Four-Part Test Screening

What it does

The agent reads the evidence for each activity and tags it as qualified, unqualified or needs review. Each tag comes with reasons, so a reviewer can agree or overrule quickly. It is a first pass to save time. A licensed CPA or enrolled agent makes the call.

The four-part test is the heart of the federal research credit. In general terms, the work must be done for a permitted purpose, such as developing a new or improved product or process. It must rely on principles of the physical or biological sciences, engineering or computer science. It must aim to eliminate uncertainty about how to develop or improve something. And it must involve a process of experimentation. The exact wording and how it applies are for professionals to interpret, and the agent organizes the evidence against each part.

A worked example (illustrative)

This scenario is invented. A team logs three kinds of work on one product. The first is a new matching algorithm, with notes about why two approaches failed. The agent tags it as likely qualified and cites the notes on uncertainty and testing. The second is a change to button colors and wording, which it tags as likely unqualified, because there is no technical uncertainty recorded. The third is a performance fix with a one-line ticket. The agent cannot tell whether the cause was unknown or routine, so it marks it as needing review and prompts the engineer for a sentence on what was uncertain. The engineer answers and the reviewer decides.

Why a first pass helps

Without screening, a reviewer has to read every record to decide what to look at. With it, the obvious cases are sorted and the reviewer's time goes to the doubtful ones. The reasons shown also make disagreement cheap: it takes seconds to see why a tag was given and to overrule it. And because the same logic is applied to every project, inconsistencies between projects stand out.

Limits

The agent can only tag what the evidence shows. Where records are missing it will say so, and it cannot supply facts that nobody wrote down. It can be wrong in either direction, tagging something qualified that a professional would exclude, or the opposite. That is why every tag is a suggestion. The agent does not decide eligibility, give tax advice or guarantee a credit, and forms are filed through an authorized IRS e-file provider or tax software.

Common weaknesses the screening finds

Most thin spots are about wording and records, not about the work. Tickets titled "fix" or "update" say nothing about uncertainty. Notes that describe what was built, but not what was unknown, make routine work and research look the same. Evidence that was written months later is harder to trust than evidence made at the time. And descriptions that are copied from one project to another suggest nobody thought carefully about either.

The agent flags these patterns and prompts for a better description. Often a two-sentence answer from the person who did the work is enough to turn a "needs review" into a clear tag one way or the other.