Skip to main content

Research

An interactive topic explorer — pick a topic from the feedback, filter by sentiment, read frequency over time, descriptors, and the actual feedback examples.

The Research page is the “qualitative explorer” of the Fast Feedback tab. Where Sentiment gives you the firm-wide read, Research lets you drill into a specific topic — pick “communication”, filter to negative sentiment, and read the actual feedback that drove the signal.

What this page answers

For any topic the firm talks about in feedback: how often is it mentioned, what’s the sentiment breakdown, what descriptors are most associated with it, and what does the actual feedback say?

How the page is laid out

Topic sidebar (left)

A scrollable list of the topics (aspects) Compass extracted from feedback in your date range — the top 100 by mention count, with the 25 most-mentioned shown first and the rest behind a More topics expander. Each entry shows:

  • The topic name

  • The count of mentions

A search box at the top of the sidebar filters topics by name. Pinned at the bottom are sentiment filter checkboxes — Positive, Neutral, Negative (all ticked by default). Once you’ve selected a topic, each checkbox also shows the count of matching mentions.

Content area (right)

When no topic is selected, you’ll see a prompt to pick one. When you select a topic, the content area loads with three sections:

  1. Topic Frequency — A line chart of how often this topic has been mentioned over time, broken into 6-month intervals

  2. Key Descriptors — A word cloud of the qualifiers (descriptive terms) Compass paired with this topic. Word size = frequency

  3. Feedback Examples — A list of cards, each showing an individual feedback entry where this topic was mentioned

Reading the feedback example cards

Each card shows:

  • Author → Recipient (Date) with a sentiment badge (green for positive, grey for neutral, red for negative)

  • Feedback text — The sentence from the feedback where the topic was mentioned

  • Descriptor tags — The qualifier terms Compass paired with the topic in this entry

Where the data comes from

Everything on this page is from fast feedback NLP:

  • Topics (aspects) — Compass extracts these from each feedback entry’s text

  • Qualifiers — Descriptive terms paired with each topic

  • Sentiment classification — Per-sentence positive / neutral / negative

The page loads in stages. The topic list and counts are immediate; the content area for a selected topic loads via AJAX when you click.

How to read it

  • Start by browsing the topic list. What’s the firm talking about? The topics at the top of the list (by count) are the dominant conversation themes.

  • Sentiment-filter to “negative only” for a concern list. Then click each negative topic in turn to read what’s actually being said.

  • Topic frequency tells you trend. A topic that’s growing in mentions is an emerging theme; one that’s declining is something the firm is moving past.

  • Key Descriptors are how people characterise the topic. “Clear communication”, “lacking communication”, “improving communication” — the descriptors paint the picture.

  • Read the feedback examples to confirm. Aggregates can mislead; the actual text never does.

Tips

  • Use the search box. With dozens or hundreds of topics in the sidebar, search is the fastest way to find what you’re looking for.

  • Filter sentiment before you scan. Pre-filtering to (say) negative only lets you scan the topic list for problems.

  • Read examples, not just counts. A topic with 30 mentions where every one is a vague throwaway is less useful than a topic with 5 mentions that are each substantive.

  • Empty page? If NLP hasn’t run yet, the page shows a notice instead of the topic list.

  • This is the qualitative complement to Sentiment. Sentiment tells you what’s happening at the cohort level; Research lets you read the actual story behind it.

Did this answer your question?