Digital businesses and analogue challenges: start-up losses incurred in web-based businesses
Mayer Brown, New York
The application of traditional tax rules is strained by the extension of internet technology to virtually every facet of business and our personal lives. In Kellett v Commissioner, the United States Tax Court updated traditional rules regarding the deductibility of startup losses for a technology entrepreneur. This article explores how US tax rules are being adapted to 21st century businesses.
‘The [decentralised finance] DeFi boom is a very near equivalent of an apocalyptic event for the traditional financial institutions,’ wrote digital entrepreneur Mohith Agadi. Given the current meltdown in the cryptocurrency markets, however, we can be thankful that traditional banks didn’t close up shop when bitcoin hit $60,000.
There has been another recent collision of the digital world with traditional finance, but this time it’s in the realm of tax. In Kellett v Commissioner, released on 14 June 2022, the US Tax Court addressed an age-old question for a 21st century business: when are web development costs deductible as startup expenses for federal income tax purposes? And in deciding this issue the court eliminated a taxpayer-friendly revenue procedure issued by the Internal Revenue Service (the IRS) that allowed taxpayers to deduct software development costs.
In Kellett, the taxpayer launched a retail website in 2002, which he operated until 2007. He then accepted a couple of jobs with internet research firms. While working full-time, the taxpayer began work on a website called ‘Vizala’ that collated demographic, social and economic data that would be useful to any number of companies. He hired remote engineers to develop the website and to develop user interfaces using open code software. The website was functional by March 2015, the bugs were then worked out, and the website launched in September 2015. Vizala did not earn any revenue until 2019.
The taxpayer claimed deductions for the amount paid to the software engineers, to marketing companies, and for home internet access and other miscellaneous expenses on his 2015 federal income tax return. The expenses were claimed on Schedule C as trade or business expenses. The IRS challenged the deductibility of these expenses. The Schedule C probably looked suspicious to the IRS because the expenses were claimed, but no revenue was reported. The IRS asserted that the expenses were startup expenses that were required to be amortised rateably over 180 months beginning in September 2015.
The IRS’ statement of the traditional law is correct. A taxpayer may deduct ordinary and necessary expenses incurred in a trade or business. If the taxpayer has not begun its trade or business, the expenses are referred to as startup expenses. Startup expenses are not currently deductible. Instead, these expenses must be capitalised or amortised over 180 months. The Tax Court used the test enunciated by the Fourth Circuit (the Circuit that would hear an appeal of the Kellett case) to determine whether the taxpayer’s activities rose to the level of a trade or business: (as in Richmond Television Corp v United States) a taxpayer does not begin carrying on a trade or business ‘until such time as the business has begun to function as a going concern and performed those activities for which it was organized.’
In Richmond Television Corp, the court held that television staff training costs were not deductible until the television station started broadcasting. The court in Kellett, however, recognised that the receipt of revenue is not a sine qua non for a venture to establish that it is a trade or business. But under the traditional test, the venture must at least try to sell goods or services. Vizala did not even try to sell anything until after 2015.
The Tax Court then imported this age-old principle into the 21st century. The taxpayer asserted that he could not successfully sell access to his website until the website was being actively used by a significant number of users. The court found that ‘[h]e therefore prioritized web traffic over revenue by charging no user fees and marketing the site to institutional customers.’ The court accepted the taxpayer’s argument and held that the 2015 activity from and after September 2015, when the website launched, would be treated as trade or business activity for federal income tax purposes. Since the taxpayer did not offer proof of the day the website launched, the court used the last day of the month, 30 September 2015, as the launch date.
The court did not focus on when the services were rendered to Vizala. Instead, the court focused on when the expenses were paid by the taxpayer. Presumably, the taxpayer was on the cash method of accounting and the use of the payment date corresponded to when the taxpayer incurred the expense for federal income tax purposes. Accordingly, the court allowed the taxpayer to deduct all engineering expenses paid after 30 September 2015 but treated all engineering expenses paid on or prior to 30 September 2015, as startup expenses.
The taxpayer estimated that his business use of internet and telephone services constituted 80 to 90 per cent of his home internet use but did not produce any evidence other than a spreadsheet showing that he spent about 49 hours per week on developing Vizala. The court, however, split the business use with the personal use based on the sum of the 49 hours per week he worked on Vizala and the 40 hours per week he worked at his ‘day job’ against the remaining hours in the week. The court would have increased the business use percentage (by decreasing the denominator) if the taxpayer had introduced evidence of the number of hours he ate and slept and therefore was not using the internet (not a joke). The court’s methodology reduced the taxpayer’s business use of the internet by approximately half.
The taxpayer argued that if his expenditures were not trade or business expenses, they should be treated as research and development expenses. (At the time, research and development expenses could be immediately deductible. Today, these expenses must be amortised over five years.) The court, quoting Treasury Regulation s 1.174-2(a)(1), noted that such expenses ‘are for activities intended to discover information that would eliminate uncertainty concerning the development or improvement of a product.’ The court held that the taxpayer in Kellett, used open code software, however, ‘to solve a complex but familiar problem.’ Since the taxpayer was not developing a new product but only using existing software to display and analyse data, the court denied the taxpayer the right to claim a research and development deduction.
For tax aficionados, the last part of the court opinion in Kellett, is extremely interesting but did not help the taxpayer. The IRS, in Revenue Procedure 2000–50, announced that it would not challenge a taxpayer’s deduction for the costs of developing computer software even if the costs did not qualify as research and development expenses. The taxpayer argued that the IRS was estopped from challenging his deductions by reason of its position in Revenue Procedure 2000–50 even if the taxpayer’s expenses were not deductible under Code s 162 or Code s 174. The IRS countered that the revenue procedure was only available if the taxpayer was engaged in the conduct of a trade or business. The court held that this IRS position was nonsense because the revenue procedure expressly did not predicate the allowance of the deduction on the conclusion that the taxpayer was engaged in a trade or business.
In general, the Tax Court has held that the failure of the IRS to adhere to a position set forth in a revenue procedure is an abuse of discretion. But in Kellett, even though the court scoffed at the IRS’s objections to allowing the taxpayer to take advantage of the IRS’s position in Revenue Procedure 2000–50, it refused to allow the taxpayer to rely on the revenue procedure. The court held that there was no basis in law for the discretion announced by the IRS in Revenue Procedure 2000–50. Thus, the court held the IRS’s refusal to allow the taxpayer to take advantage of Revenue Procedure 2000–50 was not an abuse of discretion because the IRS had no statutory discretion to allow the deduction. Since the IRS had no basis for its position in Revenue Procedure 2000–50, the taxpayer could not claim that the IRS’s refusal to allow him to claim its benefits was abusive.
In summary, the tax court succeeded in putting old wine in new bottles in the Kellett decision. It has provided us with a new interpretation of when a business commences in the internet realm that should be helpful to taxpayers and further guidance on when web-based architecture development constitutes research and development activities. It also has added a gloss on to the rules for establishing the business use of internet services. And, lastly, it has taken a jab at a taxpayer-friendly revenue procedure. As a result, the IRS now will be able to disavow its own guidance and seek tax from taxpayers who relied on the IRS’s position.
* Mark Leeds is a tax partner in the New York office of Mayer Brown, an international law firm. His legal practice includes advising startup internet businesses and digital asset transactions. The views expressed herein are solely those of the author and should not be imputed to Mayer Brown.
 T C Memo 2022–62.
 See Section 195(b) of the Internal Revenue Code of 1986, as amended (the ‘Code’).
 Code s 162(a).
 Richmond Television Corp v United States, 345 F.2d 901, 907 (4th Cir 1965), vacated and remanded per curiam on other grounds, 382 US 68 (1965).
 Code s 174.
 2000–2 CB 601 Rev Proc 2000–50 has been modified by Rev Proc 2007–16, 2007–1 CB 358; Rev Proc 2019–43, 2019–48 IRB 1107; and Rev Proc 2022–14, 2022–7 IRB 502.
 Capitol Federal Savings & Loan Ass’n v Comm’r, 96 TC 204 (1991).
 The phrase in text means a change or innovation applied or added to an established or longstanding method, system, or organisation.