Junior data jobs come down fastest. Staff jobs stay up for months.

By ,

Junior postings come down fastest. 32% of junior data job postings were gone within two weeks, against 22% of all data postings and 14% of staff ones.

We read employers’ own hiring systems every day, so we see the day a posting appears and the day it comes down. These figures cover 13,537 US data jobs first posted between April and September 2026.

22%
of data job postings came down within two weeks
32%
of junior postings did
5%
of postings on Ashby, a hiring system used mostly by startups

How fast postings come down by level

  • Intern708 postings15%
  • Junior996 postings32%
  • Mid-level2,406 postings25%
  • Senior5,637 postings21%
  • Lead914 postings22%
  • Staff699 postings14%
  • Principal550 postings20%
Share of postings at each level that were gone within two weeks. The dashed line is the share for all data jobs.

A junior posting is more than twice as likely as a staff one to be gone in two weeks. Within a month, 45% of junior postings were down, against 31% of senior ones.

Junior roles draw the most applicants, and an employer can close the posting once it has enough of them. If you are early in your career, the posting date matters more for you than for anyone else. Apply in the first week.

Internships are the exception at the junior end. Only 15% came down within two weeks, because employers post them months ahead and leave them up through a hiring season.

1,176 postings gave no sign of their level. 27% of those came down within two weeks.

Large employers take postings down sooner

  • 1 to 4 postings17%
  • 5 to 19 postings19%
  • 20 to 99 postings29%
  • 100 or more postings28%
Share of postings that came down within two weeks, by how many data postings the employer had in this set.

At employers with fewer than five data postings, 17% came down within two weeks. At employers with 20 to 99, 29% did.

The likeliest reason is staffing. A large employer has recruiters who close a posting when the role is filled or enough people have applied. At a small one, a posting can stay up until someone remembers it.

The hiring system tells you about the employer

  • Down within two weeks
  • Still up after three months
  • SmartRecruiters638 postings
  • Phenom754 postings
  • Workday4,302 postings
  • Eightfold622 postings
  • SuccessFactors433 postings
  • Oracle693 postings
  • iCIMS709 postings
  • Greenhouse2,171 postings
  • Lever406 postings
  • Ashby463 postings
For each hiring system, the share of postings gone within two weeks and the share still up after 90 days. The further apart the dots, the more that system’s employers lean one way.

You can usually see which hiring system an employer uses in the address of its application page. 45% of postings on SmartRecruiters came down within two weeks. On Greenhouse it was 10%, and on Ashby 5%.

The long end is the reverse. 64% of Ashby postings were still up after three months, against 16% on SmartRecruiters and 40% of all data postings.

The system itself is not what closes a posting. Greenhouse, Lever and Ashby are mostly used by startups and technology companies, which tend to keep a posting open while they keep hiring for it. Workday, Phenom and SmartRecruiters are mostly used by large employers.

How fast postings come down by industry

  • Down within two weeks
  • Down in weeks three and four
  • Still up after a month
  • Healthcare
  • Manufacturing
  • Finance
  • Retail
  • Technology
  • Consulting
What had happened to postings in each industry a month after they appeared.

Healthcare employers took down 27% of their data postings within two weeks. Technology employers took down 13%, and consulting firms 9%.

Consulting firms often hire for the same role all year, so their postings stay up. The same goes for machine learning and AI engineer roles in any industry. 13% of those came down within two weeks, against 22% to 24% for the other data roles.

Some fast postings come back

40%
of postings that were later posted again came down within two weeks
20%
of postings that were posted once did

1,137 of these postings were put up again later as a new posting for the same role. They were about twice as likely to have come down fast the first time.

So a posting that disappears in a few days has not always been filled. If you missed one, look at the employer’s page again a week or two later. Our job pages say when a posting has been posted before.

What made no difference

Remote, on-site and hybrid postings came down at about the same rate in the first two weeks: 24%, 21% and 19%. Remote postings do come down sooner over a longer stretch, but the gap opens after the first month.

Whether the posting stated pay, whether it offered visa sponsorship, how long the description was and which weekday it went up all made little or no difference.

What these figures leave out

A posting that comes down has not always been filled. Some roles are cancelled, some are closed to new applicants while interviews go on, and some are posted again.

We assign each posting a level. For 8,340 of them it comes from the title. For the rest we use the employer’s own level field, the level the description states, or the years of experience it asks for. 9% of postings give none of these. The level chart leaves out 451 manager, director and vice president postings.

Level, employer size, hiring system and industry overlap. A junior posting at a large healthcare employer counts in three of these charts. Junior postings still come down fastest among Workday employers alone, which is the one overlap we checked.

How we counted

These are 13,537 US postings for data analysts, analytics engineers, data engineers, data scientists, and machine learning and AI engineers, first seen on employers’ own hiring systems between 1 April and 24 September 2026. We left out postings from an employer’s first week with us, when postings that were already up would look new.

A posting counts as down when the employer closed it, it disappeared from their board, or we had not seen it for 14 days. Every posting here had at least 30 days in which it could come down before we counted. The three-month figures use the 4,545 postings first seen by 19 July, which all had 90 days.

This is a smaller set than the one in how long postings stay up, which also counts postings from October and from employers’ first week.

You’re welcome to quote these figures or redraw the charts. Please link to this page as the source.