Why New Business Counts Look Low at First
On the morning of August 13, 2026, Florida's records showed 11 new business formations dated August 12. The Wednesdays before it, once settled, averaged 2,511. The number is not wrong and nothing is broken. Florida publishes formation records on a lag, so a filing day fills in over roughly two weeks after it ends. If you read a recent week the day it closes, you are looking at about a fifth of what happened.
Florida formation counts fill in for about two weeks after a filing day. Measured on August 13, 2026, a day one day old showed 2% of a normal day's volume, three days old showed 18%, and a week old showed 85%. Dissolutions and other corporate events do not behave this way.
Key findings
- A day one day old shows about 2% of its eventual total. August 12 held 11 filings when measured the next morning, against a settled Wednesday average of 2,511.
- The catch-up is fast but not instant. Three days out a day reaches roughly 18%, six to ten days out roughly 85% to 92%.
- The tail is real and small. Figures we published on August 1 for March through June 2026 each gained between 1 and 17 additional filings by August 13.
- This affects formations specifically. Corporate events such as dissolutions and reinstatements settle within about a day, so those counts can be read fresh.
What a recent week actually looks like
Below is Florida's daily formation count exactly as it stood on the morning of August 13, 2026, next to the settled average for that weekday taken from the three complete weeks of July 13 to July 31. The right column is what percentage of a normal day has arrived so far.
| Filing day | Days elapsed | Recorded so far | Settled July average | Share arrived |
|---|---|---|---|---|
| Monday, August 3 | 10 | 3,012 | 3,280 | 92% |
| Tuesday, August 4 | 9 | 2,232 | 2,507 | 89% |
| Wednesday, August 5 | 8 | 2,222 | 2,511 | 88% |
| Thursday, August 6 | 7 | 1,922 | 2,268 | 85% |
| Friday, August 7 | 6 | 1,829 | 2,114 | 87% |
| Monday, August 10 | 3 | 601 | 3,280 | 18% |
| Tuesday, August 11 | 2 | 232 | 2,507 | 9% |
| Wednesday, August 12 | 1 | 49 | 2,511 | 2% |
| Thursday, August 13 | 0 | 11 | 2,268 | 0.5% |
Nothing about Monday August 10 was a quiet day for Florida business. It will land near 3,300 like every other Monday. On the morning of August 13 it held 601 filings because the rest had not been published yet.
Why does this happen?
The date attached to a Florida formation record is the entity's own filing date, a field inside the record. It is not the date of the batch file the record arrived in. Those two dates are usually close but they are not the same, and the gap is what produces the effect.
The state publishes daily files of new and updated records. A file published on a Thursday carries filings made that day, and also filings made on Monday, the previous Friday, and occasionally weeks earlier. Every one of those records is stamped with its own filing date, so it lands retroactively in a day that has already passed. Ask the database how many businesses formed on August 10, and the answer changes every day for about two weeks, and then stops changing.
This is not a defect in Florida's process or in the pipeline that ingests it. Every record eventually arrives, and the final counts are accurate. It is a property of publishing by filing date rather than by publication date, and it is shared by most administrative filing data.
How long does a day take to settle?
Most of the fill-in happens inside a week. Reading down the table above, a day picks up the bulk of its volume between day three and day seven, and then closes the last tenth slowly.
The slow tail is measurable, and you can check it against our own earlier reporting. On August 1, 2026 we published monthly formation totals for the first half of the year. Those months were already between one and five months old. Twelve days later, all four had grown.
| Month | As of August 1 | As of August 13 | Added |
|---|---|---|---|
| March 2026 | 71,283 | 71,284 | +1 |
| April 2026 | 68,561 | 68,566 | +5 |
| May 2026 | 64,579 | 64,596 | +17 |
| June 2026 | 62,263 | 62,277 | +14 |
Fourteen extra filings on a base of 62,263 is a change of 0.02%. That is the size of the tail once a month is properly aged, and it is why we treat anything older than two weeks as final for reporting purposes while still expecting a handful of late arrivals. The published figures in our first-half 2026 report have held to within a rounding error.
The mistake this causes
The common error is comparing a fresh window against a settled one. Put "the last 7 days" beside "the 7 days before that" and the recent week loses, every time, by a margin that has nothing to do with business activity. Do it across a month boundary and you can report a collapse in a month that turned out to be up.
The same trap catches anyone checking a specific company. A business that filed on Monday may not appear in a search on Wednesday. It exists, it is registered, and its record is coming.
Three habits avoid all of it:
- End your window at least 14 days back for any count you plan to publish or act on.
- Compare windows of equal age, not equal length. Last July against the July before it works. This week against last week does not.
- Treat the most recent fortnight as provisional rather than absent. The filings are real, they are simply not all visible yet.
What is not affected
Corporate events are a different feed and they behave differently. Dissolutions, reinstatements, name changes, and registered agent changes reach the published record within about a day of being filed, and a week-old event count is close to final. Counts of Florida businesses closing can be read fresh, which is why our dissolution reporting uses shorter windows than our formation reporting.
Federal tax lien filings have a different quirk again. Those records carry no meaningful lag once a file lands, but the files themselves publish irregularly rather than daily. A rolling seven-day lien count mostly measures whether a batch happened to land that week, so lien activity is better read over a month or more.
Totals that are not filtered by date are unaffected entirely. The count of active Florida businesses, or of entities in a given county, does not depend on any recent window.
Why we publish this
Anyone comparing our formation figures against Florida's own site on a different day, or against a competitor's dashboard, will find discrepancies in the most recent weeks. Those gaps are the effect described here rather than a disagreement about the facts. Knowing the shape of it means you can date-stamp a number and reproduce it, which is the point of publishing figures at all.
About these numbers
Daily and monthly counts come from Florida Division of Corporations filings as republished in the sneyk dataset. The August figures in the first table were read on the morning of August 13, 2026, from data ingested at 04:08 that day, and they will be higher by the time you check them. That is the point of the table rather than a caveat against it.
The settled July averages use the three complete weeks from July 13 to July 31, 2026, by weekday. The week of July 6 is excluded because Florida's July 4 closure pushed filings into the following days and distorts that week in both directions.
Counts are distinct corporations rather than filing rows. Filings carrying a January 1 date are excluded, since that date serves as a year-known, day-unknown placeholder rather than a real filing day. Filings dated the first of a month run higher than a normal day for a similar reason, which is visible in the August 1 figure and is why month-boundary weeks are not used as a baseline here.
Current formation activity by county is at sneyk.com/counties.
Image credits
- Three sand timers in a metal frame: Photo: wwarby via Flickr. CC BY 2.0.
- Woman writing figures in a notebook beside a laptop: Photo: Shixart1985 via Wikimedia. CC BY 2.0.
- Stack of notebooks and papers on a desk: Photo: rawpixel. CC0.
- Stack of envelopes and papers on a chair: Photo: Keith Williamson via Flickr. CC BY 2.0.
- Person typing on a laptop: Photo: rawpixel. CC0.