From raw archive to licensable scope: the yield funnel for a portfolio

What is the yield funnel for licensable data scope?#

The yield funnel is a way to track how a company's raw archive narrows to the records it can actually license, stage by stage, with the reason for every reduction written down. It gives an operating partner one view across portfolio companies without anyone guessing ratios in advance.

The funnel has five stages. Total archive is everything the company still holds. In-scope systems are the sources worth exporting. Rights-cleared records are those the company may license. Privacy-prepared records have had personal and confidential details removed. Approved release is what the company's signer agrees to deliver.

Each move from one stage to the next passes a gate with a single question. Records that fail leave the funnel, and the reason becomes part of the documentation a buyer later reviews.

The five stages and the gate at each#

The five stages of the yield funnel each have one gate question, one decision owner and a familiar set of reasons records drop out. Portfolio teams keep this table beside the inventory so every reduction is logged against the right gate.

From total archive to in-scope systems#

The first gate, from total archive to in-scope systems, removes sources that cannot be exported with their context or that hold little decision-making signal. Volume is not the test at this gate; linkage and accessible history are.

Typical losses include help desks whose older tickets were purged under a retention schedule, ERPs migrated without closed transactions, and inboxes that only hold automated alerts. Platforms can also limit what leaves them: Zendesk says AI agent tickets cannot be exported, so a company that moved first-line support to an AI agent may hold a shorter human-handled history than it assumes.

Log each reduction at system level. A source parked because its export route is unclear today may return after a migration or a request to the vendor's support team.

What the rights gate removes#

The rights gate removes records the company holds but may not license, and those losses often surprise leadership because the records feel like the company's own work.

Customer contracts with broad confidentiality or data-use clauses, client-owned deliverables and code written under assignment clauses all fall out here. Records inherited through add-on acquisitions bring the acquired company's old contracts with them, and those terms were rarely drafted with licensing in mind.

Vendor terms matter as well. HubSpot's updated developer terms, for example, restrict using data accessed through its APIs to train or improve AI models, with a carve-out for legitimate single-customer use. Terms like these attach to the extraction route, so review how records leave a system as carefully as the records themselves.

Privacy preparation and the approval gate#

Privacy preparation and approval are the last two gates, and they narrow scope for different reasons: the first for what records contain, the second for what the company is willing to release.

At the privacy gate, a record stays if personal and confidential details can be removed without destroying the decision trail. Candidate files, payroll notes and free-text fields packed with names often fail. Legal standards differ; California's definition of deidentified information, for example, requires reasonable measures against reidentification, a public commitment not to reidentify and contractual obligations on recipients. Counsel assesses deal by deal which of these laws may apply, because the answer turns on the records and the states involved.

At the approval gate, the signer can hold back anything: a key account's escalations, records tied to a dispute, or a whole record family for brand reasons. Those holdbacks are legitimate and go in the log like every other reduction.

How to measure yield without inventing ratios#

Yield is measured by recording what is present at each stage in concrete units and comparing companies on why records left, not on a percentage set in advance. Benchmarks borrowed from other companies mislead, because the rights and privacy gates behave differently in every archive.

A portfolio view then lines companies up by stage. The useful question is not which company kept the most records, but which reached an approved release with the fewest open questions.

• 1Systems: the named source and export route present at each stage.
• 2Record families: tickets, work orders, NCRs, RFIs, code reviews or other families that pass or fail.
• 3Date ranges: the years still covered after each gate.
• 4Record counts: taken from the export itself, never estimated.
• 5Reason codes: one per reduction, such as purged history, client-owned, privacy density or signer holdback.
• 6Reversibility: whether a reduction could change, for example after a consent or a contract renewal.

Illustrative: one fund, three funnels#

Illustrative: a fictional growth buyout fund runs the funnel across three of its portfolio companies: a vertical software vendor on Jira, GitHub and Zendesk; a civil engineering firm on Deltek with RFIs and submittals in Procore and Bluebeam; and a staffing agency on Bullhorn.

The software vendor's funnel stays wide until the rights gate, where tickets from an acquired product are held back because an old reseller agreement restricts data use. The engineering firm loses most drawings and submittals at the rights gate as client deliverables but keeps internal review comments and RFI reasoning. The staffing agency's candidate notes fail the privacy gate.

The fund proceeds with two approved releases, the software vendor's linked tickets and code reviews and the engineering firm's internal review records, and parks the staffing agency with its reason codes recorded for a later look.

How SourceX maps the funnel to a transaction#

SourceX maps the yield funnel onto the SourceX five-step transaction: Supply covers the first two stages, Rights and Preparation cover the next two gates, and Approval produces the release that moves to Delivery. The initial fit check works from metadata only, so nothing is shared at that point.

The reason log travels into the SourceX Evidence Packet, so a buyer reading the provenance, licensing rights, permitted use, privacy record and release authorization can also see what was left out and why. The SourceX Enterprise Data Value Framework explains why a narrower scope can still be worth more: rights and data cleanliness increase value, while preparation cost and privacy burden reduce net value.