When the IRS changes a form, it’s rarely about the form.

Section G of Form 6765 is a good example of that.

On the surface, it looks like an administrative update — more fields, more structure, more detail. But if you step back, Section G is something else entirely:

It’s a signal.

A signal about what the IRS cares about now.
A signal about how R&D claims will be evaluated going forward.
And a signal that the era of loosely reconstructed, after-the-fact R&D narratives is ending.

Why Section G Exists at All

For decades, the R&D credit lived in a gray zone.

Companies claimed credits based on:

• job titles,

• high-level descriptions,

• percentage estimates,

• and documentation assembled long after the year ended.

Audits happened, but substantiation expectations were inconsistent. In many cases, companies didn’t seriously organize documentation unless — and until — an audit notice arrived.

Section G changes that posture.

It formalizes something the IRS has been pushing toward for years:
R&D must be supported at the business-component level, with contemporaneous evidence.

Not stories.
Not summaries.
Evidence.

What Section G Is Really Asking For

Section G requires taxpayers to identify:

• their significant business components, and

• the wages tied to each, broken down by role (direct research, supervision, support).

That sounds procedural — but the implication is deeper.

The IRS is saying:

“Don’t tell us that your company does R&D.
Show us where the R&D lives, who performed it, and how uncertainty was resolved — component by component.”

This aligns directly with how R&D claims are examined in audits and tax court, even if prior filings didn’t reflect it.

Documentation vs. Substantiation (Why This Matters)

A critical distinction is emerging — and Section G reinforces it.

Documentation is having records.
Substantiation is proving a claim.

You can have:

• timesheets,

• tickets,

• commit histories,

• design docs,

and still fail to substantiate an R&D credit if those artifacts don’t clearly demonstrate:

• technological uncertainty,

• experimentation,

• and advancement tied to a specific business component.

Section G isn’t asking for more paperwork.
It’s asking for better alignment between facts and the §41 tests.

Why “We’ll Figure It Out Later” No Longer Works

A recent Tax Court trend — including cases like Kyocera — has reinforced a simple principle:

Contemporaneous means contemporaneous.

You cannot retroactively manufacture substantiation and call it contemporaneous just because underlying work existed.

Section G codifies that expectation. It doesn’t technically require new documentation — but it forces companies to admit whether they actually have it.

Waiting until:

• year-end,

• an audit notice,

• or a consultant questionnaire,

is increasingly risky.

What the IRS Is Actually Signaling

Zooming out, Section G tells us several things:

First, the IRS is not questioning whether software companies do R&D.
They are questioning whether companies can prove it cleanly and consistently.

Second, enforcement is shifting away from debating eligibility in theory and toward examining process discipline in practice.

Third, the IRS expects R&D documentation to evolve alongside modern development workflows — not lag years behind them.

This is especially relevant as AI-assisted development, rapid iteration, and distributed teams become the norm.

What “Good Enough” Looks Like Going Forward

Section G does not require perfection.

But it does require:

• clarity about what was being built or improved,

• traceability from people → activities → business components,

• and evidence that experimentation occurred as part of a disciplined process.

For most tech companies, that already exists — just not organized with tax substantiation in mind.

That gap is what Section G exposes.

The Real Takeaway

Section G isn’t about compliance theater.

It’s a warning shot that says:

“The rules haven’t changed — but our expectations have.”

Companies that treat R&D documentation as a living system — not a last-minute exercise — will adapt easily.

Companies that rely on loose estimates and reconstructed narratives will find the ground shifting under them.

The credit is still there.
The work still qualifies.

But how you prove it now matters more than ever.