The Question

“How much R&D credit can we actually expect?”

This is the most common question founders and CFOs ask — and also the one most often answered poorly.

Some sources throw out eye-catching percentages.
Others quote wide dollar ranges with no context.
A few imply certainty where none exists.

The reality is more nuanced.

The value of the R&D credit isn’t driven by a single factor. It’s shaped by how your team is structured, where your engineers sit, how mature your development process is, and how rigorously the work can be supported.

The goal of this article isn’t to promise a number.
It’s to give you a framework to understand what actually drives value — and why two companies that look similar on the surface can end up with very different outcomes.

The Short Answer

For most tech companies, the R&D credit ends up being a meaningful but bounded benefit, not a windfall.

It tends to be largest when:

• engineering payroll is substantial and well-documented

• a high percentage of work meets the technical standard

• employees are U.S.-based

• state credits meaningfully layer on top of the federal credit

• the company has consistent R&D activity year over year

It tends to disappoint when:

• expectations are set before understanding qualification rules

• documentation is thin

• payroll is small or heavily offshore

• the company is very early or very operational

• estimates are driven by optimism instead of evidence

The difference between “this was worth it” and “this wasn’t” usually comes down to how well the company understands what it’s paying for and what it can actually support.

The Deep Dive

1. What Actually Drives the Size of the Credit

At its core, the R&D credit is a payroll-driven incentive.
That surprises some founders, but it shouldn’t.

The largest component of qualified research expenses for software companies is almost always wages — specifically, wages paid to employees who are directly performing, supervising, or supporting qualified research activities.

From there, several factors materially influence the result.

Payroll mix matters.
A team composed primarily of engineers, data scientists, and product developers will generate a very different outcome than one dominated by QA, customer support, or implementation. Titles don’t determine qualification — activities do — but role composition strongly influences how much work ultimately qualifies.

Location matters.
Federal credits only apply to U.S.-based work.
State credits apply only to work performed in that state.
Two companies with identical headcount but different geographic footprints can see very different total benefits once state layering is considered.

Maturity matters.
Early-stage teams often spend a high percentage of time solving technical uncertainty, but lack documentation. More mature teams may spend less time on “pure R&D” but have stronger systems that support what they do qualify.

Neither is inherently better — but they produce different outcomes.

2. Why Two Similar Companies Get Very Different Results

This is where expectations often break down.

Two SaaS companies might both have:

• 15 engineers

• similar payroll

• similar revenue

• similar products

Yet one generates a credit that feels meaningful, while the other ends up underwhelmed.

Why?

Usually because of differences in how work is defined, tracked, and supported.

One team clearly distinguishes between experimental development and routine implementation. They document architectural decisions, tradeoffs, and failed attempts. Their project structure maps cleanly to technical uncertainty.

The other team moves just as fast — but doesn’t capture the “why” behind decisions. Everything lives in people’s heads. Looking back, it’s hard to separate genuine experimentation from execution.

The credit doesn’t reward effort.
It rewards substantiated technical risk.

That distinction explains most of the variance.

3. Federal vs. State: How the Pieces Add Up

The federal credit is the foundation.
State credits are the multiplier — but only in the right states.

For many companies, the federal credit alone represents the bulk of the benefit. State credits become meaningful when:

• a large portion of payroll is concentrated in a high-value state

• the state offers refundability or long carryforwards

• the administrative burden is reasonable

• the company is prepared to meet state-specific requirements

In states like Texas (starting 2026), Michigan, Minnesota, and California, state credits can materially increase total value. In others, they may add complexity without much incremental benefit.

The key mistake companies make is assuming state credits are “automatic.”
They aren’t. Each one has its own economics.

4. When Expectations Are Usually Too High

Expectations tend to overshoot reality when:

• companies apply a flat percentage to total engineering payroll

• qualification is assumed without reviewing activities

• offshore or contractor-heavy teams are included incorrectly

• documentation requirements are underestimated

• amended claims are treated like current-year claims

• timing is ignored (credits ≠ immediate cash)

In these cases, disappointment doesn’t come from the credit being “bad.”
It comes from expectations being built on the wrong inputs.

5. When Expectations Are Usually Too Low

On the flip side, companies often underestimate value when:

• they focus only on “big breakthroughs” and ignore iterative development

• internal tools and infrastructure work are overlooked

• algorithm refinement and performance optimization are discounted

• state credits are ignored entirely

• payroll tax offsets or refundability aren’t considered

• credits are viewed as one-time events instead of annual systems

These companies often have more qualifying activity than they realize — they just haven’t connected it to the credit properly.

6. How to Think in Ranges, Not Promises

The healthiest way to approach the R&D credit is not to ask,
“How much will we get?”

It’s to ask:

• How much qualifying activity do we actually have?

• How defensible is our documentation?

• Where is our payroll located?

• How consistent is our R&D year over year?

From there, ranges become meaningful.

They allow you to plan conservatively, avoid surprises, and integrate the credit into broader tax and cash-flow planning — without relying on optimistic assumptions.

Any approach that starts with a guaranteed number before answering those questions is starting in the wrong place.

The Takeaway

The R&D credit is neither a jackpot nor a rounding error.
For most tech companies, it sits in the middle — meaningful, repeatable, and highly dependent on execution.

Its value is driven less by how much you spend, and more by:

• how you spend it

• how you document it

• where the work happens

• how consistently you approach the credit

Companies that get the most out of the R&D credit aren’t the ones chasing the biggest number. They’re the ones that understand the system, set realistic expectations, and build processes that make the credit defensible year after year.

That’s how the credit becomes part of a strategy — not a surprise, and not a gamble.