How to sell customer support tickets for AI training

A resolved support ticket is a worked example: a customer's problem, the steps an agent took and a result you can measure. That makes ticket archives useful for training support assistants and agents that take actions, and for building evaluation sets with known answers. Labelbox Prism says it prefers high-volume consumer companies when it buys support logs.

Who buys support ticket data for AI training?

Support ticket data is bought by training-data companies and direct buyers that license operational records to AI labs. Their published terms differ on price, size and exclusivity.

Exclusivity is the term to compare first. Labelbox Prism licenses exclusively by default, while Endless Data licenses non-exclusively, and our micro1 profile explains its two published payment timelines.

How much are support tickets worth?

Support tickets have no published price per ticket. The figures available are company-level payouts that cover support data alongside other records.

• micro1: $100,000 to $500,000 for a qualified enterprise partnership, $500,000 to $3 million for large-scale or ongoing datasets
• Labelbox Prism: from around $50,000 for smaller targeted datasets to more than $1 million for large enterprise partnerships
• Mercor: $20,000 to $250,000 for call centers with 50 to 5,000 agents, for recorded customer conversations, as reported by ddiy.co in October 2026

These are payouts to data owners, not quotes. Ticket volume, years of history and how many tickets carry a clear resolution decide where an archive lands within any range.

What makes a ticket archive more valuable?

A ticket archive is more valuable when each ticket keeps its full thread, its outcome and the fix that closed it.

Bot replies matter more than sellers expect, because replies from an old chatbot teach a new model to imitate that chatbot. Our page on support ticket data sets out what AI teams check on delivery.

You export tickets from Zendesk as JSON, CSV or XML, according to Zendesk's help center, but only some formats include the conversation.

Zendesk says data exports are disabled by default, so the account owner must ask Zendesk Customer Support to enable them, and the export tools are not available on Team plans. Customers on all plans can use the Zendesk REST API instead. Other helpdesks have their own exports: whichever you use, keep ticket IDs, message order, author roles and timestamps.

What privacy and contract issues apply to support tickets?

Support tickets are full of customer personal information typed into free text, so the privacy promises you made to customers decide what you can license.

• Privacy notices: FTC staff wrote on February 13, 2024 that quietly adopting more permissive data practices through retroactive changes to terms or privacy policies may be unfair or deceptive
• CCPA: Frankfurt Kurnit Klein and Selz notes that a paid transfer of personal information is a sale unless it meets the statute's de-identification standard
• Card data: the PCI Security Standards Council says sensitive authentication data, including card verification codes, must not be stored after authorization even if encrypted
• Health information: if you are covered by HIPAA, HHS allows de-identification by expert determination or by removing 18 identifiers
• Business customers: if you sell software, tickets often contain your customers' own data, and your contracts with them govern its use

We are a broker, not a law firm. Take independent advice on your customer terms, and see our guide to UK GDPR, EU GDPR and CCPA when selling data for each regime.

How do you de-identify support tickets?

De-identifying support tickets means scrubbing the free text and attachments, not only the structured fields where names and emails are easy to find.

• Replace customer names, emails, phone numbers and addresses with consistent tokens
• Tokenize order numbers, account IDs and URLs that contain them
• Redact internal notes, which often hold more detail than customer-facing replies
• Remove signatures from tickets that arrived by email
• Exclude screenshots and attachments, or redact the images themselves
• Pseudonymize agent names so staff are not identifiable

Our vetting standard requires a documented de-identification method and a re-identification test on a random sample before any buyer sees a ticket.

If you hold UK or EU data

If you sell software to businesses in the UK or EU, you are often their processor for the personal data in tickets. Article 28(3)(a) of the UK GDPR says a processor processes personal data only on documented instructions from the controller, so licensing those tickets needs your customers' agreement. Tickets from your own consumers need a lawful basis or anonymization.

How do you sell support tickets step by step?

1. 1

   Size the archive

   Count tickets by year, channel, queue and language, and the share with a resolution and satisfaction score.

2. 2

   Choose the queues

   Keep technical, billing and product queues. Leave out queues for health, legal complaints or vulnerable customers.

3. 3

   Check notices and contracts

   Read the customer privacy notice in force when tickets were opened, and any business customer contracts.

4. 4

   Export full threads

   Use JSON, XML or the API so comments, internal notes and knowledge base links come through.

5. 5

   De-identify and test

   Scrub text and attachments, then test a random sample for re-identification.

6. 6

   Write the manifest

   Record counts, date ranges, fields and exclusions. See how to build a data manifest.

You approve every buyer, every sample and every license term, and you can withdraw before signing. Book a free consultation to find out what your ticket archive could be worth.