Inside an R&D Credit Study: A Worked Example for a SaaS Company

Most explanations of the R&D tax credit stop at the qualification rules. You learn what a business component is, you learn the four-part test, and then the article ends right before the part you actually wanted: the arithmetic. This post does the arithmetic.

Read this first. Everything below describes SaaSCo, a hypothetical, illustrative company that does not exist. It is a composite built to demonstrate the mechanics of IRC Section 41, not a real client, and the figures are not a real engagement, a real outcome, or a representative result. Ricerca is not claiming these results for anyone. Your numbers will differ, possibly by a lot, because they depend entirely on your own facts. Nothing here is tax advice.

With that said, here is what a study actually does.

The hypothetical: SaaSCo

SaaSCo is a fictional B2B analytics company. All of these facts are invented for the example:

25 employees, 15 of them technical, all working in the United States.
$4.2 million in gross receipts for the credit year. It is venture-funded and spending ahead of revenue.
Incorporated four years before the credit year, with its first dollar of gross receipts in the year it was incorporated, which falls inside the five-taxable-year period ending with the credit year, and none before that. This detail matters later.
No income tax liability in the credit year, because it is operating at a loss.

That last fact is the one that makes founders assume the credit is worthless to them. It is not, and the reason is at the end of this post.

Step 1: Identify the business components

The credit is not computed on your company. It is computed on business components, meaning each product, process, computer software, technique, formula, or invention you are developing or improving. A study starts by breaking the engineering year into components.

SaaSCo’s study identified four:

Real-time anomaly detection service. A streaming model that flags outliers in customer telemetry.
Multi-tenant data isolation re-architecture. Reworking the storage layer so tenant data is provably segregated without collapsing query performance.
Columnar query engine migration. Moving the analytics workload off row storage to hit interactive latency targets.
Connector framework for third-party APIs. A generalized ingestion layer for partner systems with inconsistent and poorly documented behavior.

Component-level structure is not a formality: it is the unit the statute uses, the workpapers follow, and the redesigned Form 6765’s Section G is built around.

Step 2: Apply the four-part test

Each component gets tested against §41(d): a permitted purpose, technological in nature, elimination of uncertainty, and a process of experimentation that covers substantially all of the activity. We walk through each part at length in the four-part test in plain English.

All four of SaaSCo’s components cleared it, but not identically. The columnar migration was the cleanest case: the team benchmarked three storage engines against a production-shaped workload and two failed the latency target. The connector framework was the closest call, and the study had to separate genuine protocol reverse-engineering from routine integration that just followed published documentation.

Step 3: Remove what does not qualify

This is where studies earn their keep, and where an aggressive provider quietly creates your audit exposure. The §41(d)(4) exclusions removed several things from SaaSCo’s year:

Post-release maintenance and bug fixes on components already in commercial use (§41(d)(4)(A)).
A customer-specific integration built under a time-and-materials contract where the customer owned the resulting IP and paid regardless of whether it worked. That is funded research under §41(d)(4)(H) and it is out.
Work by a contractor in Poland. §41(d)(4)(F) excludes research conducted outside the United States. It is a location test, not a nationality test.
Marketing site work, sales engineering, and routine cloud cost tuning, none of which involved technical uncertainty.

Removing this work is what drives the qualified-time percentages below 100%. A study that assigns your engineers 100% qualified time is not being generous. It is being wrong.

Step 4: Build the QREs

Qualified research expenses fall into a short list of statutory categories. For SaaSCo, wages dominate, which is typical for software. The study assigned each role group a qualified-time percentage supported by interviews and project records, then applied it to W-2 wages.

Role group (headcount)	W-2 wages	Qualified %	Wage QRE
Backend and platform engineers (7)	$1,260,000	75%	$945,000
Frontend engineers (3)	$480,000	60%	$288,000
ML engineers (2)	$400,000	90%	$360,000
Site reliability engineer (1)	$170,000	30%	$51,000
QA engineer (1)	$120,000	50%	$60,000
VP of Engineering (1)	$240,000	40%	$96,000
Total (15)	$2,670,000		$1,800,000

Illustrative only - every wage, percentage, and total is invented for this example.

The VP of Engineering is in the table because §41(b)(2)(B) counts wages for direct supervision of research, not just wages for performing it. The 40% reflects the share of that role spent supervising qualified projects rather than running budgets and hiring.

Then the non-wage categories:

QRE category	Basis	Amount
Wages for qualified services	From the table above	$1,800,000
Computer and cloud rental	$170,000 of a $600,000 annual cloud bill traced to non-production research environments	$170,000
Contract research	$200,000 paid to a US development firm, at 65%	$130,000
Supplies	None identified	$0
Total current-year QREs		$2,100,000

Illustrative only - qualification and every amount are fact-specific.

Two notes. Cloud spend counts under §41(b)(2)(A)(iii), which covers amounts paid for the right to use computers in the conduct of qualified research. Only the development, staging, and model-training environments qualified; the production infrastructure serving paying customers did not, and the statute also carves the clause back to the extent you receive payments from others for the right to use substantially identical property. And contract research is taken at 65% of the amount paid under §41(b)(3)(A), so $200,000 x 65% = $130,000. A 75% rate applies to a qualified research consortium and 100% to certain energy research, neither applicable here. Our qualified research expenses page covers each category.

Step 5: Compute the credit, both ways

There are two ways to compute the federal credit, compared in detail on our calculation methods page: the Regular Credit and the Alternative Simplified Credit under §41(c)(4). A persistent myth says a company founded in the 2020s cannot use the Regular method because its fixed-base percentage looks back to 1984-1988 data. Not so: §41(c)(3)(B) gives a start-up a statutory 3% fixed-base percentage for its first five credit years, so the Regular Credit is always on the table. A study computes both and files the stronger supportable result. So does this example.

The ASC first. The formula is 14% of the amount by which current-year QREs exceed 50% of the average QREs for the three preceding taxable years. SaaSCo’s prior three years, again hypothetical:

Year	QREs
Three years prior	$600,000
Two years prior	$900,000
One year prior	$1,200,000
Three-year total	$2,700,000

Illustrative only - every figure is invented for this example.

The arithmetic, step by step:

Average prior-3-year QREs: $2,700,000 / 3 = $900,000
Base amount, 50% of that average: $900,000 x 50% = $450,000
Excess of current QREs over the base: $2,100,000 - $450,000 = $1,650,000
Credit: $1,650,000 x 14% = $231,000

Now the Regular Credit on the same facts. The Regular method is 20% of QREs above a base amount tied to historical gross receipts, and the base can never be less than 50% of current-year QREs. With SaaSCo’s statutory 3% fixed-base percentage applied to its small early-stage receipts, that floor is what governs: base = $2,100,000 x 50% = $1,050,000, and the credit is 20% x ($2,100,000 - $1,050,000) = $210,000.

Here the ASC’s $231,000 beats the Regular Credit’s $210,000, because SaaSCo’s research spend grew quickly off a small base. Flip the growth curve and the Regular method often wins, especially for start-ups holding the statutory 3% rate. That is why a study computes both rather than assuming either.

If you have no prior-year QREs

§41(c)(4)(B) handles the first-time claimant. If a taxpayer has no qualified research expenses in any one of the three preceding taxable years, the ASC is 6% of current-year QREs instead. On SaaSCo’s $2,100,000, that would be $2,100,000 x 6% = $126,000. Lower, but available immediately, and it is why a first study is worth running rather than waiting for a base period to build up.

Step 6: The §280C(c) reduced credit election

Under Section 280C(c), you cannot both deduct research expenses in full and claim a credit on the same dollars without an adjustment. For tax years beginning after 2024 the mechanism runs through §174A: you either reduce your domestic research deduction (or the amounts you capitalized) by the full credit, or elect the reduced credit, which cuts the credit by the maximum corporate rate under §11(b), currently 21%. The deduction side of that interaction is covered in what Section 174A means for your taxes.

For SaaSCo, electing the reduced credit on its stronger method, the ASC:

Gross credit: $231,000
Reduction: $231,000 x 21% = $48,510
Reduced credit: $231,000 - $48,510 = $182,490 (equivalently, $231,000 x 79%)

Illustrative only - the election is fact-specific and locked in for the year.

The reduced credit is the common posture, for a practical reason: it keeps the credit-deduction interaction inside the credit computation instead of adjusting your §174A deduction through the return, state conformity, and every downstream schedule. The election is made on the face of Form 6765 on a timely filed original return, so it is not something you can revisit later.

Step 7: Turning the credit into payroll cash

SaaSCo has no income tax liability. Without §41(h), its $182,490 would sit as a carryforward waiting for profitability that may be years away.

§41(h) lets a qualified small business elect to apply up to $500,000 per year of the credit against payroll taxes instead. The $500,000 ceiling applies to tax years beginning after December 31, 2022, raised from $250,000 by the Inflation Reduction Act.

The definition is narrower than “small,” and this is the test people get wrong. Under §41(h)(3), a qualified small business must have:

Gross receipts of less than $5,000,000 for the credit year, and
No gross receipts for any taxable year preceding the 5-taxable-year period ending with the credit year.

That second prong is a hard cutoff, not a size test: a company with $2 million of receipts that first had revenue eight years ago is disqualified no matter how small it stays. SaaSCo passes both, with its first receipts inside the five-year window. The election is also limited to five taxable years under §41(h)(4)(B)(ii).

How the cash actually arrives, under Section 3111(f) and Form 8974:

The election is made on Form 6765, Section D, with a timely filed income tax return.
The credit becomes available in the first calendar quarter beginning after the date the income tax return is filed, so filing timing controls when the money starts.
The first $250,000 can only be used against the employer share of Social Security tax. Any remaining elected credit is then taken against the employer share of Medicare tax.
It is claimed quarter by quarter on Form 8974, filed with Form 941, 943, or 944. Anything unused in a quarter carries forward to the next one.

SaaSCo’s $182,490 is below $250,000, so all of it runs against the employer Social Security share, drawing down over the quarters following the return. In the example, that is cash reducing payroll deposits for a company with zero income tax liability. Our payroll tax offset page covers the timing.

What the deliverable actually contains

A credit number on its own is not a study. The package that supports it should include:

A technical narrative per business component, tying the uncertainty, the alternatives evaluated, and the experimentation to the four-part test, drawn from the engineers’ own record rather than boilerplate.
QRE workpapers showing the wage-by-wage build, the qualified-time percentage for each role and how it was supported, the cloud cost trace, and the contractor analysis including the substantial-rights and economic-risk review.
The exclusion analysis, documenting what was removed and why. What you left out is evidence of methodology.
Form 6765, including Section G, Business Component Information, on the redesigned form. Section G is optional for tax years beginning before 2026 and required for tax years beginning after 2025, with optional reporting retained for qualified small businesses claiming the reduced payroll tax credit and for original-return filers whose total QREs, measured at the controlled group level, are $1.5 million or less with average annual gross receipts for the prior three years of $50 million or less. Optional today does not mean unnecessary: it is the IRS naming exactly what it wants per component.
Form 8974 support where the payroll election applies, so your payroll provider can actually apply it.

What this means for your company

SaaSCo is invented, and the numbers above prove nothing about what your company would receive. What the example does show is that the outcome is driven by inputs you already have: engineering payroll, a defensible qualified-time percentage, your cloud bill, your contractor invoices, and your prior three years of research spend. Those are the five things that move the number, and none of them require a lab coat.

If your situation looks structurally like SaaSCo’s, a domestic engineering team resolving real technical uncertainty at a company too young or too unprofitable to use an income tax credit, the credit computed both ways plus the §41(h) payroll election is the combination worth pricing out. Sketch the shape with our credit estimator, see what an engagement involves on how it works, and start on the qualification side with what actually qualifies for SaaS companies.

Sources

IRC §41 - Credit for increasing research activities (U.S. House, Office of the Law Revision Counsel)
IRC §280C - Certain expenses for which credits are allowable
IRC §3111 - Rate of tax
IRS - About Form 6765 and the Instructions for Form 6765
IRS - About Form 8974 and the Instructions for Form 8974
IRS - Qualified small business payroll tax credit for increasing research activities
IRS - Research credit