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

# Signal Evidence

> What sits behind a signal, and how to read it before you act.

Every signal carries the evidence it was raised from. This is the part to read before you act, and usually the part you will quote in the customer conversation.

## What evidence looks like

Depending on where the signal came from:

* **Call quotes** -- the actual sentence, who said it, and which call
* **Metric values** -- the number, the comparison period, and the direction
* **Support tickets** -- the specific tickets, their status, and their theme
* **Email** -- who stopped replying, and when the thread went quiet
* **CRM records** -- the field that changed, and what it changed from
* **External records** -- the funding round, the executive change, the public event

## Reading it before you act

**Check the evidence actually supports the claim.** Signals are generated from real data, but the reading can still be wrong -- a name matched to the wrong company, a metric measured over an unintended window, a quote that means something different in context.

**Check the dates.** A signal raised from a call six weeks ago may already have been handled.

**Check the subject.** Externally-sourced signals about funding or executive changes depend on matching your customer to a public company record. When the match is wrong the signal is confidently about the wrong business entirely. The company name in the evidence is where to catch that.

<Tip>
  The evidence is usually the most valuable part of a signal for the customer conversation. "I noticed your contract volume is down about half since Q2" lands very differently from "our system flagged you as at risk".
</Tip>

## When evidence looks thin

A signal with weak evidence behind it is worth dismissing, not acting on. It is also worth mentioning to your Superhawk representative -- detection rules are tuned from exactly this kind of feedback.

## Related

* [Signals](/signals/overview)
* [Signal Lifecycle](/signals/lifecycle)
