How VMS holding companies are rethinking data in the AI era

How are VMS holding companies rethinking data?#

VMS holding companies are rethinking data by treating the records inside acquired products as something to govern across the group, not just a by-product of running software. For most of the buy-and-hold era, acquirers valued vertical software for recurring revenue, sticky customers and modest capital needs; data was a feature of the product, not a line in the investment case.

Generative AI changed two things at once. Customers now expect AI features in the specialized tools they use every day, and model developers now look for real operating records from specialized industries. Both pull holdco leaders toward the same question: what records do we actually hold across the group, and what are we allowed to do with them?

Four holdco approaches, as of October 2026#

Four broad approaches to data appear across VMS holding companies as of October 2026, and many groups combine them. The table below describes patterns, not any one company's strategy, and is dated because the field is moving quickly.

The decentralized buy-and-hold model, often associated with Constellation Software, leaves the most room for variation between products. Its strength is local judgment by the people who know each customer base; its weakness here is that no one in the group may hold a complete view of what records exist or what the contracts allow.

The data-moat thesis deserves the most scrutiny in diligence. A product can hold years of detailed customer records and still give its owner few rights to use them beyond running the service, so the moat may belong to the customers rather than the acquirer. Reading the customer data, aggregated data and AI clauses before signing is what turns the thesis into a fact.

The record distinction that decides what is possible#

The distinction that decides what is possible is between customer data stored in the product and the operating records the software company itself creates. A marina's slip reservations and work orders inside a marina management product are usually governed by that customer's agreement; the software company's own support tickets, implementation notes and engineering history are its own records.

Why internal operating records come first#

Internal operating records come first in most practical VMS data programs because they carry fewer contractual hurdles than customer data. The company created its support tickets, Jira issues and code reviews while running its business, and those records show how vertical problems actually get diagnosed and fixed.

Those records still need care. A support ticket can quote a customer's configuration or paste a screenshot of customer records, so preparation removes customer and personal details before anything leaves the business. But the starting position is usually clearer than for customer data, where the answer turns on terms written years earlier by previous owners.

What decentralization means for data decisions#

Decentralization means a holdco cannot simply order every business to contribute data. Business leaders own their results and customer relationships, and a program that feels imposed from the center will meet resistance and, often, legitimate customer concerns.

Groups that make progress tend to offer a program rather than a mandate. The center provides contract templates, a review process and a clear description of what qualifies; each business decides whether to take part and keeps approval over anything released.

• 1Publish a short description of which record types qualify and which are excluded.
• 2Offer shared legal review of customer terms and vendor terms.
• 3Agree one preparation standard for removing personal and confidential details.
• 4Let each business approve its own release under its normal financial arrangements.
• 5Share results across the group so other businesses can judge whether to join.

Five questions a holdco CEO should answer first#

A holdco CEO should answer five questions about the group's records before committing capital to any of these approaches or asking businesses to take part. None of them needs files; each can be answered from metadata and a read of current contracts.

Groups that cannot answer the first question usually find the rest stall too, which is why a simple group-wide inventory tends to come before any platform or licensing decision.

• 1Does anyone at group level hold a current list of products, systems and record families?
• 2Which products' customer agreements address aggregated data or AI use, and which are silent?
• 3Who in each business can approve a release of records, and what does the holdco need to see first?
• 4Which products face a help desk, CRM or hosting migration that could cut off older history?
• 5How would customers in each vertical react to AI features in the product compared with third-party licensing?

Illustrative: a small VMS group deciding where to start#

Illustrative: a fictional VMS holding company owns many small products, including software for self-storage operators, equipment rental yards and landscaping contractors. Each product runs independently with its own support and engineering teams.

The holdco CEO asks each business leader to answer a short metadata questionnaire. The equipment rental product has years of Intercom conversations linked to Linear issues and release notes, and its customer terms are silent on AI. The self-storage product holds rich customer data, but its terms limit use to providing the service. The landscaping product changed help desks twice and kept little history.

The group pilots outbound licensing of the equipment rental product's support and engineering history, with customer details removed. It plans updated aggregated-data and AI language for the self-storage product's terms at renewal, and leaves the landscaping product out for now.

Where SourceX fits in a VMS group#

SourceX works with the operating business that holds the records, not with the group as a single supplier. Each product that opts in runs its own SourceX five-step transaction, and the SourceX Enterprise Data Value Framework gives the holdco a consistent way to compare products on the same drivers, such as uniqueness, domain expertise, recency, rights and privacy burden, before choosing pilots.

Every released package carries a SourceX Evidence Packet, so the holdco has a dated record of what each business licensed and on what terms. SourceX's rights in a deidentified dataset are set out in the signed supplier agreement, and licensed records remain owned by the business that supplied them.