InformlyInformly Docs
Insights

Widget insights

Widget insights give you a per-widget breakdown of volume, message length, top questions, and completion rate, with side-by-side comparison.

Widget insights show how each widget in your organization is performing on its own terms. Where the Conversations sub-page aggregates everything, this page keeps each widget in its own row, so you can spot the one that's pulling its weight and the one that needs work. Go to Insights → Widgets to open it.

The widget table

Each row is one widget; each column is a metric over your selected date range.

ColumnWhat it shows
Total conversationsNumber of chats the widget handled.
Avg messages per conversationHow long the typical chat runs.
Top customer questionsThe three subjects asked most often, pulled from auto-tagging.
Completion rateShare of chats that ended cleanly (customer got an answer or accepted handoff).

Sort by any column. Click a row to drill into that widget's conversation history.

"Top customer questions" is a great place to look for documents you should be writing. If the same question shows up across two widgets, write the answer once in Documents and assign it to both.

Compare two widgets head to head

Use compare mode when you want to test the impact of a change — a new persona, a different appearance, an added skill — before rolling it out everywhere.

Pick the two widgets

Tick the checkbox next to two widgets in the table. A Compare button appears at the top.

Open compare view

Both widgets show side by side, with each metric stacked for direct comparison.

Pick the right date range

If you're testing a change, set the range to start the day after the change shipped. Compare against an equivalent window before the change for context.

Read the deltas

A green delta is the widget that's ahead on that metric. A red delta is the one trailing.

Drill into one widget

Clicking a widget row takes you to the Conversations page pre-filtered to that widget. From there you can sort by feedback, open individual chats, and follow the workflow in Working with a conversation.

Use cases

A few ways teams use this page.

Before and after a persona change

Tighten a persona, wait a week, then compare the widget against itself across the two periods using the date range comparison. Look at average messages per conversation and completion rate — those move first when the AI gets clearer instructions.

Spot a low-performing widget

Sort by completion rate ascending. The widget at the top of the sorted list is the one to investigate. It might be missing documents, missing skills, or pointing at the wrong project.

Find the workhorse

Sort by total conversations descending. The top widget is the one customers actually use — make sure its documents and persona are well maintained.

What's next

On this page