What is dark data, and can unused business records have value?

Dark data is information a business collects, processes and stores in the normal course of work but never uses again for analysis, decisions or revenue. Typical examples are closed support tickets, former employees' email, old project folders, chat history and archived systems. Some of it can be licensed for AI training when the company holds the rights.

Dark data definition

Dark data is information a business collects, processes and stores during normal operations but never uses again for analysis, decisions or revenue. It is not lost or deleted; it sits in mailboxes, ticket queues, shared drives, chat workspaces and backups that nobody opens.

Most dark data is unstructured: messages, documents, recordings, notes and logs rather than tidy database rows. That is why it stays dark. It is hard to search, nobody owns it, and keeping it feels safer than deciding what to delete.

Examples of dark data in a typical company

Where it sits	Examples	Why it goes dark	Why AI developers may value it
Help desk or support platform	Closed tickets with resolution notes	Teams only watch the open queue	Real problems solved step by step, with an outcome
Email archives	Mailboxes of former employees kept for retention	Nobody opens them after offboarding	Negotiations, approvals and escalations in context
Slack or Teams	Channels from finished projects	Search rarely reaches back years	How people coordinate multi-step work
Shared drives	Proposals, SOPs, project folders	Migrated in bulk with no index	Documents tied to decisions and results
CRM	Notes on lost deals and old accounts	Reports cover only the current pipeline	Deal histories with won and lost outcomes
Engineering tools	Old issues, pull requests, code reviews	Repositories archived after a release	Changes with the reasoning behind them
Call recordings	Calls kept for quality assurance	Sampled once, then stored	Real conversations with results, where recording notices were given
Retired systems	Backups of a replaced ERP or help desk	Kept only for compliance	Years of history that exists nowhere else

Dark data vs big data and related terms

Term	What it means	How it relates to dark data
Big data	Very large, fast-moving or varied datasets analyzed at scale	Describes size; dark data describes whether data gets used
Unstructured data	Text, audio and images without a fixed schema	Most dark data is unstructured
ROT data	Redundant, obsolete or trivial files	The part of dark data worth deleting
Data swamp	A poorly governed data lake	A common place for dark data to collect
AI-ready data	Data that is organized, documented and rights-cleared for AI use	What some dark data can become after an inventory and a rights review

Why companies end up with so much dark data

Retention rules and legal holds require keeping records long after their working life.
Migrations copy everything from the old system, just in case, without an index.
Tool sprawl: strong companies often run 10-15+ systems, each with its own archive.
Turnover: when the person who understood a folder leaves, the folder goes dark.
Nobody wants to own the decision to delete.

The cost usually stays invisible until a migration, a security review or a litigation hold forces someone to look: storage bills, exposure if an archive is breached, and the effort of searching material nobody has indexed.

Can dark data be monetized?

Some of it can, through licensing rather than selling, when the company owns the records and can still export them. The demand comes from AI developers moving from models that answer questions to agents that carry out tasks. Training and evaluating agents needs records of real work, such as multi-step workflows, decisions, tool use and outcomes, and that material is thin on the public web.

Researchers at Epoch AI estimated the effective stock of public human-generated text at roughly 300 trillion tokens and projected that, if current trends continue, language models could fully use it between 2026 and 2032 (Epoch AI). It is a forecast with wide uncertainty, but it helps explain why permissioned, non-public business records have become a scarce input. For the wider picture of earning from data, see what data monetization is.

In a license arranged through SourceX, the company keeps ownership, approves the scope and price, and receives one all-in price as a one-time payment after a buyer selects the data. Nothing is binding until it signs.

What to check before treating dark data as an asset

The company created the records itself; material it processes for clients usually belongs to those clients, as the comparison of data controller vs data processor explains.
The archives still exist, and someone can export them.
The history spans several years and several systems.
Privacy policies, customer contracts and employee notices do not rule the use out.
The records are not mainly consumer personal data or protected health information.
The same records have not already been licensed for AI training.
The company has 50+ full-time employees at peak (contractors excluded) and an owner or executive willing to consider an exclusive license.

Old promises still count. FTC technology staff wrote in January 2024 that commitments not to use customer data for undisclosed purposes, such as training or updating models, are enforceable whether they appear in privacy policies, terms of service or promotional materials (FTC staff post). That is staff guidance rather than a rule, but it is a good reason to reread past privacy promises before any customer-facing records go into scope. This is general information, not legal, tax or financial advice.

Dark data also outlives the business that created it. If a company enters bankruptcy, its archives may be property of the estate, and whoever controls the estate decides what happens to them. The data inventory builder helps an owner list systems and the records each one holds before any decision to delete or license.

Illustrative example

Illustrative, fictional company: a 210-person IT services firm replaced its help desk platform four years ago and kept a read-only export of nine years of closed tickets, along with the mailboxes of staff who had left. Leadership saw the archive as a storage cost. An inventory showed the tickets linked to engineering changes and customer outcomes, and that the firm had created them under its own contracts. The owner applied, counsel identified which client-related content had to be redacted or excluded, and those rules were agreed before any work began.

Next step

Before deleting an old archive or cancelling a legacy tool, check whether it holds years of the company's own operational history. Owners can compare their company with the who qualifies baseline and apply at sourcex.si/apply; advisors who spot dark data in clients' systems can register as a partner and introduce the owner.

Step 1Share your linkSend your personal link to a company you know.
Step 2Company appliesThe company applies itself at /apply.
Step 3Buyer selects and paysThe buyer selects and pays for the data and SourceX receives its fee.
Step 4You get your rewardYour share of SourceX fees becomes payable.

Common questions

Is dark data the same as unused data?

Mostly, yes. Dark data is stored information that nobody uses once its first purpose is served. The word dark stresses that it is often unknown as well as unused: leadership may not know which archives exist, how far back they go or what they contain, which is why an inventory comes before any decision to delete, analyze or license it.

Should a company delete its dark data?

Delete only what retention rules, legal holds and business needs no longer require, and only after checking what it is. Redundant, obsolete or trivial files are good candidates. Long operational histories, such as closed tickets, project records and former employees' email, may be worth keeping, because once an archive is deleted it cannot be analyzed or licensed later.

Is dark data a security risk?

It can be. Old archives often hold sensitive messages and personal information under weaker access controls than live systems, and nobody may be monitoring them. That is a reason to inventory and govern dark data, not necessarily to delete all of it; the goal is to know what is kept, why, who can reach it and how long it stays.

Can a small company license its dark data?

SourceX introductions require US companies with 50+ full-time employees at peak (contractors excluded), several years of documented operations, rights to license the records and an authorized sponsor. Smaller companies can still gain from understanding their dark data for analytics or cost control, but they fall outside the licensing baseline for now.

Who usually discovers dark data inside a company?

It surfaces when someone has a reason to look: an IT provider planning a migration, a finance lead reviewing storage and software costs, an advisor preparing a sale or restructuring, or a new executive asking where old project history lives. Each of those moments is a good time to record what exists before systems are retired.

Free resources

Earnout scenario calculator — Probability-weighted earnout value and its present value.
Profit margin calculator — Profit and margin across three scenarios.
Client opportunity brief generator — An editable intro email, summary and checklist.
All free tools · MCP resource center

By SourceX Partnerships Team · Published 2026-10-09 · Updated 2026-10-09

Know a US company with valuable proprietary data?

Become a referral partner from anywhere we support, get your link and introduce an owner or authorized decision-maker.

Refer a company →

I own a business

Explore licensing your company's data to AI developers worldwide. Start a short assessment; no uploads needed.

Start an assessment