> [!FOUNDER]
> "Founders treat the research credit as a windfall they discover at filing time. It is not a windfall; it is a second return on research spending that the company already did, and it is priced by a formula the company can run before the year ends. The expensive mistake is not failing to claim it. The expensive mistake is electing the wrong method in a hurry and then living with that election for years." — Enow A. Jovial, Founder & Chief Executive Officer
What IRC § 41 actually pays for
The research credit under IRC § 41 is a general business credit, claimed on Form 6765 and carried to Form 3800. It pays for increases in qualified research, not for research spending in general: both computation methods measure current-year qualified research expenses (QREs) against a historical baseline and credit only the excess. That design choice is the whole strategy question. A company whose research spending is flat year after year can spend millions on genuine research and earn a small credit, because the credit rewards growth over history, not effort in the abstract.
Qualified research itself must pass the four-part test in IRC § 41(d): a permitted purpose (a new or improved product, process, or software), technological in nature, intended to eliminate uncertainty, and involving a process of experimentation. Routine data collection, foreign research, research funded by another party, and research after commercial production begins are excluded by statute and regulation. The test is applied per business component — a product, process, or software project — which is why Form 6765 Section G now asks for component-level detail. A credit study that cannot name its components, the uncertainty each one faced, and the experimentation used to resolve it is a narrative, not a claim.
> [!KEY TAKEAWAY]
> The credit rewards growth over history, not spending. Before electing a method, compute both: the regular method wins when the fixed-base percentage is low relative to current spending, and the ASC wins when recent spending has grown fast. The difference in the worked example below is $58,000 on the same $1,000,000 of QREs — decided by arithmetic, not by preference.
The two formulas, stated exactly
The regular method, under IRC § 41(a)(1), equals 20 percent of current-year QREs above a base amount. The base amount is the fixed-base percentage multiplied by average annual gross receipts for the four preceding years, and it can never be less than 50 percent of current-year QREs. The fixed-base percentage itself is capped at 16 percent under IRC § 41(c)(3)(C), and for start-up companies it follows a statutory ramp beginning at 3 percent for the first five years with QREs.
$
ext{Regular Credit} = 20% imes max( ext{Current QREs} - ext{Base Amount}, 0)
$
$
ext{Base Amount} = max( ext{Fixed-Base Percentage} imes ext{Average Gross Receipts (prior 4 years)}, 50% imes ext{Current QREs})
$
The alternative simplified credit (ASC), under IRC § 41(c)(5), equals 14 percent of current-year QREs above 50 percent of the average QREs for the three preceding years. If the taxpayer had no QREs in any one of those three years, the ASC equals 6 percent of current-year QREs outright.
$
ext{ASC} = 14% imes max( ext{Current QREs} - 50% imes ext{Average QREs (prior 3 years)}, 0)
$
$
ext{ASC (no prior-year QREs)} = 6% imes ext{Current QREs}
$
QREs themselves have three buckets under IRC § 41(b): in-house research expenses (wages for qualified services, supplies consumed in research, and computer rental or lease costs), contract research expenses counted at 65 percent of amounts paid to third parties, and basic research payments to qualified organisations. The 65 percent haircut on contract research is a deliberate policy choice: a dollar paid to a contractor counts less than a dollar paid to an employee, because the statute assumes the contractor keeps part of the value. Companies that outsource most research therefore face a structurally smaller credit base than companies that employ their researchers — a fact that should shape hiring plans, not just tax filings.
Worked example: the same $1,000,000 of QREs, two answers
Take a software company with current-year QREs of $1,000,000, prior-three-year QREs averaging $600,000, a fixed-base percentage of 8 percent, and average gross receipts for the prior four years of $10,000,000. All figures are illustrative and chosen so the arithmetic can be checked by hand.
Under the ASC:
$
ext{ASC} = 14% imes (1{,}000{,}000 - 50% imes 600{,}000) = 14% imes 700{,}000 = $98{,}000
$
Under the regular method, the base amount is 8 percent of $10,000,000, which is $800,000 — above the 50 percent floor of $500,000, so $800,000 stands:
$
ext{Regular Credit} = 20% imes (1{,}000{,}000 - 800{,}000) = 20% imes 200{,}000 = $40{,}000
$
The ASC wins by $58,000 on identical facts. Reverse the history and the answer flips: if the fixed-base percentage were 3 percent (the start-up ramp), the base amount would be $300,000, floored up to $500,000, and the regular credit would be 20 percent of $500,000, or $100,000 — beating the ASC by $2,000. That crossover is the decision the election actually makes, and it is why a company should never elect a method on the advice that one method is generally better. The better method is a function of four numbers the company already has: current QREs, prior QREs, gross receipts history, and fixed-base percentage. Run both every year, even in a year when the election cannot change, because the comparison is also the forecast for the year when it can.
The decision table
| Factor | Regular method (Section A) | ASC (Section B) | Authority |
| :--- | :--- | :--- | :--- |
| Rate on the excess | 20 percent | 14 percent | IRC § 41(a)(1), § 41(c)(5) |
| Baseline | Fixed-base percentage × prior-4-year average gross receipts, floored at 50 percent of current QREs | 50 percent of prior-3-year average QREs | IRC § 41(c)(1)–(3), § 41(c)(5) |
| Records required | Gross receipts and QREs back to the base period, including the 1984–1988 base years for older companies | Three years of QREs | Form 6765 Sections A and B |
| Best when | Fixed-base percentage is low or gross receipts are small relative to research | Research spending has grown recently | Arithmetic, not preference |
| Election durability | Revokes an ASC election when filed on a timely original return | Carries forward until revoked the same way | Form 6765 instructions, Items A and B |
| No prior-year QREs | Base amount rules still apply | 6 percent of current QREs, no baseline | IRC § 41(c)(5) |
The § 280C election: credit size versus deduction size
Every claimant faces a second election on Item A of Form 6765: take the full credit and reduce the research deduction by the credit amount, or elect the reduced credit and keep the full deduction. With the election, the regular credit is figured at 15.8 percent instead of 20 percent, and the ASC amount is multiplied by 79 percent — both reductions equal to the 21 percent maximum corporate rate. Without it, IRC § 280C(c)(1) reduces the domestic research deduction by the credit amount.
Since the One Big Beautiful Bill Act added IRC § 174A, restoring immediate expensing for domestic research for tax years beginning after December 31, 2024, this coordination has real cash consequences: the deduction being reduced is now an immediate deduction rather than a five-year amortisation, as explained in the Section 174 software capitalization guide. A company in a high marginal bracket may prefer the reduced credit and the full deduction; a company with no current tax liability may prefer the full credit and the payroll election below. There is no default answer, and the election is irrevocable for the year once made on a timely filed original return — another reason the comparison belongs in the planning meeting, not in the filing week.
The payroll offset: the credit for companies with no tax bill
A qualified small business can elect on Form 6765 Section D to apply up to $500,000 of its research credit against employer payroll taxes instead of income tax, claimed quarterly through Form 8974. Qualification is narrow by design: generally a corporation or partnership with under $5 million of gross receipts for the year and no gross receipts before the five-year period ending with that year. The election must be made on a timely filed original return, and the credit can be used for up to five years of payroll offset.
This is the provision that makes the credit real for pre-revenue startups, and it is also the provision most often forfeited by timing: a startup that files late, or files an original return without the election and tries to add it on an amended return, loses the payroll use for that year entirely. Founders who fund research from venture money or savings should treat the Form 6765 payroll election as a calendar item with the same seriousness as a financing closing date, because unlike a financing, it cannot be renegotiated after the deadline.
> [!WARNING]
> An unsupported credit is worse than no credit. The failure mode is examination adjustment of the QRE base — wages for staff who did no qualified work, contract research counted at 100 percent instead of 65 percent, or components with no documented uncertainty — followed by accuracy-related penalty exposure under IRC § 6662 on the underpayment. Contemporaneous documentation is the defence: time records tied to components, project narratives written during the work, and a QRE ledger reconciled to payroll and vendor invoices before the return is filed. A credit reconstructed from memory a year later rarely survives contact with an examiner.
Entity form, payroll mechanics, and the book-tax gap
Entity form changes the filing mechanics, not the credit math. A C-Corp claims the credit on its own return through Form 3800, while an S-Corp or an LLC taxed as a partnership passes the credit through to its owners, who apply it subject to their own limitations. The payroll election runs through the employer's EIN on Form 8974, where the credit offsets the employer share of FICA taxes quarter by quarter. Have a CPA reconcile the QRE ledger to payroll records before filing, because the wage base is where examiners start. Note also the book-tax gap: under GAAP, FASB ASC 730 requires research and development costs to be expensed as incurred in the financial statements, while § 41 QREs are a tax-only measurement with their own inclusions, exclusions, and the 65 percent contract research haircut. A company whose financial statements show large research expense does not automatically hold a large QRE base, and the reconciliation between the two is part of the credit file.
Section G: the reporting change that arrived with the 2026 tax year
Form 6765 (Rev. December 2024) carries seven sections, A through G. Section G breaks QREs out by business component, with descriptions of the technical uncertainty and the experimentation for each component. For tax years beginning before 2026, Section G is optional for all filers; for tax years beginning after 2025, it is required unless an exception applies. The practical consequence is a documentation standard that must be met during the year, not after it: a company whose 2026 research has no component-level records by December cannot manufacture them credibly in April. The companies that will comply cheapest are the ones that already track research by project in their engineering and accounting systems — the same systems that feed the cash vs accrual accounting guide and the Section 83(b) election playbook, where equity compensation for research staff also enters the QRE wage base when the work qualifies.
The election playbook, in order
1. Identify components before counting dollars. List every product, process, and software project, and test each against the four-part test. Components that fail are excluded now, not argued later.
2. Build the QRE ledger from source systems. Wages from payroll records for qualified services, supplies from purchase records, and contract research at 65 percent from vendor invoices. Reconcile the total to the general ledger before any credit math.
3. Compute both methods on the same QRE base. Regular and ASC from identical inputs, with the fixed-base percentage and gross receipts history documented. The method comparison is only honest when the inputs are shared.
4. Model the § 280C choice in cash terms. Full credit with reduced deduction versus reduced credit with full deduction, at the company's actual marginal rate, including the § 174A interaction for domestic research.
5. Test the payroll election against the calendar. If the company qualifies as a small business, confirm the election can still be made on a timely filed original return, and diary the Form 8974 quarterly claims that follow.
6. Prepare Section G during the year. Component narratives, uncertainty descriptions, and experimentation records written while the work is happening, not reconstructed at filing.
7. File, then keep the file. The credit study, the QRE ledger, the method comparison, and the election record stay together for the examination period. A credit without its file is a liability with a due date.
Execution checklist
> [!TIP]
> Run the dual computation in October, not in March. The ASC election for the current year is made on the original return, but the planning value of knowing which method wins — and whether the payroll election is available — belongs to the year being planned, when hiring, contractor, and project decisions can still change the QRE base itself.
Related reading on this site: the Section 174 software capitalization guide, for the deduction that must be coordinated with this credit; the foreign developer payroll § 174 guide, for the foreign-research schedule that interacts with the same payroll; and the Section 83(b) election playbook, because equity compensation timing and research timing hit the same return. To price the credit in cash terms, run the forecast through the tax estimator and the hiring plan through the break-even calculator.