Data export size estimator
Before you migrate systems, answer a legal hold or hand records to a buyer, you need to know how big the export will be. Enter what your company uses for email, chat, files, CRM, help desk and databases, and the estimator works out the raw and compressed size, the number of files and how long the transfer will take. Every assumption is visible and editable, so you can swap in your own averages once you have a sample export.
Size by source
Where the gigabytes come from. Plain text is the readable content left after attachments, markup and metadata are removed.
| Source | Items | Raw size | Exported | Plain text | Share of export |
|---|---|---|---|---|---|
| Total |
How the estimate works
The estimator counts how many items each system holds, gives every item an average size, then applies the overhead of the export format and the saving from compression. The steps run in this order for every source, and the totals are added up at the end.
The default sizes come from typical exports of Microsoft 365, Google Workspace, Slack, HubSpot and Zendesk accounts. Open the assumptions panel to replace them with your own numbers.
-
Count the items
Email and chat are counted from people and time, because few companies know their message totals. Files, CRM records, tickets and rows are entered directly.
emails = mailboxes × emails per day × working days × years
-
Give each item a size
Each item gets the average size from the assumptions panel. Attachments are added only for the share of emails and tickets that carry one.
email size = count × (14 KB + 22% × 420 KB)
-
Apply the export format
EML and MBOX encode attachments as Base64 text, which adds 37 percent. PDF renders every email as a page, about three times the text size. Tables grow by 1.7× as JSON and 2.3× as XML, and shrink to about 0.35× as Parquet.
-
Compress
Text and tables shrink by the ZIP saving you set, 80 percent by default. Attachments and documents are already compressed formats, so they only lose about 3 percent.
exported = text × (1 − 80%) + binary × 0.97
-
Work out time, cost and storage
Transfer time uses 80 percent of your upload speed. Storage cost is the exported size in GB times your monthly price. "Fits on" picks the smallest common drive that holds it with 10 percent headroom.
Questions about data export sizes
How accurate is a data export size estimate?
Expect the real export to land within about 30 percent of the estimate when your counts are right. The biggest swing factor is attachments: a team that mails design files or scans will sit far above the average, while a team that shares links instead of files will sit below it. If you can, export one mailbox and one month of chat first and put the real averages into the assumptions panel.
Why is my email export so much bigger than the mailbox size shown in Outlook or Gmail?
Mail servers store attachments in compact binary form and often keep one copy of a file that was sent to many people. EML and MBOX exports write every attachment into every message as Base64 text, which adds about 37 percent, and each recipient gets their own copy. PST keeps attachments binary, which is why it comes out smaller.
Which export format is smallest?
For tables such as CRM records, tickets and database rows, Parquet is usually the smallest because it is a compressed, column-based format, often a third of the CSV size. CSV is the most compact plain text format. JSON repeats every field name on every record and XML adds opening and closing tags, so both are noticeably larger. Documents and attachments stay roughly the same size whatever format you pick.
How much does zipping a data export save?
Text compresses very well. Email bodies, chat logs, CSV and JSON typically shrink by 75 to 90 percent in a ZIP or GZIP archive. Files that are already compressed, such as PDF, JPG, MP4, DOCX and XLSX, barely shrink at all. That is why an export full of attachments only gets a little smaller when zipped.
How long will it take to upload or transfer the export?
The calculator divides the export size by your connection speed and assumes about 80 percent of the advertised speed is usable. A 1 TB export takes roughly 28 hours on a 100 Mbps upload. Above a few terabytes, shipping an encrypted drive or using a cloud provider's import service is often faster than uploading.
What is the plain text figure and why does it matter?
It is the readable words inside the export once attachments, formatting and metadata are stripped out. This is the part that search tools, e-discovery reviews and AI training actually use, and it is usually a small fraction of the full export. The token count is the plain text divided by four, the rule of thumb used by most AI models.