Skip to main content

Using Highlights

Jack Miller avatar
Written by Jack Miller
Updated this week

Introduction

Highlights give you an instant, narrative view of what matters most in your data. They are generated by Lyra AI and summarise the key patterns in the feedback you are viewing.

Each Highlight is grounded in the same Observations used throughout the product - it’s never just free-form text.

Highlights are clustered into groups.

A Highlight group clusters together several related Observations that all speak to the same underlying issue. For example, Observations about “too much packaging” and “packaging is not recyclable” may be grouped under a single Highlight called “Packaging Issues”.

This grouping helps you focus on the bigger story rather than chasing many small, similar Observations.

Highlights help you move quickly from data to insight. You can copy them for reporting, add them to dashboards for ongoing tracking, or expand them to see the Observations and feedback behind them.

Where to Find Highlights

You can see Highlights in Feedback, Analytics, Dashboards, and on the Highlights details page.

In Feedback, a Highlights panel summarises the current result set so you can understand the main issues and praises without leaving the comments view. When you open an Observation from Feedback, you can expand to see the supporting Observations and Snippets.

In Analytics, Highlights appear in a panel above the chart or table and update as you change filters, date range, or breakdowns. They summarise what stands out in the view you are looking at.

In Dashboards, you can view a Highlights widget to keep key insights visible. The widget refreshes automatically with the dashboard’s filters and time period.

Interacting with Highlights

Highlights are not static text. You can work with them in several ways to make them more useful for your day-to-day analysis and reporting.

You can copy a Highlight summary to your clipboard and paste it into Slack, an email, or a report. This is the quickest way to share a key insight with a colleague or a stakeholder.

You can add a Highlight to a dashboard.

This pins that Highlight as a card in a dashboard, where it will automatically refresh whenever the underlying data changes. It’s a simple way to keep a key insight visible to your team.

You can expand a Highlight to open the Highlights Details page.

This page gives you a full view of the Highlight group and all the data behind it. When you open a Highlight from Analytics, Feedback, or a dashboard, you land on this page.

With 'Insights by Lyra' you normally see only the first three Observations linked to a Highlight, but on the Details page you can view all of the supporting Observations in that group. You can also browse all of the feedback linked to those Observations, making it easy to read the original customer comments without switching to another part of the platform.

The page includes a breakdown of feedback volume over time for the Highlight group, as well as a breakdown by data source, so you can see which channels or surveys the comments come from.

You can also take action directly from this page. Select one or more Observations and apply Quick Actions such as:

  • Merge to clean up duplicates or near-duplicates.

  • Drill Down by to focus your Analytics view on that specific issue.

  • Open in Reports to start a more detailed analysis.

  • Add to Dashboard to pin the Observation for ongoing tracking.

Low-Volume Views and Fallback Summaries

Highlights work best when there is enough data to identify meaningful groups of Observations.

If the dataset is too small - for example, when you apply very narrow filters or look at a short time range - there may not be enough volume to generate observation-based Highlights.

In these cases, Chattermill automatically switches to a fallback mode. Instead of showing grouped Highlights, Lyra analyses the raw feedback directly and provides a concise written summary.

The fallback summary still helps you understand the key points in the data you are looking at, but it does not include Highlight groups or the supporting Observations. For the richest and most actionable results, use Highlights with a reasonably sized dataset so that the model can detect clear patterns and show grouped Highlights.

Did this answer your question?