Managing synthetic scores
Every synthetic score on a dataset is listed in Manage Dataset under Synthetic scores. From there you can improve a score or delete it.

Edit or improve a score
Select the edit icon next to a score. This reopens the same steps you used to create it, so you can change the description or categories, review more sample comments, and save again. Editing is also how you improve a score's accuracy over time.
Changing the description or categories re-scores every comment in the dataset. Results in Analysis, Dashboards, and Answers will change once re-scoring finishes, so let anyone who relies on the score know. Changing only the name does not trigger re-scoring.
Delete a score
Select the delete icon next to a score. Deleting a score frees up one of your account's synthetic scores straight away.
How many synthetic scores can I have?
Each account includes a set number of synthetic scores. Here's how they're counted:
- Each saved score on each dataset counts as one
- Reviewing sample comments and re-testing is free, however many rounds you do.
- Editing and re-saving an existing score doesn't use another one.
- Deleting a score frees it up immediately.
When your account has used all of its included scores, the Synthetic scores tab shows the message You cannot add another synthetic score, and you can't start a new one. To make room, delete a score you no longer need, or contact your Thematic representative to add more.

Common questions
Can I see or edit the scoring instructions?
No. Thematic writes the scoring instructions behind the scenes from your description, categories, and the corrections you make while reviewing. To change how a score works, edit it and give feedback on sample comments.
Can I add a synthetic score to a lens?
Not yet. Synthetic scores work on individual survey datasets.
Where can I use a synthetic score?
Anywhere you can use a score: Analysis, Dashboards, and Answers. For tips on explaining AI-generated scores to stakeholders, see Best practices for using AI metrics and scores.
Was this article helpful?
Your feedback helps us improve our documentation.