The January 2026 temporary-help count was still well under the 2022 peak, so hiring seasonal staff this year means a tight market, not a leftover surplus. In this read, October 2026 unemployment was 4.1 percent.
Are you staffing a calendar spike, or is the temp market itself getting tighter?
A big November payroll jump can still be the holiday calendar rather than fresh demand, so read it against the adjusted series too. Use those figures to time recruiting. Also judge whether temp help is actually expanding past that calendar bump. They are not a picker count for one site.
The split that matters is already in the monthly jobs report, and confusing those two reads is how a normal rush gets mistaken for a looser labor market. One series keeps the seasonal bump. The other takes it out.
What Seasonal Hiring Numbers Are Actually Counting
The Bureau of Labor Statistics already runs that split every month, and the Current Employment Statistics program surveys business establishments, not households. Household data feed the unemployment rate.
TEMPHELPS is the temporary help services series from those establishment counts. It is published seasonally adjusted, in thousands of persons, every month. The CES code is CES6056132001.
Total nonfarm payroll uses the same survey under two labels, and you'll want both of them before you lock a 2026 seasonal hire. PAYEMS is the adjusted count, code CES0000000001.
PAYNSA drops the seasonal filter. Weather and holidays move employment the way the BLS already expects. School calendars do the same.
Seasonal adjustment strips those repeating patterns so the adjusted series shows what changed besides the calendar, not merely the holiday week on the schedule. Turns out, you'll want both reads.
The unadjusted count sits closer to November floor coverage. The adjusted count shows whether temp help is still expanding once that bump is gone. If you've ever stared at a November payroll jump and decided the labor market just exploded, you've probably mixed those two series, since both get quoted as jobs. Mixing them is an easy miss.
BLS releases that report at 8:30 a.m. EST, and the usual slot people watch is the first Friday. The timing gap is worth watching.
The precise slot is the third Friday after the week that included the 12th of the prior month. Employment overall tends to confirm a turn after it has already happened, which is why the American Staffing Association describes it as a lagging indicator. ASA tracks that temp series instead.
Those series won't hand you a headcount target, because they are national figures meant as context, not a roster. They won't tell a warehouse in Ohio how many pickers to book for Thanksgiving week.
Platforms sit downstream of this signal, not inside it. Wonolo's site describes seasonal and on-demand fill, including temp-to-perm work, which is a sourcing workflow rather than a BLS benchmark. Check coverage and pricing on the official page. Do that before you plan around it.
These are the reads to check.
| Signal | Source | Seasonal hiring read |
|---|---|---|
| TEMPHELPS | BLS CES CES6056132001, seasonally adjusted, monthly | Temp-market trend after calendar effects come out |
| PAYEMS | BLS CES CES0000000001, seasonally adjusted | Broader hiring backdrop |
| PAYNSA | BLS CES, not seasonally adjusted | Calendar swing in total payroll |
| Release | Employment Situation, 8:30 a.m. EST | Check both series before seasonal orders |
Pull TEMPHELPS and PAYNSA from the latest Employment Situation tables before you issue seasonal requisitions. If the adjusted temp series is flat while the unadjusted count jumps, you're staffing the calendar. You're not looking at a new trend.
Read PAYNSA when you want the calendar swing in total payroll, and read TEMPHELPS for the trend once those calendar effects come out. Then you know if the rush is only the rush.
The gap between a cleaner report and a thinner bench
A clean split between spike and trend is not the same thing as a deep bench you can hire from.
Textbook notes treat a monthly gain as a turn. It isn't, not for seasonal hiring. January 2026 temporary help rose by 9,100 jobs, to about 2.48 million seasonally adjusted, and that total still sat far under the March 2022 peak of nearly 3.2 million. You don't have a full bench.
SIA's economist said much of the late-2025 weakness reflected outdated seasonal factors, and he called the prior evidence of deterioration "a statistical mirage." Temp employment may still not have hit its trough. Penetration in January held at 1.56%.
By October 2026 the broader market was still tight, with unemployment at 4.1% after a 165,000 payroll gain that month. Seasonal employers are not fishing in a slack pool.
Gains showed up in health care, professional and business services, and transportation and warehousing.
The failure mode on the floor is mismatch, not a missing applicant count, and a warehouse or retail post written like a career job draws people who want a path and benefits, people who vanish after a weekend or sometimes never start, and the inbox looked full, the inbox looked full, and the shift still didn't. Mismatch is what breaks the floor.
Job seekers get the slower version of that mess. General job boards often bury the end date, the pay, and the actual nights inside the posting. You'll apply, sit on a status that barely moves, and then find out the role dies before the holiday rush even starts. Flexible meant four closes in a row. The posting died, and the rush had not even started.
Staffing portals can move a volume order faster, then go quiet when the client cuts headcount. A gig app can cover one night. It won't hand you a crew that already knows the dock.
The old surplus assumption is the part that aged out.
A late requisition and a bigger sponsored post don't replace a local bench that is still far under the 2022 temp peak. The 2022 gap is still the limit.
| Channel | Best for | Where it breaks | Verify first |
|---|---|---|---|
| Job boards | Broad reach for hourly retail and warehouse seasons | Vague posts and applicants who wanted a permanent role | End date, pay, schedule, and a same-week screen |
| Staffing firms | Spike volume your own team can't screen | A thin local bench and orders that get cut midweek | This season's fill time for your city and shift |
| Gig or shift apps | Same-day or one-night coverage | No continuity, thin screening on regulated sites | Whether the site allows app labor and who holds the record |
| ATS plus a rehire list | People who already worked your peak | Reviewer overload when the spike hits | How fresh the rehire list is, and a hard cutoff for new apps |
No channel repairs a role that starts in ten days with pay still blank. Open the live posting on the platform and confirm the shift facts yourself. Then staff to last year's no-show rate for that same week, not to the number on the requisition.
Count back from peak week, then hire in that order
Confirming the shift facts still leaves the calendar. Late posts lose the people you want.
BLS put October unemployment at 4.1 percent while payrolls added 165,000 jobs, and compensation costs rose 3.4 percent over the year ending September 2026. Keep the 3.4 percent in view. The live date is not a draft date.
Thing is, count twelve weeks back from your first peak week, because that date is when the role should go live, not when you start drafting it in a meeting. Put that date on a calendar.
- At twelve weeks, write the end date, weekly hours, shifts, and pay, then post on the job boards and staffing platforms you already use.
- At eight weeks, screen availability before experience. Short stints are normal. Don't let an ATS dump them.
- At six weeks, interview in small batches and send offers the same week. Don't wait on the offer.
- At four weeks, finish paperwork and run one supervised shift.
- At two weeks, publish the schedule and hold a backup bench. September quits rose, so some accepted offers will disappear.
None of that helps if the ad describes a different job than the one on the floor, with different hours or a different end date.
Where this calendar breaks
Call last year's crew before the public post goes up. A returning worker who knows your floor beats a polished resume, and gap filters will bury that person if you wait for the open listing to do the sorting. Make that call before you post.
One board in one city is a thin net, and so is a pay line that says competitive while candidates stack three offers before dinner, which is not even a dramatic example, it's just how the evening goes, and slow replies lose the good ones, they take the shift that answers. Do not sit on the reply.
A returning worker already knows your floor, which is why the text goes out before the public post, and one board in one city will not replace that person. Pull week-of help from a staffing partner. The early core is the part you can actually train.
The twelve-week plan also falls apart when your peak hangs on weather or a single weekend you can't really forecast. Don't lock August fantasy headcount numbers. Hire a smaller trained core early and add week-of help from a staffing partner. A calendar is only a plan. It's not a headcount guarantee.
Questions people actually search
When should I start hiring seasonal staff for the holidays? A late-November peak means the roles should open in early September, so you can repost if the first wave is thin. October listings hit the same warehouses and stores at once.
What should a seasonal job posting include? Put the end date, hours, shift times, and pay in the first lines. Say whether overtime is likely. Vague posts lose people in week one.
Is a staffing agency better than job boards for seasonal hires? Boards work when you can train quickly and want a rehire list you own, and a staffing firm fits when you need people in days. Plenty of employers run both channels.
How do I get last year's seasonal workers to return? Text them dates and pay before you post publicly, and to be honest that head start beats spending more on ads.
Name your peak week and count back twelve weeks. Publish the posting with an end date and a real pay figure. Settle those facts before you compare job boards and hiring platforms.