DataPacts

Broker track · Lesson 3 of 7 · 9:11

Qualifying a company's data

What you'll learn

  • The four tests of readiness
  • Can the data actually leave its systems?
  • What an inventory has to count
  • Talking about value without promising a price

Free lesson

Enter your email to watch this lesson and the rest of the course.

Transcript

A company can meet every entry rule a program publishes and still not be ready. A headcount floor is an entry rule and not an offer. And the check has a result most people forget: not proceeding is a perfectly valid outcome. Your job at this stage is to find that out early, before anyone spends money.

Chapter one. The four tests of readiness.

Readiness has four parts. Eligibility: does the company meet the program's entry rule? Data: does it hold records it can export? Permission: may it license each item? And organization: can someone decide, sign and keep records? A company can pass the first and still fail the other three.

So where do the entry rules sit? Across the programs that publish them, general headcount floors run from about twenty to fifty employees, and history from two to five years of records. Most prefer English-language data from the US, UK or Canada.

Be careful, though, because published thresholds can disagree. One program's home page states thirty or more employees. Its referral page states fifty or more. And its agreement says fifty office-based. Three programs do this, so read the page that governs your route.

Company status matters too. Several programs accept companies that are winding down or already closed. Some count headcount at the company's peak rather than today. So a shut-down company isn't automatically out. It has to match the page of the program you're considering.

Chapter two. What is there, and can it leave?

Every qualified company needs an inventory. For each source, it answers three questions: what it is, where it lives, and how much. It also records which contracts cover it and whether people appear in it. The inventory is internal and complete. What you send a buyer later is outward and selective.

Expect the question early. Most programs ask for an inventory in all but name within the first call, before any sample or contract. They ask about systems, date ranges, volumes and exclusions. Each of those is a field in the inventory row, so one document answers every intake request.

Sort records by what they are before where they sit, because buyers price data types, not systems. Six families cover most companies. Communications. Documents and wikis. Tickets and project histories. Code. CRM, sales and support records. And finance, operations and industry records.

Check the date range too. Retention rules can shorten it. A free workplace chat plan has deleted history older than one year since August 2024. So the history that exists can be shorter than the company's own history. Ask the administrator what is actually there.

Next, the question brokers skip. Can each system be exported on its current plan? Plan limits and vendor terms decide what leaves. A free chat plan exports public channels only. A cloud suite's export bucket deletes itself after sixty days. And after a cancelled subscription, data is gone within a hundred and eighty days.

Then there's the count that matters. A bankrupt US airline's court filing marks ninety-seven and a half million passenger profiles as not included. The same schedule counts a hundred million emails. Gross volume and licensable volume can be very far apart.

Code shows the same gap. A shutdown service states that roughly thirty-seven percent of raw code is removed in cleaning, as third-party or generated code and duplication. So every inventory row carries two numbers, gross and licensable, with the gap explained.

You can build all of this without moving a single record. Start from billing and sign-on records to list the systems, retired ones too. Name the administrator and the plan. Record the earliest and latest date held. Count in the units a buyer uses, and mark every estimate as an estimate.

Finish each row with a status: include, exclude or hold. A hold row stays out of every offer until its open question is answered. HR and payroll lines are a good example. They get a hold flag.

Chapter three. Permission, and the people who sign.

Permission readiness asks whether the company may license each item. A no here removes data from the sale. It doesn't just lower the price. Buyers ask the seller to confirm it owns the data and hasn't licensed it elsewhere on similar terms, and some ask for contractor IP assignments.

Researchers report that ninety-nine point nine eight percent of Americans would be correctly re-identified in a dataset using fifteen demographic attributes. That's why removing names isn't enough. And personal data in customer and staff records is the most common reason a company isn't ready yet.

Some data is excluded before anyone reviews it. One program's master terms exclude, unless the order form allows them, health information, biometric identifiers, payment card or financial account data, data of children under eighteen, and export-controlled information. Another program excludes law-firm client files and privileged communications.

Last, organization. A company is ready when one person can approve the scope and sign. And it can meet duties that keep running after delivery: consent and release records kept for seven years after the agreement ends, and security incidents notified within two business days.

Getting ready has risks of its own. Sharing samples before rights are clear. Treating a program's threshold as if it were an offer. Exposing personal data during preparation. So keep the inventory internal, and take counts from admin reports and system owners, not from exports.

Chapter four. Talking about value without promising a price.

Here's what you can say honestly. Company data is worth what an AI buyer will pay for the right to train on it. That's usually a license rather than a sale. And no public rate card exists.

Three methods are used. The cost approach values data at what it cost to create or would cost to replace. That gives you a floor for negotiation, nothing more. The market approach compares disclosed deals, and few are disclosed. The income approach looks at license income over the term.

Learn to separate claims from deals. A buyer program's payout range is that buyer's claim about its own offers. A disclosed deal value is a figure a party filed, signed, or a court recorded. And as of November 2026, no seller has publicly confirmed an amount received under the programs checked.

Buyer calculators deserve a warning label. A broker's own estimator says it plainly: a scenario, not a valuation. Another buyer's calculator returned four different ranges for the same inputs on four page loads. So never quote a calculator to a seller as a value.

Six drivers move value. Rights and chain of title. Customer consent. Personal data and de-identification. Exclusivity. Uniqueness. And volume, history and freshness. The first three can take value to zero.

You'll see unit prices quoted. Public prices exist for five units: title, hour, image, token and record. None is a market rate. They come from sellers with standard formats and clean rights. A company archive mixes formats, customer contracts and personal data, so it's priced as one package. Size times a unit rate misstates it.

And remember, not selling is a valid result. It's right when the license price is below the data's value to the company, or below the risk the sale creates. A perpetual, irrevocable grant can't be recalled after delivery. Several nos on the checklist can mean the honest answer is no.

Put it together as a checklist. Twelve questions cover the four tests. Every no marks work to finish before contacting a buyer, so a no is a task to complete.

So that's qualifying. Four tests: eligibility, data, permission and organization. An inventory that counts what's licensable. A value you describe by method and never as a promise. Next up is lesson B4, Referral programs and fees, and how to disclose them.

Read and watch the full lesson free. Enter your email to unlock every lesson of the academy.

Sources on DataPacts