Normally working with XLIFF in Translation Management System doesn't require additional configuration. In certain scenarios when you actually need to do this, here is the structure of an XLIFF Translation ZIP file generated by Scroll Translations:

<?xml version="1.0" encoding="UTF-8"?>
<xliff xmlns="urn:oasis:names:tc:xliff:document:1.2" version="1.2">
<file xmlns:ns0="k15t" datatype="xml" ns0:exportTimestamp="1486977717145" ns0:spaceKey="ALXDOCS" ns0:variantId="all" ns0:versionId="current" original="Users Guide" source-language="en" target-language="it">
<header>
<ns0:scrollHeader pageId="1179665" revision="3" scrollPageId="2EA377EC0145F04501137F043E495641" spaceKey="ALXDOCS" versionId="current"/>
</header>
<body>
<trans-unit id="ScrollPageTitle">
<source>Users Guide</source>
</trans-unit>
<group xmlns:ac="http://atlassian.com/content" xmlns:ri="http://atlassian.com/resource/identifier" restype="x-storageFormat">
...
</group>
</body>
</file>
</xliff>

Translation Descriptor

The attributes of the <file> element contain all the data of the translation descriptor (export specific data) like timestamp of the export, exported version and variant, source and target language and spacekey.

Scroll Header

The attributes of the <scrollHeader> element contain all the page specific data like pageId, revision, ScrollPageId, spaceKey and versionId

Scroll Page Title

The <trans-unit> element with the id "ScrollPageTitle" contains the original scroll page title of the exported page and needs to be translated

Page content

The <group> element with the restype "x-storageFormat" contains the actual page content which needs to be translated. 
It consists out of multiple <group> and <trans-unit> elements which represent the structure of the storage format and contain the text to be translated.