# Where an answer or a score comes from

How to see the sources behind an answer, where a value on an account came from, and how a score is put together.

## Sources

A source is one thing Justwon read from a connected tool: a deal, a ticket, a message, a meeting, an email, a note. Everything Justwon says about an account rests on sources, and a source keeps a link back to where it was read.

To see what Justwon has read on an account, open the account under **Accounts**. Its page lists the sources read on that account, with the tool, the kind of source and the date. To see what one connection has read, open it under **Integrations** and select its **Sources** tab.

## An answer in the agent panel

When an answer draws on sources, they are listed with it, and you can open them. The answer's footer shows how many sources it used.

![The agent panel under an answer: the footer shows six sources, each listed with the tool it came from and its date.](https://justwon.ai/docs-shots/answer-sources.png)

Select **View quality details** to see how Justwon rated the answer. The rating is called **Answer quality**. It has three parts:

| Part | The question it answers |
| --- | --- |
| Accuracy | Did the response correctly answer your question? |
| Grounding | Is the answer based on retrieved sources? |
| Completeness | Did the response cover all aspects? |

The rating is Justwon's own reading of its answer. It tells you how much to rely on the answer, and it can be wrong. Open the sources when the answer matters.

## Where a value on an account comes from

A fact on an account, such as an industry, an owner, a renewal date or the people involved, shows **Where this comes from**. It lists the evidence behind the value:

- **Supporting evidence** is what agrees with the value. Each item names its source and how sure Justwon is: high, medium or low confidence.
- **Conflicting evidence** is what disagrees. When some sources claim otherwise, the count is shown, and you can open them.
- **Previous values** are the values it held before.

Each value also says how it got there: **A synced record** read from a tool, **The agents' conclusion**, or something a person in your team changed or confirmed.

Justwon shows a source's own words only when it holds a verified quote. When it only has a summary, the summary is marked **Summary, not a quote:**. A value with nothing behind it says "No source recorded for this." A value whose source Justwon cannot link says **Source not on record**.

## How a score is traced

A score on an account, such as health, shows as a ring with a word: Healthy, Watch or At risk. A score marked **Early read** rests on evidence with low confidence, so treat it as a first reading.

To see what a score is made of:

1. Open the account under **Accounts**.
2. Select the ring. Its tooltip reads **Open breakdown**.
3. The dialog shows what the score is made of, the points each part holds back, the sources behind it, and the rule it was read against.

For account health, the main part is **Stakeholder health**, which is made of **Role coverage**, **Engagement recency**, **Champion strength**, **Value validation**, **Breadth** and **Sentiment and stability**. If a ceiling holds the score down, the dialog says what the ceiling is and what the score would be without it.

The dialog has a section called **Read against the Playbook**. The health score is read against the guideline called **Customer health score model**, which you can open under **Playbook**.

## When you cannot open a source

When a connection respects the other tool's permissions, you only see evidence from sources you could see in that tool. You can still see that a value rests on sources you cannot open, and how many. See [data is missing](https://justwon.ai/docs/troubleshooting/data-missing).