Deal economics
Custom data or repeatable product strategy? Start with same-company margins
By SourceX Editorial · Reviewed by Noah Loul ·
10 sections
Short answer
Treat custom data projects and repeatable products as two businesses. In 2024 filings, Gartner's syndicated research earned a 74 percent gross contribution margin against 36 percent for consulting, and Forrester's research segment reported operating income of about 66 percent of revenue against 50 percent for consulting. For an AI data company, the test is how much of each custom project it may license again, and how often it does.
Value drivers
Custom data repeatable
Key takeaways
- Within Gartner and Forrester, syndicated research out-earned custom consulting by about 38 and 15 percentage points of segment margin in 2024.
- Custom lines are run on utilization, headcount and backlog, so their revenue follows bookings rather than renewals.
- A deliverable ledger records who holds rights, who does repeat labor, who supplies tooling and who keeps recurring value for each piece of a project.
- Reuse share, second-use rate and product share show whether custom work is becoming product; the blended margin alone cannot.
- SourceX sources datasets on request under a repeatable agreement; its reuse and margin figures are not public and are shared with investors directly.
Custom projects or repeatable products: which should an AI data investor back?
Back the line whose output can be licensed again, and price custom work as the services business it is. Inside companies that run both models, repeatable information products earn markedly higher segment margins than custom engagements, and custom revenue tracks labor capacity rather than renewals. For an AI data company, the decisive question is how much of each custom project the seller keeps the right and the means to reuse.
Research and advisory firms are the closest public test. They sell information, they run a syndicated product and a custom service side by side under one brand, and they report the two lines separately. That controls for much of what usually muddies a cross-company comparison, such as brand, market and overhead. Two of them, Gartner and Forrester, file annual reports with segment results, and their 2024 figures are the evidence base here.
The thesis an investor can test is specific: an AI data company that calls custom projects a pipeline into products should be able to show, by deliverable, which components it kept, under which contract terms, and how often they were used again. Without that, a custom-heavy revenue line should be valued as services, whatever the pitch deck calls it.
What two research firms' segment reports show about the margin gap
In both firms' 2024 filings, the syndicated research line earned more per revenue dollar than the custom consulting line: 38 percentage points more at Gartner on its gross contribution measure, and about 15 points more at Forrester on segment operating income. The two measures are defined differently, so compare lines within each company, not across them.
Gartner's 2024 Form 10-K reports Research revenue of $5,125.7 million and gross contribution of $3,792.8 million, a 74 percent margin in both 2024 and 2023, against a Consulting margin of 36 percent in 2024 and 35 percent in 2023; it defines gross contribution margin as gross contribution as a percent of revenues. Its 2025 report renamed Research as Insights from the second quarter of 2025. Forrester's 2024 Form 10-K reports a Research segment with $316.7 million of research revenue and $21.1 million of consulting revenue, $115.7 million of segment expenses and $222.2 million of segment operating income, against a Consulting segment with $76.2 million of revenue, $37.8 million of expenses and $38.3 million of operating income; the margins in the table are this article's division of those figures.
The table puts the four lines on one page, with how each line earns its revenue:
| Line, fiscal 2024 | Revenue | Margin and measure | How the line is run |
|---|---|---|---|
| Gartner Research (syndicated) | $5,125.7 million | 74 percent gross contribution margin, as reported | Subscriptions, about 94 percent of segment revenue, recognized over the contract term |
| Gartner Consulting (custom) | $559 million | 36 percent gross contribution margin, as reported | Backlog, billable headcount and utilization |
| Forrester Research segment | $337.8 million, including $21.1 million of consulting | 65.8 percent of revenue, derived from segment operating income | Research revenue, with some consulting delivered inside the segment |
| Forrester Consulting segment | $76.2 million | 50.3 percent of revenue, derived from segment operating income | Consulting engagements; revenue tracks client bookings |
Custom revenue is counted in hours; repeatable revenue in renewals
A custom line is managed by how fully its people are booked, which is the clearest sign of where its costs sit. Gartner measures consulting utilization as total hours billed divided by total hours available to bill, and reports backlog as future revenue from engagements already in process. Its third-quarter 2024 report showed 960 billable consultants and 65 percent utilization, against 946 and 64 percent a year earlier. The research line runs on a different clock: subscription contracts, generally for twelve months or longer, make up about 94 percent of its revenue and are recognized ratably over the contract term.
Custom revenue also moves with bookings. Gartner's Consulting revenue grew 9 percent in 2024, faster than Research at 5 percent, while Forrester's consolidated consulting revenue fell 15 percent, which its filing attributes to lower client bookings. Either direction is consistent with a business whose output is sold once: revenue follows this year's demand for hours, not last year's installed base.
For an AI data company the translation is direct. A collection or labeling project to one buyer's specification is a consulting engagement in economic terms: its capacity metric is people or contributors booked, and its revenue resets when the project ends. A licensed dataset that several buyers pay for behaves like the research line. The investor's job is to find out which of the two each revenue dollar resembles.
The deliverable ledger: rights, labor, tooling and recurring value
Split each data project into the pieces it produces, and for each piece record four facts: who holds the rights after delivery, what labor repeats for the next buyer, who supplies the tooling, and who keeps any recurring value. A project with reusable pieces on the seller's side is a product in the making; one whose pieces all end with the buyer is a service, however well run.
The rights column decides most rows, and it is set by contract, not by the label on the invoice. An exclusive grant, an assignment or a work-made-for-hire clause can each move a piece from the seller's side to the buyer's, so the ledger should cite the clause behind every entry. The ledger below shows typical positions; each deal needs its own reading by counsel.
| Piece of a data project | Rights after delivery | Labor for the next buyer | Tooling supplied by | Recurring value kept by |
|---|---|---|---|---|
| Delivered records | Set by the license; exclusivity or assignment can end resale | None if relicensable; a full rebuild if not | Seller's pipeline | Whoever holds relicensing rights |
| Field definitions and schema | Usually the seller's unless assigned | Small edits per buyer | Seller | Seller, across projects |
| Quality checks and acceptance tests | Seller's if seller-written; buyer's if buyer-written | Rerun per delivery | Seller or buyer | Author of the tests |
| Rights and contract templates | Seller's templates; facts differ per deal | Shrinks as templates stabilize | Seller and counsel | Seller |
| Supplier and contributor relationships | No one owns them; contracts govern each | Outreach for each request | Seller's records | Seller, if suppliers return |
| Buyer's task specification | Buyer's confidential information | Rewritten for every buyer | Buyer | Buyer |
How much of a custom data project can be sold again?
Measure reuse directly instead of inferring it from the margin. Three ratios, each computed per project and per cohort of projects, show whether custom work is turning into product: the reuse share, meaning the share of a project's delivery cost spent on pieces the seller may use again; the second-use rate, meaning the share of those pieces that a later contract actually used within twelve months; and the product share, meaning the share of revenue from deliverables that existed before the contract was signed.
Mix explains why the ratios matter more than any single margin. As an illustration only, take Gartner's two 2024 margins as inputs, 74 percent for repeatable work and 36 percent for custom. A business with 80 percent of revenue from custom work would blend to 43.6 percent; at an even split, 55 percent; at 20 percent custom, 66.4 percent. Those figures describe arithmetic on two reported inputs, not a forecast for any company, and a data business's own line margins could sit far from Gartner's.
The ratios then tell an investor whether the mix can move. A company whose reuse share is high but whose second-use rate stays near zero is building assets nobody buys. One whose second-use rate climbs cohort by cohort is converting projects into product, and that conversion, not today's blended margin, is what a repeatable-product strategy has to demonstrate.
Is the margin gap a reuse premium or a reporting artifact?
Partly an artifact, so test it before relying on it. Segment measures leave out different costs at different companies: Gartner reports gross contribution, while Forrester's segment operating income is reconciled to a consolidated pre-tax income of $2.6 million for 2024 after amortization, restructuring and other items. Neither number is the profit of the line after its share of selling and overhead, and the gap could narrow or widen once those costs are allocated.
Labels also blur the lines. Forrester books $21.1 million of consulting revenue inside its Research segment, so even a clean segment split mixes some custom work into the product line. In AI data, the same blur appears when a company files bespoke collection under a platform or marketplace heading, or describes managed labeling as software. The test is the deliverable ledger, not the segment name.
Newer disclosure helps. FASB's Accounting Standards Update 2023-07 requires public companies to disclose significant segment expenses that are regularly provided to the chief operating decision maker and included in segment profit or loss, plus an other segment items amount with a description of what it contains. For a private AI data company, ask for the same layout: revenue, the main expense categories and the residual for each line, prepared the way management actually reviews them.
Where SourceX sits: sourced to order, signed on one repeatable agreement
On its published pages SourceX combines a custom trigger with a repeatable process. Its request-data page says datasets are sourced on request, not held in stock, so every licensed dataset starts as a response to a particular buyer's need, which is the custom side of the ledger. The contracting side is designed to repeat: a supplier signs one master licensing agreement after qualifying, and each later deal adds a short dataset-level confirmation instead of a new contract, while the investors page describes a standardized licensing agreement meant to shorten legal review on every deal.
The rights column matters most for reuse. Under the published terms, licenses are exclusive for AI training, meaning the supplier agrees not to license the selected data to anyone else for that purpose; the term and any alternatives sit in each deal's confirmation. The terms also state that the restriction binds the supplier and leaves SourceX's right to distribute or sublicense the de-identified dataset unaffected. In ledger terms, the de-identified dataset stays relicensable on SourceX's side, subject to each agreement. SourceX describes its own role as running sourcing and qualification, rights review, contracting, preparation, delivery and payment, so the labor that repeats per request sits with it as well.
Whether the delivered records are in fact licensed more than once, and what the reuse share, second-use rate and line margins are, is not public. SourceX shares company and commercial metrics with investors directly rather than in articles, and it is raising from accredited investors under Rule 506(c), with materials available on request through its investors page. The questions below are the ones to put to it, and to any competitor.
Five ways the repeatable-product case could fail
Five developments could weaken the case for steering toward repeatable products. Each has an observable signal, and an investor should ask which one a company considers its largest exposure.
- Custom demand outgrows the product: Gartner's Consulting revenue rose 9 percent in 2024 against 5 percent for Research, so a seller that declines custom work can give up the faster-growing line in some years.
- Custom revenue proves volatile: Forrester's consulting revenue fell 15 percent in 2024 on lower client bookings, and a data company that depends on project bookings can see the same swing when a large buyer pauses.
- Exclusivity and assignment cap reuse: where buyers insist on owning outputs, or suppliers grant exclusivity, the delivered records leave the seller's side of the ledger and only process assets remain to reuse.
- Analogies fail: research firms resell their own analysis, while data companies license records that third parties created, so rights limits, privacy review and source contracts can stop resale in ways a research library never faces.
- Margins converge once overhead is allocated: segment measures exclude different costs, and a full allocation of selling and administrative expense could shrink the gap these filings show.
Six diligence questions on custom versus product revenue
Ask for evidence by deliverable and by cohort, using definitions you set. SourceX should face the same list as any other AI data company:
- What share of last year's revenue came from deliverables that existed before the contract was signed, and how has that share moved by quarter?
- For projects completed a year ago, what share of their delivery cost went into pieces you may reuse, and what share of those pieces has a later contract used?
- Which contract clauses, such as exclusivity, assignment or work made for hire, moved pieces to the buyer, and on how many deals?
- What are the revenue, main expense categories and residual for each line, prepared the way management reviews them?
- Which capacity measures, such as utilization or contributor hours, run the custom side, and how do they move when a large buyer pauses?
- How many distinct buyers paid for the same dataset version, and what share of licensed scope was granted exclusively?
Evidence trail: filings read, figures derived, gaps left open
The comparison was assembled on 2026-10-27 from Gartner's 2024 and 2025 Forms 10-K, its fourth-quarter 2024 results release and its third-quarter 2024 Form 10-Q, Forrester's 2024 Form 10-K, FASB's Accounting Standards Update 2023-07, and SourceX's published request-data, how-it-works and investors pages. Gartner margins are as reported. Forrester margins are segment operating income divided by segment revenue, computed here: 222.183 over 337.834 for Research and 38.331 over 76.159 for Consulting, in millions of dollars, rounded once to one decimal. The Forrester segment figures tie exactly in thousands of dollars: revenue less segment expenses equals segment operating income for each segment. The filings were read through search excerpts on sec.gov, which blocks direct access from the research environment, so every figure should be checked against the filing before reuse.
The deliverable ledger, the three reuse ratios and the mix illustration are this article's frameworks. The mix figures are arithmetic on two reported margins, labeled illustrative, and are not projections for SourceX or any other company.
Not measured: line margins, reuse shares or second-use rates at any AI data company, SourceX included; how the margin gap changes after full cost allocation; and whether research-firm economics carry over to companies that license third-party records. Each needs company data that is not public.
Frequently asked questions
Why do repeatable data products usually earn higher margins than custom projects?
Because the cost of building a repeatable product is paid once and spread across every buyer, while a custom project's labor starts again for each client. Gartner's and Forrester's 2024 segment results show the pattern inside single companies. The size of the gap depends on how each company measures segment costs.
Can a custom data project become a repeatable product?
Yes, if the seller keeps the rights and the means to reuse what the project produced, such as schemas, quality checks, templates or, where the license allows, the records themselves. The evidence is a rising second-use rate across project cohorts. Contract terms like exclusivity or assignment can block reuse of the records.
What metrics separate a services business from a product business in AI data?
Services lines are run on utilization, billable headcount and backlog; product lines on renewals, buyers per dataset version and revenue from deliverables that existed before the contract. Ask for line-level revenue, main expense categories and residuals, plus reuse share and second-use rate by cohort.
Are research firms a fair comparison for AI data companies?
They are a useful analog, not a match. Research firms resell their own analysis, while data companies license records that third parties created, so rights limits, privacy review and source contracts can stop resale. Use the analog to frame questions and the company's own deliverable ledger to answer them.
Does SourceX publish its margins or reuse rates?
No. SourceX does not publish company or commercial metrics; it shares them with investors directly, and investor materials can be requested through its investors page. Its published terms describe datasets sourced on request, a repeatable master agreement and exclusive licensing for AI training that binds the supplier.
Sources
- Gartner's 2024 Form 10-K reports Research revenue of $5,125.7 million, gross contribution of $3,792.8 million and a 74 percent margin in 2024 and 2023, a Consulting margin of 36 percent (35 percent in 2023), Research subscriptions at about 94 percent of segment revenue recognized over the contract term, and defines utilization and backlog. Source
- Gartner's 2025 Form 10-K reports that the Research segment was renamed Insights in the second quarter of 2025. Source
- Gartner's fourth-quarter 2024 results release reports Consulting revenue of $559 million for 2024, up 9 percent. Source
- Gartner's third-quarter 2024 Form 10-Q reports 960 billable consultants and 65 percent utilization, against 946 and 64 percent a year earlier. Source
- Forrester's 2024 Form 10-K reports segment revenue, segment expenses and segment operating income for Research, Consulting and Events, consulting revenue down 15 percent on lower bookings, and a reconciliation to consolidated pre-tax income. Source
- FASB's ASU 2023-07 requires disclosure of significant segment expenses regularly provided to the chief operating decision maker and an other segment items amount with a description. Source
- SourceX's request-data page states that datasets are sourced on request, not held in stock. Source
- SourceX's how-it-works page describes one master agreement reused by returning suppliers with a per-deal confirmation, and exclusive licensing for AI training that binds the supplier and leaves SourceX's right to sublicense the de-identified dataset unaffected. Source
- SourceX's investors page describes a standardized licensing agreement and says company and commercial metrics are shared with investors directly. Source
Request SourceX investor materials
SourceX shares company and commercial metrics with investors directly. Any offering is made only to verified accredited investors.