> ## Documentation Index
> Fetch the complete documentation index at: https://docs.videntic.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Build your Prompt Library

> Add, organize, and review the customer questions your Project measures.

Your **Prompt Library** contains the questions Videntic tests with AI Platforms. An **Active** Prompt is included in future Analyses; a **Suggested** Prompt needs your review; an **Archived** Prompt is retained but excluded from future measurement.

> [!NOTE]
>
> Prompts are questions you choose to monitor. Their results do not show how often real people ask those questions in private AI conversations or provide search-volume estimates.

## Start with useful questions

1. Open the Project, then **Dashboard → Prompts → Library**.
2. Review the Active and Suggested tabs before adding more. Look for repeated wording and gaps in the customer journey.
3. Choose a few questions that represent each important product or customer need. Include an early research question and a more specific comparison or buying question where relevant.
4. Write each Prompt as a customer would ask it, in the language and Market you intend to measure. Keep a stable core of Prompts so later Analyses are easier to compare.

For example, “How do I compare suppliers for a small team?” tests a broader need; “Which supplier supports our required integration?” tests a more specific decision. These are examples to adapt, not evidence that customers are already asking them.

### Plan a small, balanced starting set

Suppose a supplier serves small teams that need regional delivery and an accounting integration. A starting set could cover different decisions without repeating the same question:

| Customer question | Topic | Search Intent | Why keep it |
| - | - | - | - |
| “How do small teams compare suppliers that deliver in our region?” | Supplier selection | Informational | Tests an early research need. |
| “Which suppliers support our accounting integration?” | Integrations | Commercial | Tests a concrete shortlist criterion. |
| “What does it cost to start using a supplier with that integration?” | Pricing | Transactional | Tests a later decision. |

Adapt the wording to your customers and Market. Use Tags for a second grouping, such as a campaign or audience, if you need to review those Prompts together. Keep Topic and Search Intent distinct: the Topic describes the subject, while Search Intent describes the kind of decision. Do not add many near-identical phrasings just to fill the Plan allowance; each Active Prompt changes the set measured in future Analyses.

## Review suggestions and remove overlap

1. Open the **Suggested** tab. Read each question alongside the Prompts already in **Active**; suggested wording is a candidate, not a measured customer question.
2. Accept a suggestion only when it represents a relevant customer need that the Active library does not already cover. Reject irrelevant or overlapping suggestions. Check the available Prompt capacity before accepting a group.
3. Search the Active library for questions with the same underlying need, even when their wording differs. Keep the clearest customer-style version and archive an Active Prompt you no longer want measured. Use **Delete** only when you intend to remove the Prompt instead of retaining it for later review.
4. Confirm the Active tab contains the intended questions. Compare later Analyses with care after changing this set.

For example, “Which suppliers connect to our accounting system?” and “What suppliers offer our accounting integration?” may cover the same need. Keep both only if their different wording or context serves a deliberate measurement purpose.

## Add Prompts

Select **Add** in the Prompt Library. The current add flow offers these methods:

| Method | Use it when | Review before saving |
| - | - | - |
| **Generate with AI** | You need candidate questions for a Topic. | Keep only questions that fit your Brand, Market, and audience. |
| **Write my own** | You have one customer question. | Enter the wording and a Topic; Search Intent is optional. |
| **Bulk upload** | You already have a list or spreadsheet. | Check mapping, duplicates, language, Topic, tags, and remaining Plan capacity. |

### Import a prepared list

1. Select **Add → Bulk upload**. Choose **Paste** for one question per line with shared Topic, Tags, and Language, or **File** for a CSV or Excel sheet with different values per row.

2. For a file, include a header row and a column containing Prompt text. A useful starting structure is:

   ```csv theme={null}
   Prompt text,Topic,Language,Tags,Search Intent
   Which suppliers support our accounting integration?,Integrations,en,small-team;accounting,Commercial
   How do small teams compare suppliers in our region?,Supplier selection,en,small-team;research,Informational
   ```

3. Upload the file and check **Map columns**. Confirm which column supplies Prompt text, Topic, Language, Tags, and Search Intent. A Tags cell can contain multiple tags separated by a semicolon. For Paste, enter a Topic and choose the shared defaults below the text box.

4. Set defaults for missing file values. Values supplied by the file win for their rows. Check that the Prompt Language fits the Project's Market; a display translation does not change the wording measured by an Analysis.

5. Select **Preview prompts**. Review the proposed rows and the counts for new, duplicate, invalid, and over-limit Prompts. Preview does not save the batch.

6. Import only when the preview matches your intended library. Return to **Active** and confirm the new Prompts, Topics, and Tags.

A single import supports up to 1,000 Prompts. The File tab accepts CSV, XLSX, or XLS files up to 4 MB and reads up to 5,000 data rows, but a file with more than 1,000 Prompts must be split into separate imports.

> [!WARNING]
>
> If **Replace all existing prompts** appears for your role, it removes the current library before import. Use the ordinary import for additions. Check the preview carefully before any replacement.

## Organize and maintain the library

* **Topic** groups a Prompt by subject, product, service, or use case. It is required when you write a Prompt and should not be a Search Intent label.
* **Search Intent** classifies the question as Informational, Commercial, or Transactional. Use it to compare questions at different decision stages.
* **Tags** let you group Prompts for your own workflow. Add them in the import defaults or file, edit the Tags cell afterward, or use the available bulk action for selected Prompts.
* **Source** tells you how a Prompt entered the library. A generated question is a candidate for your review, not proof of real-world demand.

After a change, confirm the Prompts appear in the intended tab and that the Active count matches your Plan allowance. Archive irrelevant questions you may want to restore later. Record significant additions, removals, or wording changes when you interpret trends: a changed Prompt set can change the measurement even when your Website has not changed.

## If an import looks wrong

Check that the Prompt text and Topic columns were mapped correctly, that the language matches the questions, and that you have room for the Active Prompts. Review duplicate and invalid-row counts in the preview. Correct the input and preview again before importing; the preview does not save the batch.

For Plan limits, see [Plans](/billing/plans.md). To interpret the resulting measurements, see [Review your first Analysis](/how-to/run-your-first-audit.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.