Three headcount tests decide whether a planned layoff or plant closing in California falls under the state's Worker Adjustment and Retraining Notification (WARN) rules: the size of the employer, the size of the site, and how many people are separated inside a 30-day period. Get those counts right before anyone plans a schedule or a notice.
If you recruit or place people for a company that might cut jobs, close a plant, or move a site in California, the first question is rarely about the layoff itself. It is about counting. Someone has to know whether the employer is large enough, whether the site is large enough, and how many people are affected in a short window before the rules can be applied at all.
Every number, rule, and time window here comes from the California Employment Development Department's page on WARN, Worker Adjustment and Retraining Notification (WARN), and the quoted lines are linked to that page. Where that page does not answer a question, this post says so and leaves the answer open instead of filling the gap with a guess.
The scope is narrow on purpose. This is about headcount: who is counted, over what period, and against which line. It does not cover what a notice must say, who must receive one, or what follows if an employer gets the count wrong, because the material used here does not address those points.
Does the employer need 100 full-time employees to count?
The first test looks at the whole company. The EDD page sets its line at 100 full-time employees and states the rule this way: "Applicable only to employers with 100 or more full-time employees who must have been employed for at least 6 months of the 12 months preceding the date of required notice in order to be counted." ( EDD WARN page)
| Condition | What the quoted line states |
|---|---|
| Employer size | 100 or more full-time employees |
| Worker type | Full-time employees only |
| Tenure | At least 6 months of the 12 months preceding the notice date |
Two conditions sit inside that sentence, and both apply person by person. The first is the type of worker. Only full-time employees are counted, so part-time staff sit outside this test. The second is tenure. Each person must have been employed for at least six months of the twelve before the notice date. An employee who started four months before that date does not count, even though that person is on the payroll and works the same shifts as everyone else.
That makes the first test a look-back, not a snapshot. A headcount pulled on the day a cut is announced can give the wrong answer if several people joined recently. Start from the notice date, step back twelve months, and ask of each full-time employee how many of those months they were employed. A roster with hire dates turns this into a sorting job instead of a memory test. Sort by start date, mark who reaches six months inside the window, and total them.
Picture an employer whose full-time headcount is just over 100 on the day a cut is planned, with several full-time employees who started only a few months earlier. Those recent starters fall outside the count, which can bring the qualifying total to exactly 100 or just below it. The employer sits at the line, not comfortably above it. A result that tight turns on individual start dates, so one wrong date can change the answer.
Now consider the full-time filter. Suppose an employer's total headcount looks large because it includes a big part-time group. If a meaningful share of the full-time group started within the last six months, the qualifying count can land well under 100. The part-time staff do not change the result here, even though they make the total look much bigger. A team that counts everyone on the roster gets a larger number and a different answer, which is why the wording matters.
Three cautions keep this test honest. The quoted sentence does not say how to handle a partial month, so use exact start dates rather than rounding them to the nearest month. It says nothing about employees on leave, so those cases need a decision from the employer rather than from the staffing team. And the sentence refers to the employer's own employees. If your team places workers at a client's site, the page does not say whether those workers belong to the client's count, the placing firm's count, or both. Do not assume either until someone with authority has settled it.
Does a single site need 75 employees to be covered?
The second test applies to a "covered establishment" that "employs or has employed in the preceding 12 months, 75 or more full and part-time employees." ( EDD WARN page) This test moves the count from the company to one location. Unlike the first test, it includes part-time staff and looks back over the year before the event instead of applying the six-month tenure condition. Fix the window's two dates before counting, so every person is judged against the same span.
That difference matters when you build the list. The site sentence has no six-month condition, so this post does not apply one here. On the plainest reading of the words, anyone who worked at the site at any point in that look-back belongs in the count, including someone who left last spring. A site with most current staff still on the payroll and a group who left during the past year could easily total above 75 on that reading. The quote does not explain how to count people who left part way through the window, so treat this as one reading of the words rather than a settled answer.
Consider a site whose count sits one short of the line. If a rehired person appears twice because of a duplicate row, the site would look like it reaches the line when the true figure is still one short. Duplicate rows are a common reason a roster looks bigger than the workforce really is, so match records by employee ID before adding anything up. A person who moved between sites needs a clear date for each site they worked at, because the quoted line does not say how to count someone who appears at two sites.
The sentence also uses "covered establishment" without defining it on that line. Where a business works out of several buildings, or where some staff work on-site and others remotely, the way you draw the boundary can change the number considerably. A team that treats a campus as one site and a warehouse across the road as another may reach a different answer than a team that does the reverse, and the quoted line does not settle which approach fits.
For the roster, tag each person with a site code, a full-time or part-time status, a start date, and an end date if they have left. With those fields the 75 count becomes a filter and a total. Without them, the team is searching old emails and badge logs to reconstruct who worked where, which is slow and easy to get wrong.
How many layoffs inside 30 days trigger the third test?
The EDD page sets layoff triggers in several lines, with a separate line for 500 or more. Two of them read differently on whether a share of the workforce matters. The third test counts people affected inside a window, so it helps to see the wording side by side. The page states these lines, quoted as written:
| Quoted line | Share requirement |
|---|---|
| Plant closings involving 50 or more employees during a 30-day period | Not stated |
| Layoffs of 50 to 499 full-time employees at a single site | At least 33% of the full-time workforce |
| Layoffs of 500 or more | Covered regardless of percentage of workforce |
| Layoff of 50 or more employees within a 30-day period | No share requirement |
- "Plant closings involving 50 or more employees during a 30-day period."
- "Layoffs within a 30-day period involving 50 to 499 full-time employees constituting at least 33% of the full-time workforce at a single site of employment."
- "Layoffs of 500 or more are covered regardless of percentage of workforce."
- "Layoff of 50 or more employees within a 30-day period regardless of % of workforce."
Start with the window. The page says "30-day period," not calendar month. Separate reductions add together when they fall within any 30 days, even if the dates sit in two different months. Suppose one group leaves in early March and a second leaves at the start of April, with both dates inside a single 30-day span. They combine for this test. A planning team that counts March and April separately would see two smaller figures and miss the combined one. A practical habit helps here: list every planned separation with its date in one column, then slide a 30-day span down the list and note which dates fall inside it.
Next, the measure. The 50-to-499 line uses a share: at least 33 percent of the full-time staff at one location. The fourth quoted line sets no share at all. The two lines do not match, and this post will not pick between them. Because the answer can change a notice decision, it is a question for the employer to settle before anything is final.
Consider a site where the full-time workforce is large enough that a planned separation of a bit more than 50 employees is well under a third of it. Read alone, the 50-to-499 line would not reach this site, because the planned group falls short of the 33 percent share. The line for 50 or more with no share requirement does reach it, because the planned group is more than 50. The same numbers give different answers depending on which line governs. That is why the governing line needs settling before the plan is final, not after the separations happen.
The page is clear on one point, though. Layoffs of 500 or more fall under a line of their own, so a share calculation adds nothing at that size. A team that reaches 500 separations in a window can treat the layoff as covered without working out the site's share.
The plant closing line needs the same care. It counts "employees" and does not say whether part-time staff are included. Take a hypothetical closing where the full-time group is just under 50 and a smaller part-time group brings the two together to just over 50. If only full-time staff count, the total falls under the line. If part-time staff count, it clears. The page does not settle which reading applies, so this count cannot be finished from the page alone.
Which test should you run first when a cut is planned?
Start with the 30-day count, then run the employer and site counts, and compare all three before anyone decides the rules do not apply. This order is a suggestion, not something the page sets out. The planned cut tells you which people are in play, and the 30-day count often changes fastest when a plan moves. The quoted lines do not say how the three tests combine, so this post does not merge them into one answer.
The practical reason is that the 30-day count runs on the people being separated and their dates, which the planning team usually already has. The employer and site counts depend on the full roster, which is often the slower job. A smaller set gives a first answer quickly and shows whether the roster work is worth finishing right away.
Here are the thresholds from the EDD page in one place, as the page states them.
| Test | Threshold on the page | What the quoted wording counts |
|---|---|---|
| Employer size | 100 or more full-time employees | Each must have been employed at least 6 of the 12 months before the notice date |
| Covered establishment | 75 or more full and part-time employees | Employs or has employed in the preceding 12 months (one reading: anyone who worked at the site in that year) |
| Plant closing | 50 or more employees | Within a 30-day period |
| Layoff band at one site | 50 to 499 full-time employees | At least 33% of full-time staff at one location, within a 30-day period |
| Large layoff | 500 or more | Covered at any share |
| Layoff of 50 or more | 50 or more employees | Within a 30-day period, with no share requirement |
The table shows only what the quotes state. It does not show how the rows interact or what a notice must contain.
Here is how the rows can work together in one hypothetical case. A company has a full-time workforce above 100 plus a part-time group, spread across two sites. Its plan separates a group of full-time employees from site A within 30 days, just above 50, and site A's full-time workforce is small enough that the planned group clears 33% of it. If the company's full-time count clears 100 after the six-month condition, and site A's full and part-time count over the look-back year clears 75, then every test points the same way. Under these lines, the choice between the two layoff lines does not change the answer.
Change one number, though. If site A's full-time workforce grows until the planned group falls below 33% of it, the share line no longer reaches the site, while the line for 50 or more with no share requirement still does. The answer then depends on which of the two lines governs. For a staffing team, the practical move is to sort the rows into two piles: those the roster answers cleanly, and those that depend on an open reading. Only the second pile needs to go forward.
What should the headcount file contain?
The file should be one roster with a row for each person and five fields that cover every test: full-time or part-time status, start date, end date if they have left, site, and whether the person is in the planned cut. Those fields answer all three tests, so you need no separate list for each.
| Field | Employer test | Site test | 30-day test |
|---|---|---|---|
| Full-time or part-time status | Filter to full-time only | Keep both groups | Filter to full-time employees |
| Start date | Compare with the notice date for six months of service | Confirms presence in the window | Not used |
| End date | Not used | Counts people who left within the look-back year | Gives the separation date |
| Site | Not used | Filter to one site code | Group planned separations by site |
| In planned cut | Not used | Not used | Filter to the planned group |
A row might read as a single line of plain text: an employee ID, full-time, a start date, a blank end date because the person is still employed, site B, and yes in the planned cut. Someone who left in the spring keeps an end date and stays in the file, because the site test looks back over the year and needs that person.
Keep dates in one format, and keep a dated copy of the file before the plan changes. Write the notice date at the top, because the first test's look-back depends on it. Note which version of the EDD page you used and which lines you relied on, so someone else can follow the count without redoing it. That record also shows later what was counted and on what basis.
Share only the fields the decision-maker asks for. Pay, addresses, and other personal details are not needed to run the three counts, and every extra copy is another place that information can leak. A summary sheet with the totals, the notice date, and the sources is usually easier to pass along than the full roster.
What do you do when a count lands near a line?
Work the count again from the dated rows, not from memory, and do it before anything goes to the decision-maker. A count near a line is the case that most needs this, because one uncertain date can move the total across the threshold.
Timing is the other trigger. The six-of-twelve test is anchored to the notice date, so if the planned date moves, the count can change even with no hiring or leaving. Suppose the planned date moves two weeks later. Someone who started just under six months before the original date will have passed six months by the new one, so that person now counts where they did not before. Tie each count to its notice date, and when the date moves, rerun the count from scratch instead of adjusting the old total.
When a result lands exactly on the line, as in the case of exactly 100 full-time employees above, the total is seldom the real question. The open question is usually one start date nobody has verified. Pin that date down first, then recalculate, and say which source the date came from. A total with a clear source is far easier for the decision-maker to accept than one that rests on someone's recollection.
If the uncertain date cannot be confirmed in time, present the result as open and show both totals: one with the date counted and one without it. That gives the decision-maker the range instead of a single figure that hides the uncertainty.
Where does this leave a small staffing team?
Your job is a verified count and a short list of open questions. The decision belongs to the employer and its counsel, and this post is no stand-in for that advice. The EDD page supplies the thresholds used here, and it is the only source this post relies on for them.
In practice, a staffing team can start by pulling the roster with status, hire, end, and site fields, running the 30-day count from the planned list, and then working through the other two counts. That order gives a first answer on the same day the plan is drafted.
The questions to carry forward are these:
- How to count a partial month for the six-of-twelve condition on the first test.
- Whether employees on leave count toward the first test.
- Whether workers placed by a staffing firm count toward the employer total or the site total.
- Whether part-time staff count toward the plant closing line.
- Which of the two lines covering layoffs of 50 or more governs a given cut.
Send the three counts, the notice date, the open-question list, and a summary sheet with the sources to the employer; send the full roster only if they ask for it and only with the fields they need. Keep the list open until each question has an answer from someone with authority to give one. A correct count that is still open on these five points leaves the team with a decision it cannot make on its own.
Related guides
- Interview Scorecard Example for Consistent Hiring Teams
- Hiring Decision Debrief: Notes to a Recorded Decision
- Conviction timing rules for 3 jurisdictions