Search guides, questions and resources

Exit readiness is how prepared a company is to run a sale process on good terms: clean financials, documented operations, clear ownership of assets and a team that can answer diligence quickly. A complete assessment also covers data assets: which records exist, who owns them, and whether any were licensed.

Exit readiness is the state in which a company can go to market on its owner's chosen timeline and come through buyer diligence without avoidable discounts: verified financials and KPIs, documented processes, a management team that runs without the seller, clean contracts and rights, and organized records that support a credible equity story.

Form 1042-S is the US information return a payer uses to report certain US-source income paid to foreign persons and any tax withheld, in place of a 1099 for US persons. Whether a referral partner receives one depends on their status, the payment and the payer's classification, so check the program terms and a tax adviser.

GDPval is a public AI evaluation, released in September 2025, that measures how well models produce real professional work products, such as reports, plans and spreadsheets, for occupations drawn from the largest sectors of the US economy. Experienced practitioners wrote the tasks and blind-graded the results, making authentic workplace deliverables the yardstick for AI progress.

Ground truth data is the verified, trusted outcome that a model's output is checked against. In business records, paid invoices, resolved tickets and won or lost deals act as ground truth because they record what actually happened, which is why AI buyers value records that carry outcomes.

Institutional knowledge is an organization's accumulated know-how about how it actually works: processes, exceptions, customer history and the reasons behind decisions. Tribal knowledge is the unwritten part held by long-serving people. Once written down in tickets, threads, SOPs and decision records, it becomes the record of real work that AI developers look to license.

IT due diligence is the review a buyer or investor runs on a target company's technology before closing: systems, infrastructure, security, software contracts, IT staff and costs. Its systems inventory, listing each application, its owner and how far back its history goes, is also an early signal of whether the company holds licensable operational records.

Micro private equity is control investing in very small private companies by small funds, independent sponsors or holding companies, usually below the deal size most lower-middle-market firms consider. It overlaps with search funds and holdcos but differs in who runs each company after closing, how capital is raised and how many companies the investor owns.

Model collapse is the gradual degradation of AI models trained, generation after generation, on data produced by earlier models rather than by people. Rare cases vanish first, then outputs grow narrower and less accurate. Researchers debate how severe it is in practice, but the risk adds value to verifiably human-made records, such as company archives.

Post-merger integration (PMI) is the work of combining two companies after a deal closes: aligning leadership, finance, people, customers, operations and IT so the combined business delivers the value the deal assumed. It runs in phases from pre-close planning to steady state, and it often retires systems whose record histories deserve an inventory first.

Pretraining data is the very large body of mostly public text, code and other content a language model learns from first, before fine-tuning for tasks. Private business records were never included because they are confidential and sit behind company systems, which is why buyers license them.

Process mining is a technique that rebuilds how a business process actually runs by analyzing timestamped event logs from systems such as ERP and CRM. For ERP consultants, the same event histories that reveal multi-step workflows are also a signal that a company holds licensable workflow records.

Proprietary data is information a company created or lawfully controls, keeps private, and has the right to use and license. Internal tickets, deal histories and engineering records usually qualify. Data a company merely holds for clients, or personal data it has no licensing basis for, usually does not, and that is the first thing a licensing review checks.

Quality of revenue analysis is a diligence review that tests whether a company's revenue is recurring, real and collectible, reconciling customer-level data to the ledger. A one-time data license payment is presented on its own line, outside recurring revenue, so buyers can trace it to a signed agreement.

Rights-cleared data is data whose licensor has confirmed it can license it for the stated use, with third-party content and personal information addressed before delivery. For company records, clearance checks authority, origin, contracts and privacy, then ends in a signed agreement and the company's authorization.

Section 365(n) of the Bankruptcy Code lets a licensee of intellectual property elect to keep its rights under the license for its term if a bankrupt licensor rejects the contract. Whether a dataset is covered depends on the statutory definition in section 101(35A) and the rights attached to it, so counsel should review.

Seller financing is when a business owner accepts a promissory note from the buyer for part of the price, paid over time with interest, instead of all cash at closing. The owner becomes a lender to the new operator, so the note carries buyer-performance and collection risk that a one-time cash or license payment does not.

Services-as-software is a business model in which a company sells a finished outcome, such as a closed month-end or a resolved support ticket, delivered mostly by AI agents with human review, instead of selling software seats for staff to use. Building those agents depends on detailed records of how real companies performed the same work.

Synthetic data is artificially generated data that imitates the statistical patterns of real data without being a direct copy of it. It is produced by rules, simulations or models, and it is commonly seeded or checked against real examples, which is why licensed real business records stay valuable to AI developers.

Tech-enabled services are businesses that sell an outcome delivered by people, with proprietary or deeply configured software making the work faster, more consistent or easier to measure. Investors place them between SaaS and labor-based services. Because they keep both platform logs and human work records, those with 50+ full-time employees at peak can be strong data licensing candidates.

The private equity holding period is the time a sponsor owns a portfolio company from acquisition to exit. Bain's 2026 report puts buyout holds at exit around seven years, up from five to six in 2010-2021. Longer holds make levers that add cash without a sale, such as a one-time data license, more valuable.

Under the AICPA Code, a commission is pay for recommending or referring a product or service that someone else supplies, while a referral fee is pay for recommending a CPA's services or a payment made to obtain a client. A reward for introducing a client to a data licensing service such as SourceX therefore reads as a commission.

The Lehman formula is a sliding scale for M&A success fees: 5% of the first $1 million of transaction value, 4% of the second million, 3% of the third, 2% of the fourth and 1% of everything above. Variants such as the double Lehman raise the rates, and the fee actually owed is whatever the engagement letter says.

The lower middle market is the segment of privately held companies between small Main Street businesses and the larger middle market, usually defined by revenue or EBITDA bands that differ from firm to firm. There is no official definition, so always ask whose range is being quoted. Many of these companies are established, owner-led US businesses.

The opportunity cost of an exclusive data license is the other AI-training deals a company cannot sign for the licensed records during the agreed term. It keeps ownership, internal use and any uses the agreement leaves open, so the real comparison is one certain exclusive payment against the buyers who have actually asked for the same records.

Bankruptcy Rule 6004(h) stays an order authorizing the use, sale or lease of estate property for 14 days after entry unless the court orders otherwise. Counsel ask to waive it when delay costs value. For a SourceX records license, the stay sets the earliest safe closing and delivery date.

The value gap is the difference between what a business is worth today and what it could be worth if it performed like its best-run peers; many planners also use it for the shortfall against the value an owner needs to exit. One-time data licensing proceeds can help an owner's wealth position but never change enterprise value.

Transaction advisory services (TAS) are the deal-support work accounting and advisory firms provide when a business is bought, sold, financed or separated: financial due diligence, quality of earnings, working capital and debt analysis, tax structuring, valuation and separation support. Because TAS partners meet owners preparing for a sale or financing, they are well placed to raise adjacent options.

Unstructured data is information that does not fit a predefined table or schema, such as email, chat messages, documents, presentations and call recordings. For businesses, its value rises when it can be tied to outcomes recorded in structured systems, which is the combination AI buyers find scarce.

The cost to prepare data for licensing is mostly internal time rather than cash: a sponsor's decisions, a data inventory of systems and years of history, a rights review of contracts and notices, and someone to run exports. SourceX's fee sits inside one all-in price with no separate charges, so the main outside cost is the company's own counsel.

Company data loses value when someone else owns it, archives were deleted, the records are mostly consumer personal data, the data is already licensed for AI training, or nobody can export it. Check rights, content, history and readiness before introducing a company to SourceX.

A good referral program has five things: written terms you can read before you join, a reward tied to real revenue rather than clicks or sign-ups, a clear cap and payment trigger, a way to track your referrals yourself, and a company that actually closes the deals you introduce. If any of these is missing or vague, treat the program with caution.

A proprietary data moat is data a company holds exclusively, has accumulated over years and uses in ways competitors cannot easily copy, so it protects margins or growth. PE buyers test the claim by checking exclusivity, rights, depth of history, links to outcomes and whether the data actually changes results. A time-limited AI training license can coexist with that advantage.

Manufacturing company valuation multiples in 2026 vary with what the plant makes, for whom and how much capital it needs. Buyers pay more for proprietary products, diversified customers, certified quality systems and maintained equipment, and less for single-customer build-to-print shops facing heavy capex. Office records such as quotes, NCRs and ERP job histories can be licensed separately from plant assets.

Ohio CPAs should test a referral reward against three layers before accepting it: the Accountancy Board of Ohio's statute and rules, the AICPA's commission and contingent fee rules for members, and SEC independence rules where the firm audits public companies. Because the reward depends on a deal closing, it raises commission and contingent fee questions.

Private equity sponsors expect a portfolio company CFO to deliver fast, reliable reporting, tight cash and covenant control, a credible first-100-days plan and tracking of every value creation initiative against the thesis. Increasingly that includes vetting new revenue ideas such as licensing operational records, covering contracts, rights, revenue treatment and lender consent.

Public AI data partnership programs have mostly asked large content owners for three things: deep archives, fresh or real-time access, and rights to use the material for training and in products, usually over multiyear terms. Announced deals with news publishers and Reddit show the pattern; deals for private operating-company records are rarely announced.

A seller should keep, or contract for access to, the records needed for taxes, disputes, indemnity claims and employee matters after closing. A retained copy is usually limited to those purposes and generally does not carry a right to license the records to others.

There is no standard consulting referral fee percentage. Firms that pass a client to each other negotiate a share of the first engagement's fees, a share of first-year billings, a flat thank-you or a no-cash reciprocal arrangement, usually paid only after the client signs and pays. Agree the base and trigger in writing and tell the client.

Searchers can keep passed deals warm, refer them to other buyers with the owner's consent, or introduce owners who won't sell to SourceX for a data license. Re-screen each for 50+ full-time employees at peak (contractors excluded), years of records across many systems and rights to license them, then ask permission. You never touch the data.

Keep a former employee's mailbox for a defined period, then archive or delete it by written policy, not by default. Shared mailboxes, archived mailboxes and exports all preserve history. Deletion is permanent and removes years of work records that a company might later want to license.

Before an LOI, share only what a buyer needs to price the business, released in rounds: teaser, CIM, a limited data room, then full confirmatory access. Hold back competitively sensitive terms and personal data. The same staging applies to data licensing, where partners share fit information only.

An MSP should disclose any vendor referral fee or agent commission to the client in writing before the client buys: who pays you, how the amount is worked out, and whether it changes the client's price. A data licensing introduction differs because the client receives the payment, and your reward comes from SourceX's fee.

Partners should never promise guaranteed anonymity, that no one will find out, that a deal will happen, a price, a reward amount, or that employees need not be told. Say instead that redaction rules are agreed before work begins, nothing is binding until the company signs, and data is delivered only after authorization.

When a freight broker is shutting down, freeze deletion first, then preserve exports from the TMS, email and shared drives before subscriptions lapse. Confirm who has authority and any court approval, check the 50+ full-time employees at peak baseline, and introduce the company to SourceX.

Staffing company valuation multiples in 2026 depend less on a published average than on the quality of gross profit: segment mix, client concentration, VMS exposure, workers' compensation history and recruiter depth. Buyers pay more for specialized, diversified, well-documented firms. Job-order and placement workflows may also support a separate data license, while candidate personal data stays out.

The private equity outlook for 2027 starts from three dated 2026 findings: about 32,000 unsold companies, buyout holds near seven years at exit, and returns that now depend on operating gains rather than multiples or leverage. Plan 2027 around exit readiness, levers that prove results within a budget year, and assets companies already own, including their records.

The applicable large employer (ALE) determination is the Affordable Care Act test of whether an employer averaged at least 50 full-time employees, including full-time equivalents, during the prior calendar year, counting commonly owned employers together. ALE status suggests a client may meet SourceX's baseline of 50+ full-time employees at peak (contractors excluded), but it is a proxy, not proof.