Skip to content

Deal economics

Data deal margin vs cash conversion: bridging booked profit to cash

By SourceX Editorial · Reviewed by Noah Loul ·

10 sections

Short answer

Data deal margin is booked when revenue is recognized; cash conversion measures how much of it the deal has actually collected, net of what it has paid out. Four timing lines separate them: uncollected revenue, buyer deposits, unpaid deal costs and advances to the data owner. In this article's illustrative model, one deal books 30 points under four payment structures while peak funding ranges from 25 to 70 points.

Value drivers

Data deal margin

This page covers data deal margin vs cash conversion: bridging booked profit to cash.Illustrative

Key takeaways

  • Deal margin measures what a deal earned; cash margin measures what it has collected minus what it has paid, on a stated date.
  • A five-line bridge reconciles the two from balances an auditor already uses: receivables and contract assets, contract liabilities, payables and prepaid amounts.
  • Under Topic 606, a gap of a year or less between delivery and payment need not change booked revenue, so booked margin carries no charge for waiting.
  • In the illustrative model, the same 30-point margin needs peak funding of 25 to 70 points and 80 to 348 funded point-months, depending only on payment terms.
  • What triggers the data owner's share, and on what base it is calculated, decides who funds the wait and who absorbs a buyer's credit.

What separates a data deal's booked margin from its cash margin?

Timing separates them. Booked deal margin is the revenue recognized on a deal minus the deal costs recognized against it; cash margin is what the buyer has paid minus what has been spent on the deal, measured on a stated date. The two end at the same number only when every invoice is collected and every cost is paid, and for most of a deal's life they can sit far apart.

Data licensing widens the gap because three parties move money on different triggers. The buyer pays on its invoice terms, sometimes with a deposit at signing. The company that supplies the records is paid on whatever trigger its agreement sets: at signing, at the buyer's acceptance or when the buyer pays. Preparation, rights review and delivery are paid as the work happens, which is usually first.

That is why the cash conversion of a data deal's margin deserves its own diligence line. This page builds a deal-level bridge from booked margin to cash margin, runs one illustrative deal through four payment structures, shows which accounting balances reveal the gap and explains which familiar metrics fail to measure it. Every model value is illustrative and describes no SourceX deal.

The five-line bridge from booked deal margin to cash margin

Start from booked margin and make five adjustments, each tied to a balance an accountant already tracks. If the bridge does not land on the deal's actual cash position, a line is missing or a payment was booked to the wrong deal.

The illustrative deal has a contract value of 100 points. The data owner's share is 45, preparation and de-identification cost 18, rights review and contracting 3 and delivery 4, so the booked deal margin is 30 points once the buyer accepts. Preparation is paid in months 1 and 2, rights work in month 1 and delivery in month 3. The buyer accepts in month 4, when this illustration recognizes the revenue, and pays 60 days later. Real recognition timing depends on each contract's terms and on the company's auditors.

The table fills the bridge twice: once in month 4, when the margin is booked, for a structure that pays the data owner only when the buyer pays; and once in month 2 for a structure with a 30-point buyer deposit, when the deal shows a loss to date yet holds cash.

The five-line bridge from booked deal margin to cash margin
Bridge lineFormulaBalance to requestOwner paid when buyer pays, month 430-point deposit, month 2
1. Booked deal margin to dateRevenue recognized minus deal costs recognizedDeal-level income statement30-21
2. Less revenue booked, not yet collectedMinus receivables and contract assetsReceivable, or contract asset while the right is conditional-1000
3. Plus cash collected before revenuePlus unearned depositsContract liability0+30
4. Plus deal costs booked, not yet paidPlus amounts owed to the data owner and vendorsAccrued liabilities and payables+450
5. Less cash paid before the cost is bookedMinus advances and prepaid costsPrepaid assets00
Cash margin to dateCash collected minus cash paid for the dealCash ledger by deal-25+9

Which accounting lines reveal the gap, and why does booked margin ignore waiting?

Three revenue-side balances carry most of the gap, and the revenue standard defines each one. FASB's ASU 2014-09, the update that created Topic 606, treats a receivable as an unconditional right to consideration, meaning only the passage of time stands before payment is due. A contract asset is a right to consideration for goods or services already transferred that depends on something other than time. A contract liability runs the other way: an obligation to deliver for which the customer has already paid, or the payment is already due.

Each balance maps to a bridge line. An accepted, invoiced dataset is a receivable. Revenue recognized while payment still depends on something beyond time, such as a further delivery the invoice is tied to, sits as a contract asset. A deposit taken at signing is a contract liability until the obligation is met. The standard presents each contract as a contract asset or a contract liability and shows unconditional rights separately as receivables, so a company can produce these balances deal by deal when an investor asks.

Booked margin carries no charge for the wait. As a practical expedient, the standard does not require an entity to adjust the promised consideration for a significant financing component when, at contract inception, it expects the period between transferring the good or service and the customer's payment to be one year or less. A buyer that pays 60 days after acceptance and one that pays 11 months after acceptance can produce the same booked revenue and the same margin; only the cash records tell them apart.

Public filings show how far contracted value can run ahead of revenue and billing in data licensing. Shutterstock's annual report for 2023 disclosed about $59.6 million of contracted but unsatisfied performance obligations, relating primarily to its data offering, that were not included in deferred revenue and were expected to be recognized over five years. Reddit's annual report for 2025 put remaining performance obligations on contracts longer than one year at $143.7 million, consisting primarily of long-term content licensing contracts. Contracted value, booked revenue, billed amounts and collected cash are four separate numbers.

Data deal margin and cash conversion under four payment structures

The payment structure, not the margin, decides how much cash a deal consumes. The same illustrative deal, with its costs and recognition month unchanged, books 30 points under every structure below, yet its peak funding ranges from 25 to 70 points and its financing burden more than quadruples between the best case and the worst.

Structure A pays the data owner at acceptance and collects from the buyer 60 days later. Structure B adds a 30-point buyer deposit at signing. Structure C pays the data owner in the month the buyer pays. Structure D pays the data owner a 20-point advance at signing and the remaining 25 at acceptance, while the buyer pays 90 days after acceptance. Every structure reaches 30 points of cash if the buyer pays in full: A, B and C in month 6 and D in month 7.

Funded point-months add up the negative month-end cash balances, so 70 points outstanding for two months counts as 140. Multiply them by a monthly cost of capital to price the wait: at an illustrative 1.5 percent a month, financing costs about 1.2 points under B, 1.6 under C, 3.0 under A and 5.2 under D, so a deal that books 30 everywhere keeps between 24.8 and 28.8 after funding itself. The figure plots each structure's cumulative cash by month; the four structures side by side:

Data deal margin and cash conversion under four payment structures
StructurePayment terms (illustrative)Cash margin when 30 is booked, month 4Peak fundingFunded point-months
A. Owner paid at acceptanceBuyer pays 60 days after acceptance; no deposit-7070198
B. Buyer deposit30 at signing, balance 60 days after acceptance; owner paid at acceptance-404080
C. Owner paid when buyer paysBuyer pays 60 days after acceptance; owner paid that month-2525108
D. Owner advance, slow buyerOwner receives 20 at signing and 25 at acceptance; buyer pays 90 days after acceptance-7070348
A line chart of cumulative deal cash in points by month for four payment structures; all four book a 30-point margin in month 4 while cash bottoms at minus 70, minus 40, minus 25 and minus 70 before reaching 30.
One deal, one booked margin, four cash paths. Illustrative model; no SourceX data.

Who funds the wait: the buyer, the data owner or the balance sheet?

Someone always funds the months between the first preparation hour and the buyer's payment; the structure only decides who. A deposit moves part of the wait to the buyer, paying the data owner when the buyer pays moves it to the owner, and an advance to the owner moves more of it onto the intermediary's balance sheet.

Moving the wait onto the data owner has a price. The Federal Reserve's 2025 Report on Employer Firms, based on the 2024 Small Business Credit Survey, found that 51 percent of employer firms cited uneven cash flows as a financial challenge and 56 percent cited paying operating expenses. A company licensing its records may value an earlier, fixed payment enough to choose a different offer, so structure C can lower peak funding while weakening the intermediary's supply.

The trigger also decides who absorbs a buyer's credit. Suppose the buyer takes a 15-point credit after acceptance and pays 85. If the data owner's share is a fixed 45, the deal keeps 15 points of margin. If the share is 45 percent of what the buyer actually pays, the owner receives about 38.3 and the deal keeps about 21.8. Neither choice is right in general; both belong in the signed agreement, and the bridge should show which one applies.

Growth multiplies the wait. A company that signs one structure-A deal a month always has five deals in their negative stretch, about 198 points of contract value in funding at steady state, while every one of those deals books a healthy margin. How that scales into company working capital is a separate question; at deal level, the lesson is that a profitable book can need more cash the faster it grows.

Rejected metrics: why DSO, gross margin and the cash conversion cycle miss the deal

Each familiar metric answers part of the question and fails a deal-level test. The bridge, peak funding and funded point-months survived because they are computed per deal, include both the buyer's and the data owner's timing and reconcile to cash.

Labels deserve the same scrutiny. In its interpretations on non-GAAP financial measures, SEC staff treats free cash flow as a liquidity measure that must not be presented on a per-share basis, and it judges a measure by its substance rather than by how management describes it. A private company's cash margin figure is not an SEC filing, but the discipline carries over: ask what it measures, how it is calculated and what it reconciles to. The candidates considered for this framework:

Rejected metrics: why DSO, gross margin and the cash conversion cycle miss the deal
Candidate metricWhat it capturesWhy it fails at deal levelVerdict
Deal margin or gross margin percentWhat a deal earns on recognized revenueIdentical across all four structures; blind to timingRejected as a cash measure
Days sales outstandingHow long buyers take to payIgnores deposits and the data owner's payment trigger; an average hides one slow buyerRejected alone; kept as an input
Cash conversion cycle (DIO + DSO - DPO)Company working capital daysA data deal holds no inventory, and company averages blend deals with different structuresRejected at deal level
Operating cash flow to EBITDACompany-wide cash conversionMixes deals, overhead and growth, so it cannot locate the deal that consumes cashRejected at deal level
Five-line bridge, peak funding, funded point-monthsPer-deal timing on both sides of the transactionNeeds deal-level cash records, which is the pointKept

The strongest objection: booked margin that never turns into cash

Timing is not the only leak; some booked margin never arrives. Credits issued after acceptance, rework that the deal model charged to overhead, disputes that stall an invoice and receivables that are never collected all shrink margin after it was booked. The bridge exposes them because a receivable that ages past its terms stays on line 2 until it is paid or written off.

Concentration turns one buyer's payment habits into the company's cash flow. Innodata's annual report for 2025 shows receivables from a single customer making up 63 percent of its accounts receivable at December 31, 2025. Reddit's annual report for 2025 says that "substantially all of the contract value associated with our licensing revenue is derived from two of our partners". Neither filing describes a data intermediary's deals, but both show how few counterparties can stand behind most of what a data business is owed.

A test for this objection: rebuild the bridge at each quarter-end for every open deal and compare the sum of line 2 with the receivables aging report. Receivables past their terms, credits issued after acceptance and write-offs belong in the same review as margin, because each one is margin that was booked and did not arrive.

What could make the cash-margin bridge misleading

Five limits apply even when the bridge reconciles perfectly. Each is a reason to read the bridge beside its assumptions, not a reason to skip it.

At a glance

What could make the cash-margin bridge misleading

  • Recognition policy: the bridge starts from booked margin, so a company that recognizes revenue at delivery rather than acceptance, or over a license term, shows different line 2 balances for the same cash. Ask for the policy memo before comparing companies.
  • Buyer budgets: data buyers spend inside an AI industry that has invested ahead of revenue. Sequoia's David Cahn estimated in September 2023 that each year of GPU spending at the then-current rate implied about $200 billion of lifetime revenue needed to pay back the investment, and asked how much of the buildout reflected true end-customer demand. A pullback would show up first as slower payments and smaller deposits.
  • Concentration: with one or two buyers behind most receivables, a single dispute can move the whole bridge, as the Innodata and Reddit filings illustrate.
  • Supplier terms: the structures that look best on peak funding, such as paying the owner when the buyer pays, work only if owners accept them, and the survey evidence on owners' own cash pressure suggests many will price that in.
  • Illustrative inputs: the costs, months and shares in the model were chosen to expose the mechanism. No public source reports deal-level payment timing for data licensing, so every input should be replaced with a company's own records.

Questions on deal margin and cash conversion, SourceX included

Ask for the bridge deal by deal rather than as a company-wide ratio, and apply these definitions instead of the company's own. The same questions apply to SourceX.

SourceX helps companies license selected business records to AI developers and runs the work from sourcing and qualification through rights review, contracting, preparation and delivery to payment, so every timing line in this bridge falls inside its own process. On the supply side the clock cannot start early: no records reach SourceX until the company supplying them has signed off on the scope and terms and executed the agreement.

None of SourceX's deal margins, payment terms or cash figures are public; its company and commercial metrics go to investors directly instead. Its raise is open to accredited investors under Rule 506(c), and a request for materials on the investor page is the place to ask something specific, such as which event triggers the data owner's share.

  • For each open deal at quarter-end, what are the five bridge lines, and does the total match the deal's actual cash?
  • What are peak funding and funded point-months per deal, shown as a distribution rather than an average?
  • What share of deals carried a buyer deposit, and how large was it relative to contract value?
  • What triggers the data owner's share (signing, acceptance or the buyer's payment), and is it fixed or calculated on what is collected?
  • How are receivables aged by buyer, and what share sits with the largest buyer?
  • Which credits, disputes and write-offs followed acceptance, and which deal's booked margin did each one reverse?

Methods note: the bridge, the four structures and what was not measured

The five-line bridge, the four payment structures and the rejected-metrics table were designed for this article as editorial frameworks on 2026-10-26. The accounting definitions come from FASB's ASU 2014-09. The public evidence comes from Shutterstock's 2023 annual report, Reddit's and Innodata's 2025 annual reports, the Federal Reserve's 2025 Report on Employer Firms, SEC staff interpretations on non-GAAP measures and Sequoia's September 2023 essay.

All model values are illustrative points on a contract value of 100: a data owner's share of 45, preparation of 18, rights work of 3 and delivery of 4; costs paid in months 1 to 3; acceptance and recognition in month 4; buyer payment 60 or 90 days later; and a 1.5 percent monthly cost of capital in the financing example. Funded point-months sum negative month-end balances from signing. Not measured: payment timing, deposits, credits or cash conversion at SourceX or any other company, taxes, and how any specific contract would be recognized.

Frequently asked questions

What is the difference between deal margin and cash margin?

Deal margin is the revenue recognized on a deal minus the costs recognized for it. Cash margin is the cash collected from the buyer minus the cash paid out for the deal on a given date. They converge only when everything has been collected and paid, and the gap between them is the cash the company is funding.

Can a profitable data deal still consume cash?

Yes. In this article's illustration, a deal that books a 30-point margin is 25 to 70 points cash negative in the month the margin is booked, depending on payment terms. Preparation and the data owner's share are often paid before the buyer pays, and Topic 606 does not require an adjustment to booked revenue for a payment gap of a year or less.

Is a buyer deposit revenue?

Not when it is received. Under Topic 606, cash received before the company has delivered is a contract liability, and it becomes revenue as the obligation is satisfied. A deposit improves cash margin early without changing booked margin, which is why the bridge adds it back on its own line.

Is paying the data owner when the buyer pays always better for the intermediary?

It lowers peak funding, to 25 points from 70 in the illustration, but it moves the wait onto the data owner, who may prefer an earlier fixed payment. It also decides who absorbs a buyer's credit when the owner's share is calculated on what is collected. The terms belong in the signed agreement, and counsel should review how they are drafted.

Does SourceX publish its deal margins or cash conversion?

No. SourceX has not published deal margins, payment terms or cash metrics, and it shares company and commercial metrics with investors directly. A reader evaluating the company can ask for investor materials at sourcex.si/investors, and only verified accredited investors can take part in any offering.

Sources

  • FASB's ASU 2014-09 (Topic 606) defines a receivable as an unconditional right to consideration, a contract asset as a right conditioned on something other than time and a contract liability as an obligation to deliver for which consideration was received or is due; it also allows entities not to adjust for a significant financing component when the expected period between transfer and payment is one year or less. Source
  • Reddit's annual report for 2025 says substantially all of the contract value associated with its licensing revenue is derived from two partners, and reports $143.7 million of remaining performance obligations on contracts longer than one year, primarily long-term content licensing contracts. Source
  • Shutterstock's annual report for 2023 disclosed about $59.6 million of contracted but unsatisfied performance obligations, relating primarily to its data offering, not included in deferred revenue and expected to be recognized over five years. Source
  • Innodata's annual report for 2025 shows receivables from one customer making up 63 percent of accounts receivable at December 31, 2025. Source
  • The Federal Reserve's 2025 Report on Employer Firms (2024 Small Business Credit Survey) found 51 percent of employer firms cited uneven cash flows and 56 percent cited paying operating expenses as financial challenges. Source
  • SEC staff interpretations on non-GAAP financial measures treat free cash flow as a liquidity measure that must not be presented per share and assess a measure by its substance rather than management's characterization. Source
  • Sequoia's David Cahn (September 2023) estimated that each year of GPU capital spending implied about $200 billion of lifetime revenue needed for payback and asked how much buildout reflected true end-customer demand. Source

Request SourceX investor materials

SourceX shares company and commercial metrics with investors directly. Any offering is made only to verified accredited investors.

See if your company qualifies