, explore the evidence-backed thematics and sub-thematics. Answers include citations back to real comments. Click a citation number to open its source comment, or use **View all N citations** to review the full set. Each citation can be rated as **Helpful citation** or **Misattributed citation**. Questions and answers are kept under **Saved Q\&A**. They are saved with the grouped theme and survive a page reload. You can delete saved items when you no longer need them. This feature also needs a minimum amount of evidence before it can answer. If the grouped theme is too small, the panel shows: > At least N comments are required to answer questions. #### Going deeper on a grouped theme The grouped theme detail view includes deeper AI tools when you need more than the default summary. **Sub-Categories** The **Sub-Categories** panel is described as: > AI-detected sub-categories within this group Click **Generate sub-categories** to create a more specific breakdown inside the grouped theme. The product sets expectations clearly: > Reads every comment and groups them into specific sub-themes. Takes 5–20 minutes depending on size, nothing is sampled, so larger grouped themes take longer. **In-Depth Summary** Click **Generate In-Depth Summary** to create **In-Depth Summary** content for the grouped theme. This is a multi-section synthesis across all constituent themes' comments. It pulls together patterns, sentiment intensity, contradictions, and actionable insights. **Consensus** **Consensus** tells you how much the comments inside a grouped theme agree on sentiment. A low value means the comments broadly point in the same emotional direction. A high value means the grouped theme mixes positive and negative signals, so the evidence is more conflicted. #### AI Insights The **Insights** view goes one step further: it asks the AI to summarise the group in plain language. It returns: * A handful of headline themes with explanatory copy. * Sentiment recaps with examples. * Language patterns and recurring phrases. * Category breakdowns (where applicable). * A short list of suggested action items. Insights are generated in the background and saved, so you can keep working and come back when they are ready. #### Choosing which sources feed your grouped themes Each source in a Feedback Group has a **Sources included** toggle. Its purpose is: > Choose whether this source's feedback is included in grouped theme analysis. Leave a source as **In analysis** to let it contribute to grouped theme analysis. Switch it to **Excluded** to keep a noisy or off-topic source out of the cross-source roll-up without deleting its data. When a source is included, the tooltip reads: > This source contributes to grouped theme analysis. Turn off to exclude it. *** ### Reports Reports are **scoped** to a Feedback Group, which means a single report can summarise everything across the group's surveys, uploads, scrapers, and connectors, or you can narrow it down to specific channels inside the group. From the group's Reports widget you can: * **Create a new report** scoped to this group. * **Schedule a recurring report** (daily, weekly, monthly, or a custom interval) and have it emailed to recipients automatically. * **Filter the report's contents** to specific surveys, themes, or time ranges within the group. * **Export** the resulting report as PDF, HTML, or JSON. Reports inherit the group's structure (surveys, themes, Grouped Themes, sentiment, monitors) so a single PDF can act as a quarterly readout for an entire programme, without you having to assemble it from separate sources. *** ### Sharing & permissions Feedback Groups can be shared with other members of your organisation. Each user is assigned one of two roles per group: | Role | What they can do | | -------------- | ----------------------------------------------------------------------------------------------------------------------- | | **Read-only** | View surveys, uploads, connectors, analyses, Grouped Themes, insights, and reports. Cannot create or edit anything. | | **Read-write** | Everything Read-only can do, plus create, edit, and delete sources, run analyses, generate reports, and manage sharing. | The user who creates a Feedback Group automatically receives Read-write access. Org-level admins always have full access regardless of the group-level setting. {% hint style="info" %} **Note.** Permissions today are a two-tier model (Read-only / Read-write), not a three-tier *Viewer / Editor / Owner*. If you need to restrict who can delete the group itself, manage that at the organisation level. {% endhint %} *** ### Renaming and deleting #### Rename Click the pencil next to the group name to open the rename dialog. The new name must be 2 – 255 characters long and must not collide with another group in your organisation. Renaming does not affect the group's URL, share links, or any of its surveys, only the display name changes. #### Delete Deleting a Feedback Group is intentionally a more deliberate action. * The group itself is removed from the list. * The deletion is audit-logged for traceability (who, when, which group). * **The surveys inside the group are not deleted.** They are detached from the group and remain available; if they have no other home, they fall back into the default **Individual Surveys** group on next sync. If you want to permanently delete the data, delete the sources first (and confirm the cascade prompt for their responses), then delete the group. {% hint style="warning" %} **Heads up.** *Individual Surveys* (the default group) cannot be renamed or deleted. It always exists per organisation as a fallback. {% endhint %} *** ### The default "Individual Surveys" group Every BAI Analytics organisation has a built-in Feedback Group called **Individual Surveys**. It exists to catch any survey that was created without explicit assignment to a custom group, so no work is ever orphaned. You can use it freely, but treat it as a temporary holding area: * It cannot be renamed. * It cannot be deleted. * It cannot be shared with specific members (org-level access applies). * It is excluded from group-level comparisons that look at distinct project trends. If a survey is going to be re-run, compared, or reported on as part of a programme, move it out of *Individual Surveys* into a custom Feedback Group as soon as you can. *** ### A note on Custom Monitoring A monitor (an AI-driven detection rule used for tagging and notifications, see [Custom Monitoring](/boundaryai-docs/analysing-your-feedback/custom-monitoring.md)) can be scoped two ways: to **one question** on an individual survey, upload, connector, or scraper, or to **every source in the group at once**, continuously, with a single alert and a trend line over time. The second option is set up from the group's Quick Overview instead of configuring the same monitor on each source individually. Group-level monitors can also **cover new sources automatically**: with a monitor's *Auto-cover new sources* toggle on (the default for new monitors), any source added to the group later, whether an upload, connector, agent, or API push, is watched from its very first analysis. There is no monitoring step to fill in when adding a source. Either way, **reports can aggregate monitor results across every source in the group**, so you can report on them globally from the group's Reports widget regardless of which scope they were defined at. *** ### Best practices * **Use one Feedback Group per real-world initiative.** A group should map to a project, programme, or environment that has a name in plain English ("Q2 NPS: Europe", "Onboarding Survey 2026"). Avoid creating a group per survey. * **Mix channels deliberately.** The point of a group is cross-channel analysis. Pulling in scraped reviews alongside survey responses inside the same group lets Grouped Themes surface patterns that no single channel could. * **Move work out of *****Individual Surveys***** once it has a home.** Treat the default group as a triage area, not a long-term destination. *** ### What's next * Add the first source: a survey, an Excel upload, a connector, or a scraper. * Style the surveys to match your brand. See [Customizing a Survey](/boundaryai-docs/bringing-in-your-feedback/surveys/images-and-media.md). * Track lifecycle and run cycles cleanly. See [Managing Surveys](/boundaryai-docs/bringing-in-your-feedback/surveys/managing-surveys.md).