For years, companies claimed the R&D tax credit using a familiar approach:

• pull wage data,

• identify high-level activities,

• assemble narratives later if needed.

That model is breaking down.

Beginning with the rollout of Form 6765, Section G, the Internal Revenue Service is signaling a clear shift in expectations — not just what must be claimed, but how R&D work must be substantiated.

This change doesn’t eliminate the credit.
But it does change the operating assumptions around documentation.

The Big Shift: From Aggregated Narratives to Business Components

Historically, the IRS did not require R&D documentation to follow a rigid format at the time of filing. Taxpayers relied on:

• time data,

• job roles,

• activity descriptions,

• and post-hoc narratives if questions arose.

Section G changes that posture.

Going forward, the IRS expects R&D activity to be organized and supported by Business Component.

A Business Component includes any:

• product,

• process,

• software,

• technique,

• formula,

• or invention
  developed or improved for use in the business or for sale.

This is not a new definition — it has always existed in the statute.
What’s new is the enforcement emphasis.

What Section G Is Asking For

Section G requires taxpayers to:

• identify significant business components,

• associate qualified wages to each component,

• and distinguish between:

  • direct research,

  • direct supervision,

  • and direct support activities.

More importantly, the IRS expects the four-part test to be supportable at the business-component level, not just in aggregate.

This is a meaningful change in how substantiation is evaluated.

Why “Contemporaneous” Matters More Than Ever

Recent Tax Court decisions have reinforced a distinction many companies gloss over:

• Documentation is having records.

• Substantiation is proving a specific tax claim.

The court has been clear:
substantive support must be contemporaneous, not reconstructed after the fact.

Waiting until an audit to reverse-engineer business components, experiments, and uncertainties is increasingly risky — especially when Section G disclosures are part of the original filing.

Put simply:
you can’t create “contemporaneous” evidence retroactively.

Why This Is Harder Than It Sounds

Most engineering teams already generate large volumes of artifacts:

• Jira tickets,

• Git commits,

• pull requests,

• test results,

• design docs.

The challenge isn’t lack of data.

The challenge is that these artifacts:

• aren’t organized by business component,

• don’t explicitly tie to uncertainty or experimentation,

• and weren’t created with tax substantiation in mind.

Section G doesn’t require new work.
It requires new structure.

What Smart Preparation Actually Looks Like

Preparing for Section G does not mean turning engineers into tax reporters.

It means:

• identifying business components earlier,

• aligning existing artifacts to those components,

• and making uncertainty and experimentation easier to see in hindsight.

Examples include:

• framing Jira tickets around technical uncertainty rather than tasks,

• using pull request discussions to capture design tradeoffs,

• documenting why alternatives were rejected — not just what was built.

This is about visibility, not bureaucracy.

Why Waiting Is the Riskiest Option

The IRS has delayed mandatory Section G requirements until the 2026 filing year, which gives companies time.

That delay should not be mistaken for leniency.

Companies that wait until the first mandatory year to rethink documentation will be forced to:

• retrofit processes,

• reconstruct narratives,

• and defend decisions made long after the fact.

The companies that fare best under increased scrutiny are the ones that:

• adapt gradually,

• treat documentation as an operational habit,

• and build defensibility as work happens.

The Real Risk Isn’t Compliance — It’s Economics

Poor R&D substantiation doesn’t just create audit risk.

It creates economic risk.

When credits are challenged:

• management time is consumed,

• professional fees escalate,

• interest and penalties compound,

• and transactions can be delayed or discounted.

In extreme cases, the cost of defending weak documentation can exceed the value of the credit itself.

That’s not hypothetical. It’s observable.

The Takeaway

Section G doesn’t change what qualifies for the R&D credit.

It changes how clearly companies must be able to prove it.

The credit still rewards:

• technological uncertainty,

• experimentation,

• and human judgment.

What’s different is that the IRS increasingly expects those elements to be:

• visible,

• structured,

• and contemporaneous.

For companies that treat R&D as a system — not a year-end exercise — this shift is manageable.

For those relying on reconstruction and luck, it’s a warning.