The Shop Records You Already Keep: How Custom Manufacturers Can Turn Travelers, First Articles, and Scrap Tags Into R&D Credit Support

TL;DR

• Job shops already generate much of what supports an R&D credit claim: routers and travelers, CAM program revisions, first-article reports, scrap and rework tags, and engineering change requests.
• The weak link is usually the connection, not the paperwork. A one-line note on why a change was made, and which project a person worked on, can turn routine records into useful support.
• Work after a part or process is ready for commercial production generally falls outside the credit, so records that show where development ended are as valuable as records that show where it started.

Why this is a records problem, not a paperwork problem

The federal research credit (IRC Section 41) rewards businesses that spend money on qualified research: work aimed at developing or improving a product or process through a process of experimentation, to resolve real technical uncertainty. Custom manufacturers often do this work every week. A new fixture geometry, an unfamiliar alloy, a tolerance no one has held before.

What owners and CFOs often tell us is that the work happened but “nobody wrote it down.” In practice, a lot did get written down. It just lives in five different systems and nobody labeled it as development. The Treasury regulations ask that a claimant keep records in sufficiently usable form and detail to show the credit is allowable (see Treas. Reg. 1.41-4), and they do not prescribe a single format. That leaves room for a shop to use what it already has, depending on the specific activities, documentation, and facts.

The records you may already have

Here are the shop records that most often carry useful information, and what each can show.

Routers and travelers. A traveler with handwritten notes (“changed feed rate, chatter at op 30,” “re-fixtured, second setup added”) can show that the process was not routine. A clean traveler for a repeat job shows nothing. The notes are the evidence.

CAM and CNC program revision history. If your programming software keeps version history, it is a dated trail of iteration. Version 1 failed to hold position, version 4 did. Even a short revision comment such as “tool deflection at 3x depth, changed toolpath” helps connect the change to a technical problem.

First-article inspection reports. A first-article that failed, led to a change, and then passed documents the loop of trying something, measuring the result, and adjusting. Keep the failed one. It is often the most useful document in the file.

Scrap, rework, and nonconformance tags. These record what did not work and sometimes why. Failure is not a reason a project cannot qualify, and a log of failed attempts can show that you were testing alternatives, not repeating a known process.

Fixture and tooling drawings. Dated drawings and markups for custom fixtures show design iteration. Pair them with a note on what problem the fixture was meant to solve.

Quotes and engineering change requests. Customer ECRs, deviation requests, and your own internal change notes show where a spec could not be met as drawn and what was tried. They also help with a separate question: whose risk and whose rights govern the work.

Test and trial material records. Material requisitions for trial stock, consumed test material, and gauge or CMM results can tie supply costs to a specific development effort.

The one-line habit that does most of the work

Most of these records answer “what happened.” The credit conversation also needs “why.” A single line next to a revision or a rework tag, stating the technical problem being solved, does more than a page of retroactive narrative.

Three small habits tend to pay off:

1. Add a “why” to changes that are not routine. One sentence on the revision, the traveler, or the ECR.
2. Name the project. Whether through a job number, a project code, or a naming convention, make sure someone can later tell which development effort a record and a person’s time belong to.
3. Keep the failures. Do not clean out scrap logs, superseded programs, or failed first articles before the year closes.

None of this requires new software or a timekeeping overhaul. It requires deciding, once, what gets a “why” and which project code it carries.

Connecting people to projects

The credit generally looks at wages for people who performed qualified research, directly supervised it, or directly supported it. Payroll alone does not tell that story. What helps is a reasonable link between a person’s time and the projects above: a production supervisor’s notes on a trial run, a programmer’s revision comments, an engineer’s markups. Contemporaneous records are generally stronger than reconstruction, but there is no required form, and many shops can build a reasonable picture from payroll, job numbers, and the technical files above.

Be aware of the boundary on the other side too. Under the regulations, work after a component is ready for commercial production, including trial production runs and troubleshooting production equipment, is generally treated as post-production and falls outside qualified research (Treas. Reg. 1.41-4(c)(2)). In a job shop that line matters. Records that show when a process was signed off for production help define where development ended, and that keeps a claim focused on work that may genuinely qualify.

What to keep, what to skip

Keep the records that show uncertainty, iteration, and outcome: notes on travelers, revision comments, failed and passed first-article reports, scrap analysis, fixture markups, and the project or job code that ties them to people and costs.

You can generally skip rewriting history. Nobody needs a formal lab notebook. Records that were created at the time of the work, in the ordinary course of business, are often the most credible.

Also keep an eye on contract terms. If a customer pays you regardless of outcome and you keep substantial rights in the work, the picture is different from a deal where payment depends on success or the customer owns the result. The records above help sort that out, but the contract language usually decides it.

Where a specialist fits

You do not need to organize any of this before talking to someone. Strata offers a $0 initial assessment, and a typical study often takes 4 to 8 weeks, depending on the business. The time commitment on your side is generally modest, because most of the work sits with the provider, and you pass the finished package to your CPA, who files the return. Strata’s service tiers range from a self-serve option to full-service and enterprise engagements, so you can choose how much of the work you carry. As one reference point, a Marine & Boat Manufacturing client recently identified $300K+ in credits. Individual results vary, and every claim depends on its own facts.

If you want to know whether your shop’s records already tell the story, book a 15-minute review at stratataxgroup.com/contact.

Sources

• IRC Section 41, Credit for increasing research activities (Cornell LII)
• Treas. Reg. 1.41-4, Qualified research for expenditures paid or incurred in taxable years beginning after December 31, 1985 (eCFR)
• About Form 6765, Credit for Increasing Research Activities (IRS.gov)

This post is for informational purposes only and does not constitute tax or legal advice. Consult a qualified tax professional regarding your specific circumstances.