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

# Claims

> The facts you state about your brand, what each engine says about them today, and since when.

Claims answers: what do the engines say about us, and is it true? You track a fact about your brand, such as a price, a free trial, an integration or a certification, and Claims shows what each engine says about it, since when, and from which source.

## Where claims come from

* **From your site.** When a project is set up, Citino reads your website and drafts the claims it states. They appear like any other claim. A project with none can use **Find claims on your site**, which reads it again.
* **From you.** **Track a claim** adds anything the site reading missed.

Each claim is read in the stored answers that name you. A new claim does not wait for new answers: its history is rebuilt from the answers already stored, and the screen shows how far that has got.

## The toolbar

* The scope line: a summary of your claims, the engines, the dates covered, and "no confidence band".
* **Engine** shows one engine’s readings, or **All engines**.
* **Order**: **Unreviewed first** or **Oldest first**.
* **Track a claim** opens the form described below.

## The list

One row per claim: its topic, the statement, and a dot per engine coloured by what that engine says. Claims carry no `±` range, unlike visibility: each reading is a count of the answers that stated it, not a sampled estimate, so read a small count with care.

## An open claim, top to bottom

**1. The claim**

Its topic, the day you started tracking it, and, while the engines disagree, how far they diverge. **You told us** shows what is actually true. **Edit** and **Archive** sit here.

**2. Your verdict**

**Correct**, **Differs from my claim** or **Wrong**. A fourth choice, **Nothing was said**, is offered only when no engine stated anything on the claim in the window. Choosing **Wrong** opens a correction work item in [Gaps](https://docs.citino.ai/guides/screens/gaps.md), with the claim, the engines that state it wrong or differently, how many answers carry it, and what is actually true. If the evidence under a verdict changes later, the claim says so and on which day; the verdict stands until you change it.

**3. What each engine says**

One row per engine: its reading, how many of its answers match, since when, the sentence it used, and the site it came from. **Read the answer** opens it in Answers; **See the prompt** opens the question it answered.

**4. History**

What each engine said about this claim, day by day. One line per engine steps between the readings below. Engines that agree overlap, so agreement reads as one line and a change reads as a fork. A break in a line is a day it said nothing we could read. Days a correction was published and days an engine drifted are marked.

### The four readings

| Reading | What it means |
| - | - |
| matches your claim | The engine states the fact as you do. |
| differs from your claim | The engine states it differently, for example in another unit or an older version. It is not necessarily wrong, so it is not called wrong. |
| contradicts your claim | The engine states something that conflicts with it. |
| not stated | The engine answered and named you, and said nothing on this topic. That is a finding, not missing data. |

## Track a claim

* **Topic**: Pricing, Trial, Integrations, Trust, General or Fit. A different topic is a new claim.
* **The claim**: one fact the answers could state or contradict.
* **The fact itself**, for the topics that have one: a price’s amount, currency and unit; a trial’s length in days; the other product, for an integration; the certification or standard, for trust. General needs none: its statement is the fact. Fit is never judged automatically; the answers that touch on it are kept for you to read and judge.
* **What is actually true**: what a buyer should hear instead. It is carried into the correction when an engine gets it wrong.

On save, the claim shows what it will look for in the answers, so you can correct it at once, and its history starts rebuilding.

**Edit**: a change of wording keeps the claim and its history. A change to the fact itself starts a new claim; the old one stops being tracked and keeps its history.

**Archive** stops tracking a claim and keeps its history. It is not a resolution: it does not mean the engines now have it right.

## Who can do what

Owners, admins and editors can track, edit and archive claims and set verdicts. **Find claims on your site** is for an owner or admin, because it reads your site again. Viewers read every claim and its history.

## Empty and error states

* **No branded prompts yet**: claims are read from the answers to questions that name you. Add one and the answers start carrying claims.
* **No claims tracked yet**: **Find claims on your site**, or **Track a claim** yourself.
* **History is being rebuilt**: a new claim shows how many stored answers have been read so far, and back to which day.
* **No claim has a reading from this engine yet**: none of its answers in the window named you on a tracked claim. **Show every engine** clears the filter.

## Related

> [card] [Gaps](https://docs.citino.ai/guides/screens/gaps.md)
>
> Where a wrong claim becomes a correction.

> [card] [Answers](https://docs.citino.ai/guides/screens/answers.md)
>
> The answers each reading comes from.

> [card] [Alerts](https://docs.citino.ai/guides/screens/alerts.md)
>
> Get told when a claim changes.

> [card] [Drift labels](https://docs.citino.ai/guides/drift-labels.md)
>
> What the drift marks on the history mean.


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