Retail Trade business applications jumped 44.5% in a single month, the largest move of any industry in the Census Bureau's July 2026 Business Formation Statistics, released August 12, 2026. That's a startling number in a report that tracks eighteen industry sectors every month, and it earns a caveat: a one-month spike isn't a trend until a second month confirms it. What doesn't need a caveat is the number sitting underneath it. The Information sector, the closest Census category to software and tech, made up just 2.3% of the 578,926 applications filed that month, in line with where it's sat most of this year. Whatever kind of business is forming right now, in retail or anywhere else, it's mostly being formed by people who don't build software for a living.
What Business Formation Statistics actually count
Business Formation Statistics (BFS) is a monthly release built by the Census Bureau in research collaboration with the Federal Reserve Board, the Atlanta Fed, the University of Maryland and the University of Notre Dame. It counts applications for an Employer Identification Number, filed on IRS Form SS-4, after excluding applications that clearly aren't an operating business (tax liens, estates, trusts, and a short list of industries with historically low conversion rates). From that base the Bureau builds four nested series. Business Applications (BA) is the full count: 578,926 in July, up 8.1% from June. High-Propensity Business Applications (HBA) is the subset whose characteristics predict a high rate of actually becoming an employer: a corporate legal structure, a stated hiring plan, a first wages-paid date, or a NAICS code on a specific list covering food service, parts of construction and manufacturing, two retail codes, two professional-services codes, one education code and two health care codes. That was 151,857 applications, about 26.2% of the total. Business Applications with Planned Wages and From Corporations narrow further still. This piece works from the broadest series, BA, which is why retail's 44.5% jump and software's 2.3% share are both counted the same way: as applications filed, not yet as businesses that have hired anyone. June's breakdown, covered here, used the same series.
The industries filing fastest, per the July numbers
| NAICS sector | July 2026 applications | Change vs. June |
|---|---|---|
| Retail Trade (44-45) | 141,241 | +44.5% |
| Professional Services (54) | 81,366 | -2.5% |
| Construction (23) | 48,972 | -0.5% |
| Other Services (81) | 44,741 | ~0% |
| Administrative and Support (56) | 37,537 | +0.7% |
| Transportation and Warehousing (48-49) | 34,982 | +0.3% |
| Health Care and Social Assistance (62) | 33,953 | +0.7% |
| Accommodation and Food Services (72) | 29,951 | +2.8% |
| Real Estate (53) | 26,431 | -1.2% |
| Finance and Insurance (52) | 20,287 | +2.2% |
| Information (51) | 13,223 | +0.5% |
Ten of the eleven biggest sectors here build nothing technical themselves. The eleventh, Information, covers far more than software: broadcasting, telecom, publishing and data processing all share that NAICS code, so even a generous read puts software's actual share below the reported 2.3%.
The 44.5% number, and what it doesn't tell you
A seasonally adjusted jump this size is unusual enough to flag and unusual enough not to over-read. Census's own methodology notes state plainly that statistical significance isn't measured for the industry breakdown, so a single month reading 44.5% higher than the last is data, not a diagnosis. The Bureau's next release, due September 11, 2026, will show whether July was the start of something or one loud month in a series that has had them before. What doesn't require waiting for confirmation is what didn't move: software's share of the total, sitting under 2.5% through the same month retail jumped by nearly half.
What this means for the founder who just filed
Filing for an EIN doesn't come with a developer. Whoever is behind July's 141,241 new retail applications, or the steadier flow from construction, professional services and health care, filed that form because they're opening a store, a practice, a shop or a firm, not because they're building software. Most of what they'll actually need from software is unglamorous: an inventory system that doesn't lose track of stock, a booking calendar that doesn't double-book, a point of sale that talks to the accounting software instead of fighting it. None of that gets built by a marketing agency, and none of it is a reason to hire a full engineering team for a business that, by this same report's own bar, has closer to a one-in-four chance of ever running payroll at all.
A worked example
Picture a founder who filed an EIN this month to formalize a specialty outdoor-gear retail brand she's been running out of a garage for a year. She has a storefront on a hosted platform, a spreadsheet tracking inventory across three sales channels that increasingly disagrees with itself, and a lease on a small warehouse starting in October. She isn't hiring a CTO for this. What she needs is a fixed-scope project: one inventory system that talks to her storefront, her wholesale channel and the new warehouse's shelving, built once and handed over working. That's a two-to-six-week build for someone who's done it before, not a six-month platform rewrite, and pricing it as the latter is how a retail founder ends up quoted for a problem she doesn't have.
What kind of help actually fits, and what doesn't
The honest exclusion first: if what a founder actually needs is a template storefront or an off-the-shelf CRM, a custom build is the wrong answer and a waste of a budget this size of business can't spare. Custom development earns its cost when the workflow is specific enough that no off-the-shelf tool fits cleanly: three sales channels that need one inventory number, a scheduling rule a generic tool doesn't know exists, a handoff between two systems that were never designed to talk to each other. That's the gap custom software development is built for, scoped to the business in front of it rather than sold as a platform. Every engagement gets priced individually against what it actually needs, not against a rate card built for a different kind of company. If the worry is losing control of what gets built once someone else is building it, here's what to keep in your own name regardless of who writes the code. And if software already runs more of your week than you'd like and nobody in-house can touch it, the signs are usually the same ones whatever industry you're in. Tell us what you're building and we'll say plainly which side of that line it's on, including if the answer is that you don't need us yet.
Sources
- U.S. Census Bureau: Business Formation Statistics, July 2026 (Release CB26-130, published August 12, 2026; accessed August 21, 2026)
- U.S. Census Bureau: Business Formation Statistics Monthly Data Release, July 2026 (Release CB26-TPS.46, published August 12, 2026)
- U.S. Census Bureau: Business Formation Statistics program page (methodology and release schedule, accessed August 21, 2026)