Most cross-border NRI businesses fail for the same reason most domestic startups fail — no customers want what is being sold. But of the few that do clear that first hurdle, a surprising number stall on the structural questions that came up in the first eighteen months and never got cleanly answered. Five frictions show up again and again in the founder conversations the NRI Globe business desk has tracked since 2023, and the order in which a founder solves them quietly determines how much of the next five years gets eaten by paperwork.
Friction one: the entity choice that decides everything else
The first real question is not "LLC or C-corp" — it is "where does revenue land and where does payroll go out." A U.S.-resident NRI selling software-as-a-service to U.S. customers, with an Indian engineering team, has a meaningfully different structure than a U.K.-resident NRI selling services to Indian customers with a U.K.-based sales lead.
The dominant pattern that emerges for the first cohort — U.S. residence, U.S. customers, Indian team — is a Delaware C-corporation that owns an Indian subsidiary, with the subsidiary on a transfer-pricing arrangement that compensates the India entity on a cost-plus basis. This setup pushes the bulk of profit into a U.S. corporate-tax environment that founders are already familiar with and keeps the India side clean for FDI compliance.
The pattern flips for the second cohort. If revenue is in INR from Indian customers, an Indian private limited company with an overseas founder shareholder is usually correct, with the founder visa status determining whether dividend repatriation is straightforward or paperwork-heavy. Mixing the patterns — say, a Delaware C-corp invoicing Indian customers in INR — tends to surface tax-residency problems three years in, when the company finally tries to raise institutional capital.
Friction two: payroll and contractor classification
The second friction shows up around month nine, when the founder has six engineers in Bengaluru, three contractors in different U.S. states, and a part-time designer in the U.K. Each jurisdiction has its own rule for what counts as employee versus contractor, what counts as residency for tax purposes, and what counts as a foreign payroll exposure.
The discipline that works is to pick a single Employer of Record vendor for each major jurisdiction early — Deel, Remote, Multiplier, Velocity Global, and a handful of regional players occupy this space — and route everyone through them until the headcount in a jurisdiction crosses a threshold where setting up a direct entity is justified. The threshold is usually five to ten employees per jurisdiction. Below that, an EOR costs more per head but absorbs the compliance overhead.
The variant that consistently breaks is paying U.S.-based engineers as contractors on 1099 from the India entity, then using their stock options for retention. Misclassification under California's AB-5 or analogous rules turns into a back-payroll-tax problem that can wipe a year of runway.
Friction three: banking and the corridor problem
Most NRI founders open the first corporate bank account at whichever U.S. or U.K. neobank approves them fastest — Mercury, Brex, Wise, Revolut Business — and then discover, eighteen months in, that none of them accept inbound INR with a usable narration field. The Indian subsidiary's outbound transfers to the parent for reimbursable expenses arrive with cryptic descriptions that take weeks to reconcile.
The fix is unglamorous: open a second account with a traditional bank that has both a U.S. branch and an Indian correspondent relationship — HSBC, Citi, Standard Chartered are the usual three — and run the cross-border corridor through it, even at higher fees. The neobank stays for domestic operating expenses. The traditional bank handles the corridor. Trying to compress this into a single bank account is the founder equivalent of trying to use one credit card for both personal and business expenses; it works for a year and then doesn't.
Friction four: profit repatriation and the dividend question
The fourth friction lands at the moment the business becomes meaningfully profitable. An Indian subsidiary that earns retained earnings on the cost-plus model can technically declare a dividend back to the U.S. parent, but the process — Form 15CB from a chartered accountant, withholding tax of 20% under the India-U.S. treaty, board resolutions, RBI filings — turns a routine corporate action into a multi-week project. Many founders just leave the cash in the India entity and use it for hiring or marketing instead.
That works for a while. Then the U.S. parent raises a Series A and the investor's due diligence asks why $1.8 million is sitting offshore in a subsidiary. The answer "because dividending it home is a pain" is not an answer that closes a round. The discipline is to plan the dividend cadence — annual or biennial — from year two, file the paperwork even when the amounts are small, and treat the corridor as something to be exercised before it has to be relied on.
Friction five: the founder visa question
The last friction, and the most personal: what visa status is the founder operating under, and does that status survive the business succeeding? An H-1B founder building a U.S. startup is constrained in ways a green-card founder is not — board control, executive title, employer-of-record arrangements all interact with non-immigrant work-authorization rules. An L-1A intracompany transferee path is real but requires a year of qualifying employment in the India entity before the application.
The O-1 visa for extraordinary ability is the fastest path for a founder who can document a track record — funding raised, press coverage, advisory roles — but it is not a path one can plan around in year one without those artefacts. Most founders end up navigating a two-to-three-year sequence: ship a product, raise a seed round, generate enough public footprint to support an O-1, then switch off the original employment-based visa.
The mistake is to leave this question for "later." Later usually means the founder is running a company that requires their unrestricted presence in the U.S. exactly when their original H-1B sponsor relationship is becoming awkward to maintain.
Order of operations
The order that tends to work is the inverse of the order most founders attempt. Solve the entity choice first — before incorporation, ideally. Solve the visa question second, before hiring. Set up the EOR relationships third. Open the corridor banking relationship fourth, before the first real revenue lands. Plan the repatriation cadence fifth, no later than year two.
The other order — incorporate, hire, ship, then panic about residency and repatriation — is how the next eighteen months get eaten. The friction is real, but it is finite. Spending a quarter on it explicitly at the start saves the same quarter, twice over, three years in.
