Does my company's data qualify?

Your data qualifies if the records still exist in your systems, they show real work being done (requests, approvals, jobs, invoices), and your company clears a provider’s floor. As of October 2026, published floors run from no minimum to a preferred 40 employees, and Handshake AI also asks for three years in operation.

What does each provider require?

Published floors run from none at Appen, Mercor, Polyshares and Troveo to 20+ full-time employees and three years at Handshake AI, 30+ employees at micro1 and a preferred 40+ at Scale AI. The table gives each provider’s own rules, as of October 2026. “No published minimum” means Data Licensing Report checked the provider’s pages and found none. A calculator’s input range is shown where it helps, but it is not an eligibility rule.

Provider	Company size	Years of history	Geography	Records it names
Appen	No published minimum	No published minimum	No country rule on the program page	Process documents, decision and approval records, systems of record (source)
Avelence (introducer)	No published minimum to submit a profile; calculator takes 20 to 10,000 employees (source)	Calculator takes 1 to 50 years	Profile form lists the US, UK, Canada, several EU countries, Australia and others; none stated as ineligible (source)	Slack, Gmail, Outlook, Jira, Salesforce, HubSpot, Zendesk, QuickBooks, NetSuite, GitHub (source)
Handshake AI	20+ full-time employees	3+ years in operation	Estimator offers USA, Canada, Europe, Other (source)	Gmail, Outlook, Drive, Dropbox, Slack, Notion, NetSuite, QuickBooks, ServiceTitan, AutoCAD (source)
Mercor	No published minimum	No published minimum	No published rule	100+ tools by connector, such as Slack, Teams, Gmail, Outlook, Drive, Salesforce; eight named industries (source)
micro1	30+ employees; its application page says 30 to 200	Asks for “operationally mature” companies with established documentation	Mainly companies that operate in English; demand strongest in the US, UK and Canada	SOPs, process documents, knowledge bases, templates, CRM data, project histories (source)
Polyshares	No published minimum	No published minimum; asks for “real operating history”	No geographic restriction published (source)	Slack, email, Drive, spreadsheets, CRM, tickets, design files, call recordings, accounting and ERP exports (source)
Replay	No explicit minimum; its valuation form accepts 1 to 9 people, while price tiers start at 20	Form accepts companies under two years old	No country rule stated; form asks what share of communications is in English	Internal docs, decision threads and messages, billing, support history, CRM (source)
Scale AI	Prefers 40+ employees, assessed case by case	Asks for years of records in established systems	No published geographic rule	Slack, Teams, Gmail, Outlook, Drive, SharePoint, Salesforce, HubSpot, SAP, Zendesk (source)
Telegraph Lab (affiliated with this site)	None stated for direct applicants; referral pages say 20+ or more than 30 employees	Referral pages say 3+ years	No country rule for direct applicants; partner introductions must be US-based	Slack, Teams, Gmail, Salesforce, SAP, Zendesk, drive files including CAD, CSV exports and scans (source)
Troveo	No published minimum; calculator runs 20 to 1,000+	No published minimum; calculator runs under 3 to 15+ years	No country rule; asks which languages the data uses	Chat, email, documents, work tools, CRM and support, finance, data and code (source)

• If you have 20 or more full-time employees and three years in operation, you clear every published size floor for direct applicants except micro1’s 30 and Scale AI’s preferred 40.
• If you have fewer than 20, start with the providers that publish no minimum (Appen, Mercor, Polyshares, Troveo), Telegraph Lab for direct applicants, Avelence, which says its calculator’s limits do not decide who may submit a profile, and Replay, whose form accepts companies of 1 to 9 people.
• If your company is outside the US, check geography first: Telegraph Lab’s partner introductions are US-only, and micro1 says its demand is strongest in the US, UK and Canada (micro1).
• If most of your records are in another language, micro1 says it is looking mainly for companies that operate in English, and Replay and Polyshares ask what share of your data is in English (Replay, Polyshares).
• If your industry is not on a provider’s list, ask before ruling it out. Mercor names eight industries, from legal to logistics, and construction is not among them; Polyshares, Telegraph Lab, Scale AI and Troveo name construction or field services (Mercor, Polyshares, Telegraph Lab, Scale AI, Troveo).

What counts as data?

Anything your people do their work in. Email, chat, shared drives and business software are the core, and records such as bids, photos and call recordings count when they sit alongside the jobs they belong to. As of October 2026, providers name these:

Record	Usually lives in	Named by
Email	Microsoft 365, Google Workspace	Avelence, Handshake AI, Mercor, Polyshares, Scale AI, Telegraph Lab, Troveo
Chat	Teams, Slack	The same seven, plus Replay
Bids, estimates, drawings	Shared drives, estimating software exports	Telegraph Lab (PDF, DOCX, XLSX, CAD), Handshake AI (AutoCAD, Dropbox, Drive)
Jobs, invoices, accounting	ServiceTitan, Procore, QuickBooks, NetSuite	Handshake AI (ServiceTitan, QuickBooks, NetSuite); Polyshares publishes guides for ServiceTitan and Procore
Call recordings	Phone system, field-service software	Polyshares; LH2 AI Labs lists recorded calls and meetings
Photos and drone footage	Phones, drives, job records	Troveo’s separate content-owner program lists drone footage among its video categories; see the video and audio list
Procedures and templates	Drives, wikis	micro1, Appen

Two practical limits. ServiceTitan’s help center describes downloading call recordings one call at a time, as MP3 files (ServiceTitan), so a large recording archive can be a bigger export job than the email. And recordings carry voices, which raises consent questions the video and audio list sets out.

Some records usually stay out. Polyshares says data a company holds for its own clients is out of scope, and Telegraph Lab says regulated, privileged or third-party information needs extra review and may be excluded. Which documents to check is covered in Is it legal to sell company emails?, and how to carve records out in Can I leave out certain customers, people or folders?

We switched from an old server to Microsoft 365 in 2017. Is the older stuff still there?

Only if it was moved, and the oldest message in a long-serving employee’s mailbox will tell you. Microsoft’s migration guide says cutover and staged migrations from an Exchange server move “All email, calendar, and contacts from user mailboxes”, while an IMAP migration moves only mail folders, not contacts or calendars. Microsoft also offers a separate service for importing old .pst files, and users can import their own.

So the pre-2017 years end up in one of three places:

• In Microsoft 365, with their original dates, if the whole mailbox was migrated.
• In .pst files on old PCs or a shared drive, if people kept archives locally or the move carried only recent mail.
• On the old server or its backups, if anything still exists.

To find out:

1. Open the oldest email in two or three mailboxes of people who were there before 2017.
2. Ask your IT firm which migration method it used and whether it set a date cutoff.
3. Ask whether the old server, its backup drives or any .pst archives still exist, and make sure nobody wipes them while you decide.
4. Check whether the old shared folders were copied into SharePoint or OneDrive, or left behind.

Write the real oldest date in your data inventory, not the founding year. A gap at the migration year will show in any sample, and Data Licensing Report’s free sample tool counts a mailbox export by month, so you can see the gap before a buyer does. The tool reads .mbox and .eml files; Outlook .pst files need to be saved as .eml first.

Ex-employee mailboxes are deleted. Does that kill the value?

No, it narrows it. A deleted mailbox loses that person’s own copy, but every message they exchanged with colleagues who are still at the company normally sits in those colleagues’ mailboxes, and their jobs, invoices and files usually remain in the other systems. Microsoft stores Teams chats in a hidden folder in the mailbox of each user in the chat, so a departed user’s chats also sit with the other participants. What is gone is their correspondence with customers and suppliers that nobody still there was copied on.

How much is recoverable depends on how long ago it happened:

• Microsoft 365: when a license is removed or the account deleted, email is kept for 30 days and then permanently deleted, unless a hold or retention policy was applied first.
• Google Workspace: deleting a user deletes their Gmail and owned Drive files, which are kept for 20 days but are only accessible if you restore the user.
• Slack: deactivating a member does not delete their messages or files.

To stop further losses in Microsoft 365, convert a leaver’s mailbox to a shared mailbox, which Microsoft says keeps all existing email and calendar data and can go without a license under 50 GB, as long as the account itself is not deleted. In Google Workspace, an Archived User license keeps a former employee’s data.

Is small business data worth anything?

Sometimes, if the record is deep and specialized; but most published floors and price tiers start at about 20 employees. Polyshares says “Nobody is indexing by company size”, and a Troveo guide says “a smaller company with rare, specialized workflows can be worth more than a larger generic one”. idler’s Y Combinator profile invites companies with 10 to 100 employees.

The published figures at the small end disagree with each other. Replay’s tiers show $10K to $100K for 20 to 50 employees, while Avelence’s calculator shows $160K to $500K at its default inputs of 20 employees, 10 years of records and a US headquarters, a figure it labels a model estimate, not a buyer offer (both as of October 2026). Both are the providers’ own statements, not completed deals. How much is my company’s data worth? sets the published ranges side by side.

Why depth can count for more than headcount, and which records count for most, is in Why would AI companies want our emails? How much work is it? covers what licensing would ask of your office and your IT firm.