R&D Tax Credit: Getting Paid Isn't the Problem, Your Contract Is

David Wiener: Work product. All drawings, calculations, designs, methods, and know-how developed under this agreement shall be the sole and exclusive property of the client. I wrote that sentence. But if you run an engineering firm, you've signed something close to it, probably more than once, and probably without anybody reading it out loud. Here's what it can cost you. Say your people spend four months on a foundation for a site where the soil had beaten everybody before them. Two designs failed review. The third one held. That's exactly the kind of work the R&D tax credit was written to reward. Sign that paragraph, and the credit on that project can be gone. All of it. How good the engineering was won't matter, and neither will how hard the problem was. Getting paid by a client isn't what costs you the R&D credit, your contract is. And most firms learn what theirs says on the day somebody from the IRS reads it back to them. This is the Tax Strategy Playbook. I'm David Wiener, and this is part two of a four-part solo series on the R&D tax credit. Part one, two weeks back, was for startups that aren't profitable yet. This one's for anybody whose technical work gets paid for by somebody else. Engineers, architects, software developers, product designers, Manufacturers building to a customer's spec. One question runs all the way through it. When a client pays for the work, whose credit is it? By the time we're done, you'll know the two tests that decide that question, and that passing one of them isn't enough. You'll know the two contract clauses that drive those tests, and why they're rarely in the document you'd check first. And you'll hear what two federal appeals courts did to engineering firms that claimed this credit, plus the piece of it that's still being fought over. The back half has a software shop that loses the credit on one project and keeps it on another. Same client, same year. It also has one setup where the credit just disappears and nobody gets it. The risk test comes first. The clause that can wipe you out comes later, so don't bail at minute 10. Engineering firms. Architecture practices, software shops, product design studios, if most of your revenue comes from technical work on somebody else's project, nearly every research dollar you spend runs through somebody else's paperwork. You're the reason this episode exists. Custom manufacturers, you're in this too. A lot of what follows lives in your purchase orders and the terms printed on the back of them. If you're on the buying side, Paying outside engineers or developers to solve problems for you, hang on until near the end. That same two questions run in reverse and they decide whether you can count what you paid. Tax professionals, a few short notes along the way are aimed at you. When a service firm's claim gets examined, funded research comes up early, and the answer lives in the contracts, not the engineering files. You're usually the one who can get a client to dig those contracts out. One more thing. Episode nine with Brian Broussard from CSSI covered what makes work qualified in the first place. Go back to that one for the basics. Here, I'm assuming the work qualifies. The fight is over who owns the credit. One word in the tax code runs this entire episode. Funded. The R&D credit has a list of things that don't count. Research outside the United States. Market research. Routine quality control, and funded research. The tax code says qualified research doesn't include, quote, any research to the extent funded by any grant, contract, or otherwise by another person or governmental entity. That's section 41D4H. Here's the idea underneath that. The credit is built to reward whoever puts money at risk trying to solve a technical problem. If somebody else paid for the work, took the risk, and walked away with the results, the firm that did the work doesn't get the credit. If anybody does, it's the one who paid. The credit follows the risk and the rights. It doesn't follow the labor. Think about it from the government side. One research project, one credit. If the firm doing the work and the client paying for it could both claim the same research, the treasury would be paying for it twice. So the rules pick one of them, and they pick by asking who was really exposed. So being paid isn't the problem. Every firm gets paid. The question the regulations ask is narrower. Were you going to be paid whether the research worked or not? And did you have any right to use what you figured out? Those are the two tests, and you have to pass both. If you fail the first and you can lose some or all of the credit on that project, fail the second and you lose all of it. Two quick notes on where the rules live. The regulation that spells this out is old, written in the nineteen eighties, but today's R&D regulations point straight back to it. It still governs and it says quote, all agreements, Not only research contracts entered into between the taxpayer performing the research and other persons shall be considered, every agreement, including the ones without contract on the cover. So let's take the first test. When the work fails, whose money pays for it? Back to the bad soil with a made up firm. It signs on to design a foundation system for a difficult site. The first approach fails review. The second fails too. The third one works. Now give the firm an hourly contract. The client pays for every hour, including the hours spent on the two designs that failed. Ask yourself who was exposed when the first two approaches didn't work. Not the firm. The firm was getting paid either way. The client carried the risk. Under the regulations, that research was funded, and the credit isn't the firms to claim. Nothing about the engineering changed. Same soil, same three attempts, same uncertainty. The only thing that decided it was who was on the hook when it went wrong. The regulation comes at it from both sides. From the side of the firm doing the work, it says the amounts that are quote. Contingent on the success of the research and thus considered to be paid for the product or result of the research are not treated as funding. Translation, if you only get paid when it works, that payment isn't funding. You were selling a result and you are carrying the risk. From the side of the company writing the checks, a companion rule says that for the payer to count the research, the agreement has to require the payer to quote. Bear the expense even if the research is not successful. Same idea from the other angle. Somebody has to be exposed. Whoever's exposed holds the credit. So what does contingent on success actually look like in a contract? This is where most firms are surprised because the language they point to usually isn't it. A promise to do the work to a professional standard of care isn't it. A promise to meet the building code. Isn't it. A clause that lets the client terminate if you fail substantially to perform isn't it. Inspections and acceptance rights by themselves aren't often enough either. Courts have looked at all of those and said the same thing. That's a promise to do the job properly. It's not a promise that the research succeeds. The Eighth Circuit drew exactly that line in 2024. Proper performance versus Successful performance. I'll come back to that case in section three a little bit later. What does count in the cases that went the taxpayer's way is payment that actually turns on the technical result. Detailed specifications the work has to meet. Payment that's withheld or has to be given back if it doesn't meet them. That's rare in ordinary professional services contracts, and that's the point. Most contracts are silent on what happens when the technical approach doesn't work, and silence usually means you get paid anyway. Manufacturers hit the same wall. Take a made up one that develops a new custom part for a customer. The customer's purchase order pays a separate development charge, sometimes called an engineering or tooling charge up front, and then pays per piece once production starts. Ask the same question. If the first three prototypes fail, does the manufacturer still keep that development charge? If it does, the customer funded part that part of the research, whatever the invoice calls it, the word on the purchase order doesn't decide it. who is exposed when it fails does. Before the contract shapes, a small favor. If you're getting something out of this, hit follow wherever you listen or watch on YouTube. It costs you nothing, and it's the main way anybody new finds this show. The common contract shapes cleared first. Hourly or time and materials with no limit. The client pays for every hour whether it works or not. That's the clearest case of funded research that there is. Then there's capped. You you bill your costs, sometimes plus a markup, up to a maximum. Firms like these because they feel risky. There's a ceiling, and if you blow through it, you eat the rest. But a federal appeals court looked at exactly these contracts in 2015 and held that the research was funded because up to the gap the client paid whether the research succeeded or not. That's the GeoSyntech case. More on it down in section three. Then there's fixed fee. One price for the job. If the work takes twice as long, that's your problem. That's the shape where reasonable people disagree, and I'm gonna label it that way rather than pretend it's settled. Hold that thought for section three because two court cases pull it in different directions on it. And one more shape that shows up constantly, milestone billing. You get paid when you finish schematic design, then design development, then construction documents. Or when you deliver the alpha, the beta, the release. That feels like payment tied to success. But read what the milestone actually is. Usually it's that a phase got delivered, not that a technical question got answered. Courts have looked at payment tied to delivering the work rather than to the research succeeding and treated it as payment for proper performance. If the client has to pay for the phase, whether or not your new approach works, The milestone didn't put you at risk. The Eighth Circuit put one more marker down that's worth knowing. Part of what it noticed was that the firm couldn't point to anything in its contracts requiring it to give back money it had already been paid if specific benchmarks weren't met. A clause that says you get paid when it works is one thing. A clause that says you give the money back if it doesn't is another. Most professional services contracts have neither one. I'm not telling you to rewrite your contracts that way. Your client may never sign it, and there are business reasons why you might not want to offer it. I'm telling you what the courts have looked for so when you and your attorney decide what to sign, you know what you're trading. Now for some good news. Failing the risk test doesn't automatically wipe out the project. If you kept your rights in the research, and that's the second test, the regulation says the research is funded, quote, To the extent of the payments you're entitled to for doing it. So the payments come off your qualifying costs. If the client paid you 100% of what the research cost, there may be nothing left. But if a project ran over what the client paid and you ate the overrun, that part can still count. Here's the simplest version, an illustration with round numbers. A firm keeps its rights and its qualifying research cost on a project came to $300,000. The client paid $200,000 and that payment didn't depend on the research working. The $200,000 is funding. The $100,000 the firm ate is the part that can still be in the credit. Not the whole project, not zero either. That's the difference between failing the risk test and failing the rights test. Tax professionals one note, when funding has to come off, and you can document the project's total research costs and show that they're more than the funding. The regulation lets you spread it across the project's qualifying and non qualifying costs pro rata instead of putting all of it against the qualifying side. But no less than 65% of the funding can go against the qualifying costs. So that's the risk test. It can cost you some of the credit, or it can cost you all of it. The second test doesn't negotiate. First a quick checkpoint. Client paying you doesn't cost you the R&D credit. A client paying you no matter how the research turns out can. And that was settled in your contract before anybody ran the first calculation. Our Tax Strategy Playbook newsletter is at Taxstrategeplaybook.com slash newsletter. It's free, and my subscribers get my 2026 tax planning guide and the playbook notes for every episode at no charge. This episode's notes have both tests on one page, plus a short list of contract questions to take to your attorney. And reply to me with one word, yes or no. On your biggest client contract, Are you allowed to use what your people figure out without paying the client for that right? If you're not sure, that's a no. Test two is the one that zeroed out the firm in the cold open. It's shorter than test one and it's harsher. Here's the regulation word for word. If a taxpayer performing research for another person retains no substantial rights in research under the agreement providing for the research. The research is treated as fully funded for purposes of Section 41D4H, and no expenses paid or incurred by the taxpayer in performing the research or qualified research expenses. Fully funded, no expenses. That's not a reduction. There's no partial credit and no overrun math. If you kept no substantial rights, the project is gone as far as you're concerned, and it doesn't matter how much risk you carried on test one. So what's a substantial right? The plain version, the right to use the results of your research in your own business. And the regulation tells you three things that don't count. One, getting better at your job. The regulation says quote, incidental benefits to the taxpayer from performance of the research, for example, increased experience in a field of research, do not constitute substantial rights in the research. Your engineers learned a lot on the project. That's not a right. Neither is a nice photo of the building for your website. Two, having to pay to use it. Quote, a taxpayer does not retain substantial rights in the research if the taxpayer must pay for the right to use the results of the research. If your contract makes you license your own work back from the client, you didn't keep anything. And then three, handing the client the exclusive rights to exploit the results. The regulation uses that as its own example. If the agreement gives another person the exclusive rights to exploit the results, the research is treated as fully funded. Now the part that's better news than people expect. You don't need exclusive rights. You don't need to own the patent. The Federal Circuit said it plainly back in 2000 in a case for involving Lockheed Martin. The right to use the research results even without the exclusive right is a substantial right. A non exclusive right to use what you learned in your own business without paying for it can be enough. So how does a firm lose a right that low? Usually they do it without noticing it. Work for higher language. Clauses that assign all intellectual property and all work product to the client. Those are ordinary in architecture, engineering, and software contracts. Clients ask for them, firms sign them, and on the day of signing it feels like a formality about who owns the drawings. It's also a decision about a tax credit made years before anybody calculates one. The distinction that matters is between the deliverable the client bought and the know-how you developed to produce it. The client can own the drawings for its building. The question is whether your firm kept the right to use the method, the calculation approach, the detail you worked out on the next job. Firms that keep the second one keep the argument. Firms that sign everything away, including the know-how, generally don't. So let's make it concrete for an architect. The client owns the design for its stadium, its tower, its hospital. Nobody's arguing with that. But in the course of that project, your people worked out a new way to detail a long span roof connection or a facade system that handles wind load differently. Can you use that approach on the next project without asking the client? If your agreement says yes, or simply doesn't take it away, you've likely kept something. If it says every concept, method, and detail developed under the agreement belongs to the client, and you need their permission to use any of it again, you may have kept nothing. Software contracts often use two terms for this. Background intellectual property is what you walked in with. Foreground intellectual property is what gets created on the project. A lot of agreements let you keep the background and hand over all of the foreground. That's the part to read slowly because research usually lives in the foreground. Tax professionals. When you're scoping a credit for a services firm, ask for the master agreement and every amendment, not just a statement of work. The rights question is answered in the regulations words under the agreement providing for the research. If the agreement that controls ownership isn't in your file, The analysis isn't done. And here's where it hides. Not in the proposal, not in the friendly two-page work order, in the master service agreement behind it, or the standard terms and conditions, or the fine print on the back of a purchase order. Remember the regulation. All agreements count. A one-page work order that references a 15-page master agreement nobody's opened since it was signed. The master agreement counts. And if it says every scrap of work product belongs to the client, the friendly language up front may not save you. Here's another made-up example built on a pattern I see all the time: a 20-person software development shop, one big client, two projects per year. Project one is built under the client's master agreement. Paragraph 14 assigns all work products. All inventions and all know-how to the client and nothing reserves anything back. The shop's developers spent months working out a way to sync data across devices that nobody on the team knew how to build. Real uncertainty. Real testing. And under that agreement, the shop kept nothing it can use without the client's permission. Project one, fully funded. Zero. Project two, the same client signed later, after the shop's attorney pushed back. The client owns the finished application. The shop keeps a non-exclusive right to use the general techniques and tools it develops along the way. Same developers, same kind of problem. Project two likely passes the rights test, and now it goes to the risk test, which turns on how the shop gets paid. Same year, same client, same people. One project is gone, the other is still in play. That's not a loophole. That's how the regulation's written to work. And I'll show you the sentence in a minute. Look at what changed between the two projects. Not the code, not the people, not the documentation. A lawyer read the ownership paragraph before the shop signed it and asked for one sentence back. That's the cheapest piece of tax planning in this whole episode, and it happened in a contract negotiation, not at tax time. One more combination because it surprises people. What if you kept no rights and you only got paid if the research succeeded? You'd think somebody gets the credit. The client didn't carry the risk, you didn't keep the rights. The regulation says in that case, quote, neither the performer nor the person paying for the research is entitled to treat any portion of the expenditures as qualified research expenditures. Nobody. The credit just basically falls through the floor. The question I get often is my biggest client will never let me keep any rights, their lawyers won't sign it. Maybe so. That's a business decision and it's yours to make. Some clients won't move. Some will move further than you think on the know-how, as long as they're deliverable and their confidential information are protected. What I'd want you to know is it's a real trade with a tax number attached instead of a formality nobody thought about. If you sign it anyway, sign it knowing that the project's research credit is probably gone and plan around that instead of being surprised by it. So that's the second test. Now let's put both tests where the IRS puts them, project by project in front of a judge. Here's the sentence I promised you. The regulation says the funding rules quote Shall be applied separately to each research project undertaken by the taxpayer. A project for this purpose is the piece of work you're actually claiming. A design, a product, a process, a piece of software. If you did 12 projects this year under four different agreements, that's 12 answers. And the agreement behind each one decides its answer. That's why the software shop can lose one project and keep another. It's also why there's no such thing as a firm wide answer. You don't get to say our contracts are fine. You have to say which contract, which project, which terms. And the form's catching up to that. Starting with tax years that begin in 2026, which for most firms means the return you file next year, a part of Form 6765 called Section G stops being optional. It breaks the credit out by the Business component by business component instead of reporting one number. There are exemptions. Startups using the payroll election are out, and so are smaller claims on an original return, generally a million and a half dollars or less of qualifying costs at a company averaging fifty million or less in gross receipts. But if you're over those lines and you've been computing one credit for the whole firm, you're about to be asked to show your work project by project. If the funded research analysis was never done at that level, this is the year it surfaces. That's part three of this series, and it's coming in a couple of weeks. Now the courts. Two federal appeals courts, two engineering firms, and the same regulation. The first is Geosyntec Consultants, decided by the Eleventh Circuit in 2015. An environmental and engineering consulting firm asked for a refund of research credits for 2002 through 2005, about $1.7 million. Its sample contracts came in two shapes. Some were fixed price, others were capped. The firm billed its costs up to a maximum. The trial court treated the fixed price work as the firm's risk. The capped contracts were the fight, and both the trial court and the eleventh circuit held that research was funded. Up to the cap the client paid whether the research worked or not. The second is Meyer, Borgman, and Johnson. Decided by the Eighth Circuit in 2024. A structural engineering firm claimed just under $200,000 of credits for its fiscal years 2010, 11, and 13. Its contracts weren't careless. The firm described them as fixed price. They had inspection, acceptance, and quality assurance provisions. The client could terminate if the firm failed substantially to perform. The firm argued that all of that put its payments on the line. The tax court disagreed, and the Eighth Circuit affirmed. Its reasoning was the line I mentioned in section one meeting a professional standard of care as proper performance. It isn't successful performance. None of the contracts made payment depending on the research itself succeeding. Notice how that case was decided on summary judgment. That means the court didn't need a trial about the engineering. Nobody had to prove whether the structural work was hard or uncertain or innovative. The court read the contracts, and the contracts answered the question. A firm can have the best engineering records in the country and still lose this one on paper. That's the thesis of this whole episode in one court decision. The case the courts kept pointing to as the one that worked is an older one called Fairchild, an Air Force contract with more than a thousand pages of technical specifications. Where the contractor stayed at risk until the work actually met them. That's the bar the courts are holding these contracts up against. Now here's the part I want to label clearly because it's contested. Put those two cases side by side and they pull in different directions on fixed price work. In GeoSyntach, the trial court treated fixed price contracts as putting the risk on the firm. In Meyer Borgmann, Fixed price contracts with inspection and termination clauses weren't enough. How much exposure a contract has to create before the risk test is satisfied, and whether a fixed fee where you eat the overruns gets you there is something courts and practitioners are still arguing about. Nobody can tell you from a podcast how that comes out for your contracts. Anyone who says the answer's obvious hasn't read them. What isn't contested is the lesson underneath Both cases. It's not fixed price good hourly bad. The label on the contract doesn't decide it. The terms do. And the terms get read closely years later by somebody looking for one specific thing. Now run that forward to an exam. On a services firm's R&D credit, funded research is a natural early question because the firm's whole business is client work. And the answer doesn't come out of your engineering notes or your timesheets. It comes out of your agreements. Both of these appeal cases were decided on contract language, which means the records worth having here aren't only technical. They're the signed agreements. The master agreement, the work orders, the amendments, and the change orders, because a change order can quietly change who pays for what. Kept with the project they belong to so anyone can see which terms governed which work. If you're three years past a project and the only copy of the master agreement is in the inbox of somebody who left the firm, that's a problem you can fix this week, not during an exam. There's good news buried in all of this. Everything in this section is knowable in advance. You don't have to guess what an examiner will think of your engineering. You can read the same documents they'll read today and know where you stand on most of your projects, where you can't tell that's the conversation to have with your tax professional before the claim goes in, not after. If your client's a governmental agency, two things. First, the statute names governmental entities directly, so government money can fund the research the same way a private client's money can. Grants count as funding too. Second, there's a narrower rule for government contractors. Under federal contracting rules, some of a contractor's independent research and development costs and its bid and proposal costs. get allocated to its government contracts. The regulation says those amounts generally are not treated as funding the underlying research except to the extent that they are quote properly severable from the contract. Government contracting is its own specialty and this is a place for specific advice, not a general rule from a podcast. Last piece, for the companies on the other side of the table, if you pay a An outside firm to do research for you here in the United States, you can generally count 65% of what you pay toward your credit. But the same two questions are going to decide it. The agreement has to be in place before the work's done, and it has to require you to pay even if research isn't successful. And the work has to be done on your behalf, which means you have the right to the results. If you only pay when it works, you're buying a result and it's not a contract research expense. Notice the symmetry here. The hourly contract that's bad news for the engineering firm can be good news for the company paying it. The client paid for every hour, success or failure, and if it got rights to the results, the client's the one carrying the risk, the credits meant to reward. Whether the client can actually claim it still depends on the client's own facts, starting with whether the work's qualified research for the client's business. But the contract already pointed the credit in the client's direction. So between you and your contractor there's one credit, and it generally goes to whoever carried the risk and kept the rights. Sometimes, as you heard, it goes to nobody. The contract decides which. That's the framework. 15 quick ones. One, if a client pays me, is the work automatically disqualified? No. Being paid is not the same as being funded. Two, does an hourly contract usually mean the research was funded? Usually, yes. The client pays whether it works or not. 3. Does a fixed price contract guarantee the credit? No. It helps on risk. It's contested and you still need the rights. 4. Do I need to own the intellectual property? No. A non-exclusive right to use the results can be enough. 5. Does the experience my engineers gain count as a right? No. The regulation calls that an incidental benefit. 6. If I have to license my own work back from the client, then you didn't keep a substantial right. 7. Does the master service agreement matter if the work order looks fine? Absolutely. All agreements count. 8. Can one project qualify in another not for the same client? Yes, it's decided project by project. 9. If a client paid less than the research cost me, The overrun may still count if you kept your rights. ten. Does a code compliance cause make payment contingent on success? Courts have said no, that's just proper performance. 11. Do government grants count as funding? Yes. The statute names governmental entities by name. 12. If I pay a contractor to do my research, what can I count? Generally 65% if you carry the risk and get the results. 13. Can anybody end up with no credit at all? Yes. No rights plus pay only on success means nobody claims it. 14. Does a purchase order count as the agreement? Yes. So do your standard terms and conditions. And 15, can I fix this on contracts I've already signed? Talk to your attorney. Future contracts are where you really have room. Now for the playbook, six moves and the first one matters the very most. This week, pull your biggest three client agreements. Not just the proposals and the work orders, the master agreement and the standard terms behind each one. Find two sentences in each. What happens to your payment if the technical approach doesn't work and who owns the results? If you can't find either sentence, that tells you something too. This is the whole episode in one task because the credit was decided in those two sentences before anybody did the work. Number two, answer one question for each of these agreements. Do you keep the right to use what you figured out in your own business without paying the client for it? Yes, no, or I don't know. Write an answer. Write the answer next to the project name. Every no is a project where the credit may be zero, no matter how good the engineering was. Three. Start a project list for this year. Every client project with real technical uncertainty, the client's name, how you get paid, and which agreements control ownership. The IRS looks at this project by project, and if you're above the Section G lines, so will your 2026 return. Building that list now takes an afternoon. Rebuilding it in 2028 takes weeks, and half the people who'd know won't be there. Put a copy of the controlling agreement in each project's folder while you're at it. 10 minutes per project. Four, before you sign the next contract, put your contract's attorney and your tax professional in the same email. Ask the attorney one question. Can we reserve a non-exclusive right to use the general know-how we develop separate from the deliverable that the client's buying? Those two people rarely talk to each other, and that's most of the reason this problem exists. Number five, if you've already claimed the R&D credit on client work, put this to your tax professional. Was the funded research analysis done project by project with the contracts in the file? If the answer is yes, good. If it's no, you'd rather find that out from your own side of the table. And if you never claimed the credit because you assumed client work didn't qualify, ask that too. Some of it may. And episode nine covered picking up prior years. And number six, if you'd like a second set of eyes on whether your firm has a credit worth claiming and what your contracts do to it, I'll run it at no cost through CSSI. Grab a time at calendly.com/david-wiener/cs spelled W-I-E-N-E-R, or just call me at 770-224-8504-option two. No obligation, no sales pitch. If your contracts already gave the credit away or the numbers are too small to justify a study, you'll hear that from me directly. Start with number one though, this week. Last thing before I let you go. Who do you know that owns an engineering firm, an architecture practice, or a software shop? Somebody whose whole business runs on client contracts. Send them this episode. Odds are nobody told them the paragraph above about who owns the drawings is also a paragraph about the tax credit. Engineers talk to engineers, architects talk to architects. This one slips through because the person who signs the contract and the person who calculates the credit hardly ever compare notes. A 10-second text from you connects them. Just one person. The name that came to mind first is usually the right one. If you do technical work for clients, getting paid was never the problem. The R&D credit follows the risk and the rights and the labor doesn't move it. Who pays when the research fails? Who keeps what you figured out? Both answers got written down before the work started in documents most firms never connect to a tax credit. Read them before you sign the next one. This is education, not tax advice for your firm. Your facts need a tax professional who knows you, knows them, and on this topic, an attorney who knows your contracts too. My job today was making sure you know which two sentences to ask them about. In a couple of weeks, part three, the questions the IRS asks when it reviews your R and D credit and the records worth building before the year's out. Next Tuesday, the regular show's back in between. Remember, tax evasion is a crime. Tax avoidance is mandatory. I'm David Wiener, Mr. Cash Flow. See you next Tuesday.