Before you start

Prompt Effectiveness

The Prompt Effectiveness page shows you how your role prompts are performing across the organization. Use it to see which prompts people actually use, how they’re rated, and which ones need attention, so prompt authors can iterate and improve.

Grengin prompt effectiveness

Before you start

To open this page from the navigation, you need the View Analytics permission. To actually load the page, you also need the View Roles permission. Both can be granted organization-wide or scoped to a department.

Because the two requirements differ, it’s possible to see the menu link but be blocked when you open the page (or the reverse). If you can see Prompt Effectiveness in the menu but the page won’t load, ask your administrator to confirm you have both permissions.

You’ll find the page under Control Hub → Monitor → Prompt Effectiveness.

What you’ll see

The page opens with a header, a filter bar, and a row of five summary cards (KPIs):

  • Total Usage — how many times the prompts in view were used.
  • Total Feedback — how much feedback those prompts received.
  • Thumbs Up and Thumbs Down — an estimate of the positive/negative split based on usage and ratings.
  • Avg Rating — the average effectiveness rating, shown as a percentage and color-coded green, amber, or red so you can spot trouble at a glance.

Below the cards, four tabs let you explore the data in different ways.

Filtering the data

Use the filter bar at the top to narrow what every card and tab shows:

  1. Pick a role from the Role menu to focus on that role’s prompts.
  2. Pick a prompt from the Prompt menu to focus on a single prompt.
  3. Click Refresh at any time to pull the latest numbers.

If you arrive here from the Prompt Library page by clicking through on a specific prompt, the Prompt filter is pre-selected to that prompt for you.

Reviewing per-prompt usage (Overview)

The Overview tab lists every prompt in view with its name, its role, usage count, feedback count, thumbs up/down, and a rating bar. This is the best starting point for a quick scan of how each prompt is doing.

Finding your best and worst prompts (Rankings)

The Rankings tab shows two short lists side by side:

  • Most Effective — your top five prompts by average rating.
  • Least Effective — the five lowest-rated prompts that have actually received feedback. These are your best candidates for a rewrite.

Seeing department adoption (Adoption)

The Adoption tab shows, per department, the member count, prompts used, total usage, and an adoption-rate bar. The table pages 10 departments at a time; Previous and Next buttons appear only when there’s more than one page.

Treat adoption numbers as a rough indicator of reach rather than an exact per-department breakdown.

Comparing two prompts (Compare)

The Compare tab lets you put two prompts side by side:

Grengin prompt effectiveness compare
  1. Choose a prompt in the Variant A menu.
  2. Choose a prompt in the Variant B menu.
  3. The two stat cards appear with usage, feedback, and rating, plus a badge showing which one leads (or that they’re tied).

Until you pick both prompts, you’ll see a prompt to select both. Note that this comparison view is a preview and isn’t backed by live A/B test data yet, so use it as a visual aid rather than a source of truth.

Good to know

  • Loading and empty states. While data loads, you’ll see a spinner. If there are no metrics at all, the whole page shows an empty-state message. Individual tabs show their own “no data” message when a filter leaves them empty.
  • Errors. If something goes wrong while loading, you’ll see a brief error notification rather than a full error panel. Click Refresh to try again.
  • Thumbs up/down are estimates. These splits are calculated from feedback volume and average rating, not stored as separate counts, so rounding can make a single prompt’s up and down figures not add up exactly to its total feedback.
  • Adoption is indicative. Adoption figures use a simple formula and apply the same prompt set across departments, so “Prompts Used” and “Total Usage” may look identical from one department to the next. Read them as directional, not precise.
  • No date range or export. Filtering happens entirely in the browser. There’s no date-range, search, sort, or export control on this page.
  • Refreshes on demand. Data loads once when you open the page and updates only when you click Refresh. Changing a filter resets the adoption table to the first page.
  • Up to 100 departments. The adoption table shows up to 100 departments. In very large organizations, some departments may not appear.