# Track a score over time with Scorecards Canonical page: https://www.questionpunk.com/support/scorecards Follow an NPS, CSAT, or rating question across 7- or 30-day windows, see what low scorers mention, and get told when the score drops. A scorecard follows one score question from a survey, such as onboarding NPS or support CSAT. It compares the latest window with the one before, shows the trend, lists what low and high scorers said, suggests fixes, and alerts you when the score drops by more than chance would explain. Last reviewed: 2026-11-08 Time: 5 minutes Level: Beginner Audience: Product teams, Customer experience teams, Researchers ## Before you start You need a survey with an NPS, rating, opinion scale, or slider question that is collecting responses. Scorecards are personal: each person creates their own for the surveys they can open. ## Steps ### 1. Create a scorecard In the sidebar, open **Analysis → Scorecards** and click **New scorecard**. Or, in a survey’s **Results → Summary**, click **Track this score** under a score question’s chart. ### 2. Pick the survey and score question Choose the **Survey** and the **Score question**: an opinion scale (including NPS), rating, or slider question. Give the scorecard a **Name**, such as “Checkout”. ### 3. Choose how it is scored Under **Score as**, pick NPS (% answering 9-10 minus % answering 0-6), CSAT (share of answers in the top quarter of the scale), or the average answer. For an average, tick **Lower is better** when a lower answer is the good one, as in “how much effort did it take?”. Under **Compare**, choose **Last 7 days** or **Last 30 days**. Add an optional **Target**, then click **Track scorecard**. ### 4. Read the status and trend Each scorecard shows **On track**, **Below target**, **Dropping**, or **Not enough data**, with the change against the previous window. The chart shows the last six windows; the shaded band is the range chance alone could explain, and a hollow point has too few answers. Switch to **Show table** for the numbers. ### 5. Review what to fix Under **What to fix**, AI suggests fixes from what low and high scorers said, with verbatim quotes and links to each response. When you act on one, click **Mark as done**: a dotted line marks the date on the chart, and the scorecard compares the score before and after. ## Details A window needs at least 10 answers to get a status. A drop needs 20 answers in each window, a difference larger than chance would explain (Welch’s t-test at 95%), and at least 5 points (5% of the scale for an average). Each response counts once, at its latest answer. Test, owner, and fraud-flagged responses are left out. Scorecards are checked daily. A new drop sends an in-app notification when **Response Alerts** is on, and at most one email a day when **Response Alerts** and **Account Notices** are both on. Set these in **Settings → Account Preferences**; see [Notifications](https://www.questionpunk.com/support/notifications). Fixes need at least 5 low scorers in the window. If low scorers did not say why, add an open-ended question after the score question, such as “What is the main reason for your score?”. Running [Themes](https://www.questionpunk.com/support/qualitative-analysis) on the survey lets fixes cite the themes low scorers mention more often. Deleting a scorecard stops tracking and deletes its recommendations. The survey and its responses stay. ## Check the result The scorecard appears under Scorecards with its status, trend, and, once enough low scorers have answered, suggested fixes. ## If something goes wrong If no survey is offered, add an NPS, rating, or opinion scale question to a survey first. If the status says **Not enough data**, share the survey to collect at least 10 answers in the window. ## Related articles - [View your survey report](https://www.questionpunk.com/support/view-report.md) - [Qualitative analysis and coding](https://www.questionpunk.com/support/qualitative-analysis.md) - [Notifications and email preferences](https://www.questionpunk.com/support/notifications.md)