Creating your first Box and analysing documents

Last updated: 23 July 2026

What you'll do

This walkthrough ties the main pieces of Smartbox together so you can go from an empty workspace to a reviewed, exported document set. You'll create a Box, upload documents, let Smartbox analyse them, review what it found, redact anything sensitive, and export. Each step links to a more detailed article if you want to go deeper.

Smartbox is a powerful assist, not a replacement for your judgement. It surfaces likely sensitive data and duplicates so you can review far faster — but the final decisions stay with you.

Step 1 — Create a Box

A Box is a top-level workspace that holds your documents and folders. From your Boxes list, create a new one and give it a name (names must be unique within your account). You'll choose between two setups:

Default — the quickest start. The Box uses all of your organisation's configured term lists and pattern lists, so detection runs across everything your account has switched on.
Advanced — for more control. You can choose which detection categories and lists apply, set sharing and download permissions, and set a per-box page cap.

A new Box starts empty. Counts and analytics only appear once you've uploaded documents and they've finished processing. For more on the two setups, see Creating and configuring Boxes.

Step 2 — Upload your documents

Add files by uploading them into the Box. Smartbox converts everything to PDF so you get a consistent view and searchable text across formats. It handles a wide range of file types — Word and text documents, presentations, spreadsheets, images and scans, calendar files, web pages, and PDFs themselves.

Container files are expanded automatically: email containers (such as PST, EML, MSG, MBOX and OLM) break out into individual messages, and archives (ZIP, 7Z, RAR, TAR, GZ) are unpacked with their folder structure preserved.

Nothing is dropped silently. If a file can't be converted — no extension, an unsupported type, password protection, or a damaged archive — Smartbox keeps it and flags it so you can follow up. Very long documents are split into roughly 20-page parts, and oversized spreadsheet worksheets are flagged too.

Scanned documents need text recognition (OCR) before their content can be read. How much text comes through depends on the quality of the scan — a clean scan reads well, a poor one may leave gaps. Bear in mind that search looks at the extracted text only, so content that wasn't extracted from a poor scan may not be findable. See Uploading documents and supported file types for the full list.

Step 3 — Let Smartbox analyse

Once processing finishes, Smartbox reads the extracted text and automatically detects likely sensitive data. It combines a language engine that works across many categories, pattern matching for shaped types like phone numbers, emails, postcodes and card numbers (card numbers must pass a validity check), and face detection in images (it notes that a face is present, not whose it is). Detected items are grouped into readable categories such as Addresses, Names, Identifiers, Locations, Numbers, Organisations and Time.

Detection is only ever as good as the extracted text, so a poor scan can miss content. It's strongest in English, names are caught conservatively, pattern types can occasionally mis-match, and a document always proceeds even if one check couldn't finish — which means a partial result can occasionally look cleaner than it is. Always review, and see How sensitive-data detection works for the full picture.

You can extend detection with your own Dictionary terms and Regular Expression patterns for anything specific to your organisation.

Step 4 — Review culling results

Smartbox also helps you reduce volume. It groups exact duplicates by content fingerprint, flags near-duplicates, and can de-thread emails where a shorter message is quoted in full inside a longer one. It records a report of what it proposes to remove. Review that report before you delete anything — culling is about cutting noise, not losing evidence. See Reducing volume with culling for how each type of match works and where it stops.

Step 5 — Redact and finalise

To hide sensitive content, mark items — either draw redactions manually or bulk-redact whole detected categories — then finalise. Finalising flattens each redacted page to an image with the box baked in, so the text underneath is genuinely removed on those pages, while your original file is preserved separately. Redacted pages become non-selectable and non-searchable.

Finalising is irreversible, and bulk redaction only covers items Smartbox detected — so review before you commit. In rare cases a page that can't be rebuilt keeps its original text under the box, so always verify your exported PDF by trying to select and search behind the redactions. See Redacting documents for colour and exemption codes.

Step 6 — Export

When you're happy, export the reviewed set. Take a moment to open the exported PDF and confirm redactions are solid and the right documents are included. From here you can move into a compliance Workflow — see Running a Workflow — where a Disclosure step can package the result and deliver it via a time-limited link with delivery tracking.

Where to go next

Creating and configuring Boxes — Default vs Advanced in detail
Uploading documents and supported file types
How sensitive-data detection works
Reducing volume with culling
Redacting documents
Running a Workflow