Skip to content
Outrich Data

Data type

License your customer success records for AI training

Customer success records can often be licensed for AI training after accounts and contacts are de-identified. Outrich Data is the direct buyer of health score histories, playbook runs, success plans and renewal outcomes, and pays you for a licensed copy. Customer NDAs and master agreements are checked first, and restricted accounts are removed before anything is assembled.

What counts as customer success records

Customer success records track what happens after the sale. For each account they hold the onboarding plan, adoption milestones, a health score that moves over time, risk flags a manager raised, notes from business reviews, playbooks that were triggered, escalations and the renewal, expansion or cancellation at the end of the term.

They differ from support tickets and from CRM pipeline data. A ticket is one problem. A pipeline record is one deal. A success record is the long view of a relationship across years, mostly in a customer success manager's own words. Accounts you manage on behalf of another vendor under a reseller or outsourcing arrangement are that vendor's records and are excluded.

Why AI labs want it

Predicting which customers will leave is a common commercial use of machine learning, and good training data for it is private by nature. A health score history shows what the vendor believed about an account each month. The renewal result shows whether that belief was right. AI labs want both halves.

The interventions are just as useful. A manager saw usage fall, ran a re-engagement playbook, brought in an executive sponsor and saved the renewal, or did not. Those records describe a judgment, an action and a result in a business setting. Labs use them to train account management assistants and to test recommendations against real outcomes.

Removed before licensing

What comes out first.

  • Customer company names, logos, domains and industry details narrow enough to identify them
  • Names, titles and contact details of customer stakeholders and champions
  • Customer success manager and executive sponsor names
  • Contract values, renewal amounts and discount terms
  • Customer strategy, roadmap and budget details shared in confidence during reviews
  • Verbatim survey comments that identify a person or company
  • Usage screenshots and exports embedded in review decks

Data title search

What gets checked.

  • Master agreements and NDAs. What a customer tells you in a business review is often their confidential information. Check each agreement's definition and permitted uses.
  • DPAs. Stakeholder names and survey responses are personal data you may hold as a processor or as a controller, depending on the field. The roles affect what can be licensed.
  • Survey terms. Check what respondents were told when feedback was collected, including any promise that responses were for internal use.
  • Reference and publicity clauses. A customer that agreed to be a named reference has not agreed to data licensing. The two permissions are separate.
  • Regulated customers. Banks, hospitals and public bodies frequently impose stricter confidentiality terms on their vendors.

Who signsThe chief customer officer or head of customer success sponsors it. The general counsel approves after reviewing customer agreements, and the success platform administrator exports the records.

Value

What moves the price of customer success records.

FactorEffect on value
Score history with outcomesHealth scores tracked over time and tied to the renewal result are generally worth more than a current snapshot.
Recorded interventionsPlaybook runs and save plans show what was done about a risk, not only that it existed.
Narrative notesManager notes explaining why an account was flagged carry reasoning that a score cannot.
Churn coverageLost accounts with reasons are essential for balance and are often the scarcest part.
Agreement clearanceThe proportion of accounts whose agreements permit the use sets the usable volume.

Industries where this is common

Questions

Licensing customer success records: what companies ask.

Our QBR notes contain customers' internal plans. Can those be used?
Not in that form. Information a customer shared in confidence is removed, and accounts with strict NDAs are excluded outright. What can remain is the generalized pattern: the kind of risk, the action taken and the result.
Is a health score meaningful outside our company?
By itself, not very. With its component inputs, the formula or weights and the eventual outcome, it becomes a labeled prediction that a lab can learn from or test against. Including a short description of how the score is calculated helps.
We have few churned accounts. Does that reduce interest?
It can, because models learn from contrast. Downgrades, late renewals and accounts that were saved after a risk flag also count as informative outcomes, so a low churn count is not disqualifying.
Will our customers find out?
Your company is not named to buyers, accounts are coded and identifying details are removed, with the aim that no customer can be picked out of the dataset. Whether you must or should notify customers depends on your agreements. Outrich Data raises the question and your counsel answers it.

Find out if your data qualifies.

Eleven questions, about five minutes, no call required. You get an answer on screen and a written follow-up.

esc