# The Search Launch Calendar, One Role's Hires Curve to a Dated Start Week

*Build your industry's 12-month NSA hires curve, pick a sourced application-to-offer lag, and leave with a dated launch week and a monthly send quota for your tracker.*

- Canonical URL: https://www.refolk.ai/candidates/guides/job-search-launch-week-hiring-calendar
- Pillar: Reading the market
- Format: Teardown
- Published: 2026-11-08
- Last reviewed: 2026-11-08
- Reading time: 24 min
- Keywords: best time of year to look for a job, best month to apply for jobs, when do companies hire the most, is december a bad time to job search, how long does it take to get hired after applying, when should I start my job search

## Key takeaways

- In five months of total nonfarm hires, the seasonally adjusted series ranks March highest (5,535 thousand) and the not-seasonally-adjusted series ranks it lowest (5,220 thousand), so you can only read seasons from a FRED series whose ID starts with JTU.
- In Ashby's data, the median time from application to hire is about 30 days for business roles and 40 days for technical roles. US federal hiring averages 101 days, so the same hiring peak gives very different launch weeks.
- A launch week is the target hiring month, minus any notice period, minus the application-to-offer lag, minus about one week as a buffer for the 7.2-day median from application to first screen.
- December is not wasted: the not-seasonally-adjusted hires rate rose from 2.4 in December to 3.4 in January, and an application sent in early December with a 30 to 40 day lag lands as a January offer.
- January job searches ran up to 31% above the early-December average, but postings never rose above their early-December level, so people who start in the first week of January are competing for December's leftover openings.
- Refolk's index holds 944 US and 109 UK profiles titled campus, university or early careers recruiter. These programs run on fixed calendars, with offer deadlines in late October and mid-February, so they decide before the general market moves.

This guide is for job seekers deciding which week to start sending applications and how hard to push in each month after that. It follows one role, a mid-level technical recruiter in the US, through not-seasonally-adjusted federal hires data, a free job postings index, and a published application-to-offer lag, wrong turns included. It ends with a dated launch week and a monthly send quota. Run your own occupation and industry alongside it and you will finish with the same three things: a 12-month hiring curve, a lag in days, and a quota you can paste into your tracker.

## Why do answers to "the best month to apply for jobs" contradict each other?

They contradict each other because they measure different things: how many people are searching, how many jobs are posted, how many people are hired, and how one tool's users did. Each of these peaks at a different time. That is why January, May to June, and September to October can each be right about one measure and still wrong for you.

#### What each "best month" answer is really measuring

1. **Searcher volume** - How many people are looking; peaks in the first weeks of January
2. **Posting volume** - How many openings are advertised; January postings peak at month end
3. **Hires** - People added to payroll; the measure that shows when roles are actually filled
4. **One tool's interview rates** - What one product's users saw; usually no named dataset behind it

*Each of these four things moves at a different time, so a page that measures one of them cannot tell you about another.*

Only one of these measures answers "when do companies hire the most": hires. The source is JOLTS, the Job Openings and Labor Turnover Survey, a monthly Bureau of Labor Statistics (BLS) survey of employers. In JOLTS, a hire is any addition to the payroll during the month. Being one of those hires is the goal.

The trouble is that most pages that quote JOLTS use the headline number, and that number is seasonally adjusted. Seasonal adjustment is a statistical correction that strips out the predictable calendar pattern so analysts can see the trend underneath. It is the right tool for asking "is hiring getting better?" and exactly the wrong one for asking "which month is best?", because it removes the very pattern you are looking for.

Before you start, set aside two other kinds of answer:

- **Answers based on search volume.** They tell you when the crowd of job seekers is biggest, not when roles get filled. They are useful later, as a measure of crowding.
- **Pages with no named dataset.** Some ranking pages claim January brings 200 to 300% more applications per posting, or quote "23% higher response rates", with no method shown. I leave out any figure that does not name its dataset, and you should too.

## The worked example, and the wrong turn of reading adjusted data

My first pull for this example used the seasonally adjusted hires series, which ranked March as the best of five spring and summer months. The not-seasonally-adjusted series ranks March the worst. You can only read seasons from the not-seasonally-adjusted series, whose IDs on FRED (the St. Louis Fed's free economic data site) start with JTU.

The worked example is a mid-level technical recruiter in the US, looking for a new in-house or agency job.

**28,256 - US profiles titled Technical, Healthcare or Nurse Recruiter in Refolk's index**

This is the occupation in the worked example. In a 25-profile sample, 17 were Technical Recruiter and 3 were Healthcare Recruiter, but a sample that small cannot tell you the split across all 28,256.

### Fork 1: which industry does the role sit in?

JOLTS publishes hires by industry. BLS Table 9 shows hires levels and rates by industry and region, not seasonally adjusted. For a recruiter, the industry that counts is the employer's, not the recruiting function. An in-house technical recruiter at a software company is counted in the software employer's supersector (the broad industry group JOLTS reports). An agency recruiter is counted in professional and business services, where staffing firms sit. The example seeker is open to both but has spent their career at agencies, so I mapped the role to professional and business services.

To find the series ID, search FRED for "hires" plus your supersector name. Open the series notes and confirm they say "Not Seasonally Adjusted". The ID should start with JTU and end in HIL (hires level). If it starts with JTS, you have the adjusted version.

### Fork 2: the wrong turn

My first pull was JTSHIL: total nonfarm hires, seasonally adjusted. It looked sensible and was no use for timing. Here are the same five months side by side. I'm using total nonfarm because that is the series where I have verified monthly figures to show you.

| Month | NSA hires (k) | SA hires (k) | NSA/SA |
|---|---|---|---|
| Mar | 5,220 | 5,535 | 0.94 |
| Apr | 5,597 | 5,215 | 1.07 |
| May | 5,967 | 5,252 | 1.14 |
| Jun | 6,139 | 5,332 | 1.15 |
| Jul | 5,513 | 5,054 | 1.09 |

The adjusted series puts March first and July last. The raw series puts June first and March last. I calculated the ratio column myself. It only approximates the seasonal factor, because BLS mixes additive and multiplicative models, but the direction is clear. In March, the adjustment adds hires that did not happen. In May and June, it removes hires that did.

> **Rule:** Check the third letter of the series ID
>
> Only read seasons from a FRED hires series whose ID starts with JTU and whose notes say Not Seasonally Adjusted. If the ID starts with JTS, the seasons have been removed, and its month ranking tells you nothing about timing.

> A seasonally adjusted series is built to erase the very pattern you are trying to read.

One limit I want to be plain about. For the not-seasonally-adjusted data, I recovered hires levels for total nonfarm only, March to July, plus the hires rate for December to April. I do not have four verified years for professional and business services, so the arithmetic below uses total nonfarm as a stand-in. When you do this for yourself, your industry's series replaces it line for line.

## How do you build a seasonal hiring index from NSA hires?

Divide each month's not-seasonally-adjusted hires by that year's monthly average, then average each calendar month across the years. A value of 1.08 means that month hires 8% more than a typical month. Build the index twice, once with 2020 and once without, so you can see how much the April 2020 collapse drags down the spring months.

BLS does not publish this simple ratio method. Its own adjustment uses X-13-ARIMA-SEATS, a statistical model that also deals with outliers. Since its pandemic-era benchmark revision, it has allowed for temporary changes and level shifts to capture seasonal patterns without removing the pandemic's effect. You don't need any of that. You need a rough shape across 12 months, and this simple index gets you there in a spreadsheet.

**Seasonal index sheet (paste into row 2, fill down)**

```
Column A  Year
Column B  Month number, 1 to 12
Column C  NSA hires level (thousands), from the JTU series
Column D  Year average     =AVERAGEIF($A$2:$A$61,A2,$C$2:$C$61)
Column E  Ratio to year    =C2/D2

Index, all years, month 1      =AVERAGEIF($B$2:$B$61,1,$E$2:$E$61)
Index, without 2020, month 1   =AVERAGEIFS($E$2:$E$61,$B$2:$B$61,1,$A$2:$A$61,"<>2020")
Difference, month 1            =(without-2020 cell) - (all-years cell)
Repeat the three index rows for months 2 to 12.
Sanity check - the 12 values in each index should average to about 1.00.
```

*Assumes 60 rows of NSA hires in A2:C61; widen the ranges if you pull more years.*

### What the example's partial index shows

With only five months of total nonfarm levels, I can only build a partial index, measured against the average of those five months (5,687 thousand). It is not the full 12-month index the method calls for, and the numbers are my own calculation:

- March: 0.92
- April: 0.98
- May: 1.05
- June: 1.08
- July: 0.97

The swing is modest, about 8% either side of the average. That matters later: a fairly flat curve means a fairly flat quota. Some industries swing much harder. In Table 9, retail trade hires went from 554 thousand in September to 673 thousand in October, a jump of about 21% in one month. Transportation, warehousing and utilities went from 251 thousand to 419 thousand, about 67%. Both percentages are my own calculation from the published figures.

### What the rate series adds for December and January

The hires rate is hires as a percent of employment. The not-seasonally-adjusted rate for total nonfarm (JTUHIR on FRED) fills the winter gap: 2.4 in December, 3.4 in January, 2.6 in February, 3.3 in March, and 3.4 in April. December is the low point of that stretch, and January is about 40% higher. Keep that in mind: it changes the answer to the December question in the procedure.

### The 2020 test

Nobody has published how much a simple index like this moves when 2020 is included or left out, so measure it on your own sheet and note the difference for each month. If April and May move by more than a few hundredths, use the version without 2020 and write down that you did.

## Do job postings confirm the hiring curve?

Only partly. The free national postings index has a not-seasonally-adjusted version that shows the shape of the year. Its sector series, though, are published only as seasonally adjusted change since February 1, 2020, based on fixed 2017 to 2019 patterns, so they flatten exactly the sector seasons you are looking for.

The free source here is the Indeed Job Postings Index. It covers Australia, Canada, France, Germany, Japan, the UK and the US at country and sector level, with February 1, 2020 set to 100. Each series, including the occupational sectors, is adjusted separately. Use the national not-seasonally-adjusted series for shape and the sector series for trend only. I could not establish whether unadjusted sector series are publicly available anywhere. Indeed also asks users to request permission before using the data, so treat it as a private cross-check, not something to republish.

What to look for: whether postings peak one to two months before hires. Sources disagree on which comes first, and I have not seen the gap documented for any one sector. Write down what you see in a one-line note rather than forcing it into the launch date.

### The gap between searches and postings

The posting data does settle one question cleanly. Laura Ullrich and Sneha Puri of Indeed Hiring Lab reported that job searches in January ran up to 31% higher than the early-December average. A secondary write-up quotes Indeed as saying postings never rose above their early-December level at any point that January. Indeed's UK data shows the same timing: the first or second Monday after New Year's Day typically has the most job searches of the whole year, while January postings typically peak at the very end of the month.

> **Watch out:** The first week of January is the crowded week, not the hiring week
>
> Searches peak on the first or second Monday after New Year's Day, but January postings peak at the end of the month. Launching in the first week means competing with the year's biggest crowd for openings left over from December.

Be careful with that 31%. One person runs many searches, so 31% more searches does not mean 31% more applicants. Nobody has published applicants per opening by month, so treat any page that quotes such a figure as unverified.

## How long does it take to get hired after applying?

For a private-sector business role, the median is about 30 days from application to hire. For a technical role it is about 40 days, and US federal roles average 101 days. Use a figure whose clock starts when you apply, because that is the clock your launch date runs on.

| Source | Clock | Business | Technical/exec |
|---|---|---|---|
| Ashby (median) | Application to hire | 30 | 40 |
| Ashby (median) | Job opened to first hire | 56 | 76 |
| SHRM (median) | Requisition to offer accepted | 39 | 45 (exec) |
| OPM (average) | Federal, all roles | 101 | n/a |

The rows are not interchangeable. Ashby's 30 and 40 days run from application to hire, across 109 million applications and 247 thousand jobs. Its averages from application to accepted offer are longer, about 38 and 48 days. Ashby's 56 and 76 days run from the day the job opens to the first hire, which includes time before you could have applied. SHRM's 39 and 45 days run from the day the employer opened the requisition (its internal request to hire) to an accepted offer, which is another clock that starts on the employer's side.

Other published figures help you calibrate:

- **Employ:** time to hire of 46.2 days, and 7.2 days from application to first screen, across 6,640 customers.
- **Greenhouse:** average time to fill rose from 43.6 days to 59.7 days between two measurement years, across 6,000+ companies. These lags drift, so re-check yours every time you re-plan.
- **OPM:** time to hire for mission-critical federal occupations ranged from 72.1 to 94.6 days, and Executive Order 14170 directs that government-wide time to hire be cut to under 80 days. Check OPM's current dashboard before relying on 101.
- **Seniority:** one secondary source puts senior roles at 71 days and junior roles at 52, 37% longer, but it does not say what it measures, so I would not subtract it. Use it only as a reason to lean toward the longer end if you are senior.

### The example's choice

A technical recruiter counts as a business role in Ashby's split, even though the hiring is for technical teams, so the example takes 30 days. The wrong turn I nearly took was using 56, the larger business figure from the same source, which counts from the day the job opens. Subtracting it would have pushed the launch 26 days, nearly four weeks, too early.

## Step by step: from hires curve to a dated launch week

The launch week is your target hiring month, minus any notice period you owe, minus the application-to-offer lag, minus a one-week buffer for the first screen. These eight steps take you from a blank sheet to that date and a monthly quota. I found no practitioner who documents this exact sequence, so treat it as assembled from the sources above, not as an established standard.

#### Hires curve to launch week and quota

1. **Map the role to a JOLTS industry** - Map your employer's industry, not your job function, to one supersector. Done is one series ID that starts with JTU and ends in HIL.
2. **Pull NSA hires levels, not SA** - Download the JTU hires level for the last four full years plus 2020. Done is a 60-row sheet whose series notes say Not Seasonally Adjusted.
3. **Build the seasonal index twice** - Divide each month by its year's average, then average each calendar month across years, with and without 2020. Done is two 12-value indexes and the difference for each month.
4. **Cross-check against postings** - Compare the national not-seasonally-adjusted posting series with your index, and use the sector series for trend only. Done is a one-line note on whether postings peak one to two months before hires.
5. **Choose one lag in days** - Pick the median from application to offer for your role type: about 30 for business, 40 for technical, 101 for federal. Done is one number with its source and clock written beside it.
6. **Back-date the launch week** - From your target month, subtract any notice period, the lag, and a 7-day buffer for the first screen. Done is a Monday date in your tracker.
7. **Overlay fixed hiring cycles** - Mark campus offer deadlines, federal lead times, and retail or logistics peak season if your role sits in one. Done is the weeks to avoid and the weeks to push harder marked on the calendar.
8. **Set the monthly send quota** - Divide your application budget across the months of your search, weighting each month by the index of the month those applications will land in. Done is 12 numbers in the tracker that add up to your budget.

#### Working back from the hiring peak to a launch week

1. **Target month** - The first month of your peak run in the not-seasonally-adjusted index
2. **Notice period** - Subtract any notice you owe before you can start
3. **Lag** - Subtract the median days from application to offer for your role type
4. **Buffer** - Subtract about 7 days, the median wait for a first screen
5. **Launch week** - The Monday of the week you are left with

*The launch week is the target hiring month, minus any notice period, minus the application-to-offer lag, minus a one-week buffer for the first screen.*

### The example, dated

- **Target:** May is the first month at or above 1.00 in the partial index, so the median offer should land on May 1. Aiming at June, the single highest month, would have put the launch in late April and spent the first half of the busy season waiting for offers.
- **Notice period:** the example seeker is between jobs, so this is zero. If you owe notice, subtract it here.
- **Lag:** 30 days back from May 1 gives April 1.
- **Buffer:** Ashby's 30 days already includes the first screen, so subtracting Employ's 7.2 days counts it twice. I subtract it anyway, on purpose, as a one-week safety margin: by definition, half of all offers take longer than the median.
- **Launch week:** the last full week of March. That Monday is the date that goes in the tracker.

Two variations on the same curve:

- **Technical role, 40 days:** May 1, minus 40 days, minus 7 days, lands in mid-March.
- **Federal role, 101 days:** to be hired by October 1, apply around June 21. That is roughly three times the private-sector lead time.

### Is December a bad time to job search?

No. December's hires rate (2.4) is the low point, but December is the launch month for January (3.4). An application sent in the first week of December, with a 30 to 40 day lag, lands as a January offer, and it arrives ahead of the crowd that starts searching in the first week of January. December's slowdown tells you when people start new jobs, not when they sent the applications that got them there.

## Which fixed hiring calendars override the curve?

Campus and early-careers programs, federal hiring, and the retail and logistics peak season run on fixed calendars and make their decisions before the general market moves. If your role sits inside one, work back from that calendar's deadlines, not from the hires curve.

#### Curve-led or calendar-led

Horizontal axis runs from Flat hires curve to Strong seasonal swing. Vertical axis runs from No fixed calendar to Fixed calendar and deadlines.

| Quadrant | What it means |
| --- | --- |
| Calendar-led | Work back from the program's offer deadline or start date; ignore the curve |
| Calendar plus swing | Work back from the deadline and push hardest in the run-up month |
| Steady state | Launch when ready and keep the quota flat |
| Curve-led | Work back from the peak in the not-seasonally-adjusted index and weight the quota by it |

*The more your role runs on a fixed calendar, the less the general hires curve should set your launch week.*

### Campus and early-careers programs

Roughly 70% of employers recruit in both fall and spring. The share of offers made in fall and winter has fallen from 81% to just under 64%, so check the current share before you rely on the fall round alone. At the University of Virginia, full-time offers made during fall recruiting carry a deadline of October 30 or 3 weeks, and spring offers February 15 or 3 weeks. NACE puts employers' time from interview to offer at 27.3 days, against a 9-day window for candidates to answer. At that pace, an interview in mid-September produces an offer around mid-October, in time for a late-October deadline. NACE also advises employers not to hire for positions more than 12 months before the start date.

| Market | Campus recruiter profiles | Top sample employer |
|---|---|---|
| US | 944 | Jane Street (4 of 25) |
| UK | 109 | Jane Street, Teach First (3 of 25 each) |
| US to UK ratio | 8.7x | n/a |

In Refolk's index, the same quantitative trading firm tops both the US and UK samples of campus, university and early careers recruiters. The UK search also included Graduate Recruiter. That fits structured programs tied to the academic calendar. But each sample is 25 profiles, so read it as a sign of where calendar-driven hiring sits, not as a ranking of employers.

### Retail and logistics peak season

These industries hire heavily in October: in Table 9, retail trade went from 554 thousand hires in September to 673 thousand in October, and transportation, warehousing and utilities from 251 thousand to 419 thousand. One Indeed Hiring Lab report found seasonal postings up about 10% year on year as of September 30. With a lag of about 30 days, an October hire means sending in September. That makes September the month to push hardest, and late August the launch.

### Federal hiring and what I could not establish

Federal roles need about three times the private-sector lead time, as dated above. I could not establish how federal posting volume moves around October 1, or when K-12 and higher-education postings open. If your role is in one of those, look at last year's postings from the agency, district or university and note the dates they opened.

## How does this go wrong?

Most failures come from reading the wrong series, mixing up clocks that start in different places, or treating a stand-in measure as a real count. Each failure below says what it looks like and the check that catches it.

1. **Reading a seasonally adjusted series as if it showed seasons.** It looks like March or another quiet month coming out on top. The adjusted series ranks March first and the raw series ranks it fifth, so the ranking says nothing about timing. Check: the series ID starts with JTU.
2. **Using Indeed's sector data for seasonal shape.** It looks like a sector curve that is suspiciously flat. The adjustment is fixed to 2017 to 2019 patterns, so the seasons have been smoothed out. Check: the column name is index_NSA, not index_SA, and if no unadjusted version exists, use the national series for shape.
3. **Treating search counts as headcount.** It looks like "31% more competition in January". One person runs many searches. Check: label any applicants-per-opening figure as unverified, because none has been published.
4. **Mixing up clocks.** It looks like a launch date a month too early. Ashby's 40 days (application to hire) and Ashby's 76 days (job opened to first hire) measure different things. Check: write down where each figure's clock starts and stops before you subtract it.
5. **Letting 2020 skew the index.** It looks like a spring that seems weaker than it really is. The April 2020 collapse drags down the spring months. Check: run the index with and without 2020 and note the difference for each month.
6. **Citing pages that call themselves "data" but show no source.** It looks like confident percentages, such as 200 to 300% more applications in January, with no method. Check: leave out anything without a named dataset.
7. **Reading too much into a small sample.** It looks like "17 in 25 recruiters are technical, so that is the market". That split comes from 25 profiles, not 28,256. Check: present sample splits as samples, with the sample size beside them.
8. **Ignoring revisions to preliminary figures.** It looks like a peak month that moves after you have already set your launch date. The latest month is marked (p) for preliminary, and seasonally adjusted data for the most recent 5 years is recalculated every year. Check: pull the data again after each January JOLTS release.

## Setting the monthly send quota and keeping the calendar current

A month's quota is your application budget divided by the number of months in your search, multiplied by the index of the month those applications will land in. Round so the total still matches your budget. Weight by the month the applications land in, not the month you send them, because the lag moves every application forward by about a month.

The example seeker set a budget of 60 applications from the launch week to the end of June. That number is their own choice, not a benchmark. Applications sent in the last week of March count toward April's quota.

| Send month | Offers land in | Landing index | Applications |
|---|---|---|---|
| Late Mar and Apr | May | 1.05 | 20 |
| May | Jun | 1.08 | 21 |
| Jun | Jul | 0.97 | 19 |

The spread is narrow because total nonfarm swings only about 8% either side of average. A retail or logistics curve would spread the same budget much more unevenly. For a 12-month plan, use your full index: divide your annual budget by 12, multiply each month's share by the index of the following month (for a lag of about 30 days), and rescale so the 12 numbers add up to your budget.

Quality is what slips in a 21-application month. Each one needs a resume tailored to that posting and a cover letter that answers it. [Refolk](/candidates) writes your resume from your own work history, tailors it to each posting and drafts the cover letter, so an application in your busiest month gets the same care as one in your slowest. Refolk also scores how well you fit each posting, which is the right way to decide which postings make it into the quota and which you skip.

**Launch plan block for your tracker**

```
LAUNCH PLAN - [role], [level], [industry]
Hires series: FRED [JTU series ID], NSA, pulled [date], years [list], 2020 excluded [yes/no]
Target month: [month] (index [value])
Lag: [days] days, source [name], clock = application to hire
Buffer: 7 days (median application to first screen)
Notice period: [days]
Launch week: week of Monday [date]
First-screen checkpoint: [launch Monday + 7 days]
Fixed-cycle weeks: avoid [weeks] / push harder [weeks]
Quota (total [budget]):
Jan [n] | Feb [n] | Mar [n] | Apr [n] | May [n] | Jun [n]
Jul [n] | Aug [n] | Sep [n] | Oct [n] | Nov [n] | Dec [n]
Re-pull: after the next January JOLTS release
```

*Fill every bracket from your own sheet; keep the clock and source lines, they are what you check when the plan is revised.*

### Keeping the calendar current

Pull the hires series again after each January JOLTS release, rebuild both indexes, and change the launch week only if your target month changes. Check your lag source at the same time, because published lags drift. If you have no first screens by the checkpoint date, the problem is the applications, not the calendar.

#### Before you call the launch plan done

- [ ] The hires series ID starts with JTU and its notes say Not Seasonally Adjusted.
- [ ] The sheet holds at least four full years plus 2020, with the latest preliminary (p) month flagged.
- [ ] Both index versions exist, with and without 2020, and the plan names the one it uses.
- [ ] The 12 values in the index you use average to about 1.00.
- [ ] The lag's clock starts at the application, and its source is written beside it.
- [ ] The launch week is a Monday date, not just a month.
- [ ] Any fixed-cycle deadlines that apply to the role are marked as weeks to avoid or weeks to push harder.
- [ ] The monthly quotas add up to the stated application budget.
- [ ] No figure in the plan comes from a page without a named dataset.
- [ ] The tracker holds a date to pull the data again after the next January JOLTS release.

## Frequently asked questions

### What is the best month to apply for jobs?

No single month is best for everyone. The best month is roughly one application-to-offer lag before your own industry's hiring peak. In the total nonfarm months I pulled, not-seasonally-adjusted hires peaked in June, so for a business role with a 30-day median lag, most applications should go out in May. Industries with sharp seasons, like retail before October, move that window a lot. Build the curve for your own industry before you trust any calendar.

### Is December a bad time to job search?

No. December is often the right time to apply for a January start. The not-seasonally-adjusted hires rate fell to 2.4 in December and rose to 3.4 in January. An application sent in early December, with a 30 to 40 day lag, lands as a January offer. It also gets in ahead of the crowd that starts searching in the first week of January, when searches spike but postings have not yet risen above their December level.

### How long does it take to get hired after applying?

For private-sector roles, the median is about 30 days from application to hire for business roles and about 40 days for technical roles, based on Ashby's data across 109 million applications. Averages from application to accepted offer run longer, about 38 and 48 days. US federal hiring averaged 101 days. Always check where the clock starts. Figures that start when the job opens, such as 56 or 76 days, include time before you could have applied.

### When should I start my job search?

Start in the week you get by working backwards from your target hiring month. Subtract any notice period you owe, then your role's median application-to-offer lag, then about one week as a buffer for the first screen. For a business role aiming at a May 1 offer, that means launching in the last full week of March. Technical roles start about ten days earlier, and federal roles need roughly three times the private-sector lead.

### Why does the JOLTS headline number show different peak months than this method?

The headline JOLTS figures are seasonally adjusted, which means the calendar pattern you are looking for has been deliberately removed. In five months of total nonfarm hires, the adjusted series ranks March first while the raw series ranks it last. Use the series whose FRED ID starts with JTU and whose notes say Not Seasonally Adjusted. The JTS series is the right tool for spotting trends, but useless for picking a month.

### Can I use Indeed's sector posting data to find my industry's hiring season?

Not for the seasonal shape. The sector series are published as seasonally adjusted percent change since February 1, 2020, and the adjustment is based on fixed 2017 to 2019 patterns, which flattens the seasons you want to see. Use the national not-seasonally-adjusted series for shape and the sector series for trend only. Indeed asks people to request permission before using the data, so keep it as a private cross-check.

---

*From the Refolk guide library. I revise these guides rather than replacing them, so the current version is always at https://www.refolk.ai/candidates/guides/job-search-launch-week-hiring-calendar*