The Research Credit for a Technology Consulting Firm: The Four-Part Test, the Funded-Research Exclusion, and Who Actually Qualifies

The research credit is real, valuable, and narrower for a consulting firm than the marketing suggests, because of a rule that follows the money. The credit: a credit for increasing research activities, computed on qualified research expenses (wages for employees performing, supervising, or supporting qualified research; supplies used in it; a percentage of contract research paid to third parties; and cloud computing costs used in the research) above a base amount — with the regular method and the alternative simplified credit (a percentage of the excess of the year's qualified expenses over half the prior three years' average — the method most small firms use), the payroll-tax offset for qualified small businesses (a startup with gross receipts under US$5 million and no gross receipts more than five years back may apply the credit against its employer payroll taxes, up to US$500,000 a year — the provision that makes the credit usable for a firm with no income tax yet), and the state credits many states layer on. The four-part test — what qualifies as research: (1) permitted purpose — the activity is intended to develop a new or improved business component (a product, process, software, technique, formula, or invention) in its function, performance, reliability, or quality; (2) elimination of uncertainty — at the outset, the firm was uncertain about the capability or method of developing the component or its appropriate design; (3) process of experimentation — the firm evaluated alternatives through modeling, simulation, systematic trial and error, or testing; and (4) technological in nature — the experimentation relied on the principles of the physical or biological sciences, engineering, or computer science; software development that meets the four parts qualifies (with the additional rules for internal-use software — software developed primarily for the firm's own general and administrative functions faces a higher bar: the high-threshold-of-innovation test, requiring that the software be innovative, involve significant economic risk, and not be commercially available). The funded-research exclusion — the consulting firm's problem: research is excluded from the credit to the extent it is funded by a grant, contract, or another person — and research is "funded" when the firm is paid regardless of the research's success (the client pays for the hours whether or not the solution works — the consulting engagement's normal structure) or when the firm does not retain substantial rights in the results (the client owns the deliverable and the intellectual property — the consulting engagement's normal terms); a technology consultancy building a custom system for a client on a time-and-materials contract where the client owns the code fails both prongs — the work may be genuinely experimental and it is the client's research for credit purposes (the client, if it bears the risk and keeps the rights, may claim the credit on what it paid the consultancy as contract research, at the contract-research percentage), not the consultancy's; the exceptions run on contract terms — a fixed-fee engagement where the consultancy bears the risk of the work's failure (payment contingent on the solution working, or a fixed fee against uncertain effort) meets the risk prong, and a contract under which the consultancy retains rights to reuse the methods, code, or components it develops (a license back, or retained ownership of the underlying tools with the client receiving a license) meets the rights prong — so a consultancy's contract drafting determines its credit eligibility on client work, and most standard consulting agreements are drafted the wrong way for it. Where the consultancy's credit actually lives: its own products (a software product or platform the firm develops and sells or licenses — its own research, its own risk, its own rights); its internal tools and accelerators (the frameworks, code libraries, and platforms the firm builds to deliver engagements faster — the firm's own research, with the internal-use software rules applying where the tool is for internal administration rather than for delivering services to clients — a delivery accelerator used on client engagements is generally not "internal use" in the administrative sense, and the distinction is analyzed); its unfunded development (a proof of concept the firm builds on its own account before a client signs, or a solution the firm develops and then licenses); and the risk-bearing, rights-retaining client work its contracts are drafted to support. The 2025 change to the expense side: the 2017 law had required research and experimental expenditures to be capitalized and amortized (over five years domestic, fifteen foreign) beginning in 2022 — a rule that made research more expensive for every firm and that the 2025 legislation reversed for domestic research: domestic research and experimental expenditures are again deductible currently (alongside the consultancy's other deductions) for tax years beginning after 2024 (with a new provision governing the deduction, and with small businesses permitted to apply the change retroactively to their 2022–2024 years by amended return or accounting-method change); foreign research remains subject to the fifteen-year amortization; the deduction and the credit interact (the credit reduces the deductible expenses, or the firm elects the reduced credit to preserve the deduction) — and a consultancy that had been amortizing its research costs since 2022 re-runs those years. The documentation — what survives examination: the research credit is examined heavily, and the firm's file has to show, per project, the business component, the uncertainty at the outset, the alternatives evaluated and the experimentation process, the technological basis, the personnel and their time on the research (contemporaneous time records, or a defensible allocation), the supplies and cloud costs, and — for any client work claimed — the contract terms establishing the firm's risk and retained rights; a credit claimed on a study that reconstructs projects from memory and allocates wages by percentage guess is the one that is disallowed, and the credit's providers who take a percentage of the credit have an incentive the firm's own examination risk doesn't share. The claim: Form 6765 with the return (with the expanded reporting now required — the business components, the officers' wages, the acquisitions, and the detailed project information, with the business-component detail of Section G mandatory for tax years beginning in 2026), the payroll-tax offset election for qualified small businesses, the state credit forms, and the accounting-method or amended-return filings for the 2025 change's retroactivity. The honest assessment for a technology consultancy: the credit is worth pursuing for the firm's own products, tools, and unfunded development, and for client work only where the contracts were drafted to support it; a consultancy whose revenue is entirely time-and-materials client work under client-owns-everything terms has a small credit at best, and the pitch that its billable projects qualify is the one to decline.

Key takeaways

• The four-part test: a new or improved business component; technical uncertainty at the outset; a process of experimentation; a basis in the hard sciences, engineering, or computer science — software qualifies, with the higher internal-use bar for administrative software.
• The funded-research exclusion is the consultancy's problem: research paid for regardless of success, or whose results the client owns, is the client's research — and standard time-and-materials, client-owns-the-code contracts fail both prongs.
• Contract terms decide eligibility on client work: a fixed fee with the firm at risk, or retained rights to the methods and code, can bring client work in — most standard agreements are drafted the wrong way.
• The consultancy's credit lives in its own products, delivery tools and accelerators, and unfunded proofs of concept — its own risk, its own rights.
• The 2025 legislation restored current deductibility of domestic research expenses for tax years after 2024, with retroactive relief for small businesses' 2022–2024 amortization); foreign research stays on fifteen-year amortization.
• Documentation per project — component, uncertainty, alternatives, experimentation, technology, contemporaneous time, and contract terms for client work — is what survives the examination the credit reliably draws; Form 6765's expanded reporting is the claim.

The consultancy's research credit screen

Per project: business component; uncertainty at the outset; experimentation process; technological basis. Funding: who bears the risk (payment contingent or fixed?); who owns the results (client, or rights retained)? Own products and tools: yes; client T&M with client ownership: no; fixed-fee or rights-retained client work: analyzed. Expenses: wages by time record; supplies; cloud; contract research. 2025 change: prior-year amortization re-run. Form 6765 with the expanded reporting; the payroll-tax offset if a qualified small business; state credits. The funding line is the one the pitches skip.

Worked example

A twenty-person data engineering consultancy is pitched a research credit on its entire US$4 million of client project revenue. The screen: US$3.4 million of the work is time-and-materials under agreements where the client owns all deliverables and code — funded research on both prongs, no credit (the clients may claim contract research on what they paid, if they qualify); US$400,000 is a fixed-fee engagement where the firm bore the risk of an uncertain integration and the agreement grants the client a license while the firm retains ownership of the integration framework it built — the firm's risk, the firm's rights, and the framework meets the four-part test: eligible; and the firm spent US$350,000 of engineering wages on its own delivery accelerator (a reusable pipeline platform used across engagements) with genuine technical uncertainty and documented experimentation — its own research, eligible (the platform delivers client services rather than administering the firm, so the internal-use bar doesn't apply). The claim: the alternative simplified credit on the eligible wages and cloud costs, Form 6765 with the project-level reporting, the state's credit — a credit in the mid five figures, not the six figures the pitch projected on US$4 million. The 2025 change: the firm had amortized its accelerator research since 2022 — the retroactive relief for small businesses is applied to 2022–2024, recovering the deferred deductions. Going forward: the firm redrafts its standard agreement to retain rights in reusable components and offers fixed-fee structures on engagements with genuine technical uncertainty — the contract terms that make future client work eligible. The competitor that claimed the credit on all of its T&M client revenue with a study that allocated 60% of every engineer's time to "research" from memory: the examination disallowed the client-work portion under the funded-research exclusion and the rest for want of contemporaneous records, and the study provider's fee had been a percentage of the disallowed credit.

Official sources

The IRS's audit techniques guide for the research credit states that "to be considered 'qualified research', the taxpayer must be able to establish that the research activity being performed meets ALL four of the above tests," and that "the exclusion for 'funded research' under section 41(d)(4)(H) provides that the credit shall not be available for qualified research to the extent funded by a contract, grant, or otherwise by another person." — Internal Revenue Service, Audit techniques guide: Credit for Increasing Research Activities (IRC § 41), https://www.irs.gov/businesses/audit-techniques-guide-credit-for-increasing-research-activities-ie-research-tax-credit-irc-ss-41-qualified-research-activities

Publication 538 states that "under the cash method, you generally report income in the tax year you receive it, and deduct expenses in the tax year in which you pay the expenses," and that "a corporation or partnership, other than a tax shelter, that meets the gross receipts test can generally use the cash method." — Internal Revenue Service, Publication 538, Accounting Periods and Methods, https://www.irs.gov/publications/p538

Practitioner note

The research credit is pitched to every technology consultancy on its billable revenue and lives, for nearly all of them, somewhere else — in the firm's own products, its delivery tools, and the client work whose contracts put the risk and the rights with the firm. Our screen runs the funded-research prongs before the four-part test, because a client who pays for the hours and owns the code has bought the research; and we redraft the standard agreement so the next engagement can qualify — which is the advice the percentage-fee study providers don't give.

See also: For related guidance, see the consulting firm's entity structure and the QBI cap; and browse every small business tax guide, by situation.

Next step

Fairlight Accounting is a cross-border accounting and tax practice with a U.S. Tax Desk and a Canadian Tax Desk. Our U.S. Tax Desk handles research credit analysis for technology consultancies — the funded-research screen on client work, four-part-test documentation for own products and delivery tools, contract redrafting for future eligibility, the 2025 expense-deduction change and retroactive relief, and Form 6765 with the payroll-tax offset and state credits. See pricing or book a call.