Internship Deadlines vs Reality: What 14,000 Postings Show About When Jobs Actually Close | Intern Insider
Internship Deadlines vs Reality: What 14,000 Postings Show About When Jobs Actually Close
A third of internship postings are gone before their own stated deadline arrives. We looked at when postings really close, and which deadlines you can trust.
About a third of internship postings are already gone by the time their own stated deadline arrives.
Students ask us a version of this constantly. A posting says "apply by March 15," so does anything actually happen on March 15? Because we track when postings appear on company career pages and when they come down again, we can answer it directly.
The short answer is that it happens less than half the time. The more useful answer is that the deadlines which fail do so in a pattern, and the pattern is predictable enough to plan around.
Intern Insider monitors the career pages of 200,000+ companies to detect newly posted roles and to notice when they come down. For this analysis we took every posting that stated an application deadline in its description, and compared that date to the date the posting actually left the employer's careers site. That gave us 14,077 postings from March 2025 to July 2026.
For students
Looking for an internship right now?
We watch the career pages of 200,000+ companies and post new internships the moment they go live. Set your preferences once and get matched with roles that fit.
Fewer than half of postings close when they said they would#
We counted a posting as closing on time if it came down within three days either side of its stated deadline.
When postings actually closed, relative to their stated deadline
More than 3 days early
31%
Within 3 days
44%
3 to 7 days late
9%
More than 7 days late
16%
A posting counts as on time if it closed within three days either side of the stated date.Source: Intern Insider postings, Mar 2025 - Jul 2026 (n = 14,077)
So 44% land on target, 31% vanish early, and 25% linger past the date. About 8% were still listed a month after their deadline had passed.
That is better than we expected going in, and it is worth saying so. Most employers are not making the date up. But a quarter of postings outlive their own deadline, and close to a third never reach it.
For a student, the interesting question is not whether a deadline is accurate in either direction. It is whether the job will still be there when you get around to applying.
Share of postings already closed, counting backwards from the stated deadline
30 days before
16%
21 days before
19%
14 days before
22%
7 days before
27%
On the deadline
34%
Read this as: with 7 days still to go before the stated deadline, 27% of postings had already come down.Source: Intern Insider postings, Mar 2025 - Jul 2026 (n = 14,077)
A full month before the stated deadline, 16% of these postings were already gone. A week before, 27%. If you saw a posting, told yourself you had until the deadline, and came back at the last moment, about one in three times there was nothing to come back to.
This is the sturdiest finding in the analysis, and it held steady across every version of it we ran. The deadline is not a countdown. It is an optimistic upper bound that a third of postings never reach.
Deadlines are not uniformly unreliable. Their reliability depends almost entirely on how far ahead they were set.
The first column below is how many postings fall into each band, so it adds up to every posting in the analysis. The three columns after it are what happened to that band, so each row adds up on its own.
Stated deadline was
Share of postings
Closed early
Closed on time
Closed late
Under 2 weeks out
33%
9%
58%
33%
2 to 4 weeks out
26%
21%
51%
28%
1 to 2 months out
20%
36%
43%
21%
2 to 4 months out
15%
64%
17%
19%
4+ months out
6%
92%
6%
2%
A deadline two weeks away is a real operational date. Somebody intends to review applications and close the posting, and they mostly do. A deadline six months away is closer to a placeholder, and the posting behaves like any other posting. It comes down when the role is filled, which is typically about three weeks later, nowhere near the stated date.
The share column is worth reading alongside the rest. That bottom row looks alarming at 92% closing early, but only 6% of postings set a deadline that far out. Most employers are in the top two bands, where the date is set within a month and holds up more than half the time.
That is why the median posting in this set stayed up 20 days against a median stated window of 23 days. The two roughly match in aggregate. It is the variance that catches people out, and the variance is driven almost entirely by how far ahead the date was set.
Canadian employers hold to their dates more often#
The raw comparison here looked dramatic, and part of it turned out to be an artifact, so it is worth showing the work.
American employers set deadlines much further out than Canadian ones: a median of 32 days versus 14. Given what we just established about long deadlines, that alone could produce a gap that is not really there. So we compared only postings with similar deadline lengths.
The gap survives, and it is large. Each figure below is the share of that country's own postings that closed on time, so the two columns are separate rates rather than a split of one total.
Stated deadline was
US closed on time
Canada closed on time
Difference
Under 2 weeks out
50%
66%
16 points
2 to 4 weeks out
46%
58%
12 points
1 to 2 months out
39%
56%
17 points
Canadian employers are 12 to 17 points more likely to close on the date they named, at every deadline length we can compare. That is not a rounding difference.
Our best guess is the co-op calendar. Canadian student hiring is organised around university terms with fixed application windows, and a posting tied to a co-op cycle has a real reason to close on a specific day. We cannot prove that from this data, so treat it as a hypothesis rather than a finding.
Industry differences are large and they survive the same control. These are all postings whose stated deadline was two to four weeks out, so employers who default to long deadlines are not penalised.
Share of postings closing within three days of their stated deadline, by industry
Utilities
75%
Pharmaceuticals
69%
Banking
68%
Hospitals and Health
66%
Financial Services
59%
Government
52%
Non-profits
43%
Aerospace
15%
Restricted to postings whose stated deadline was 2 to 4 weeks after posting, so industries with longer default deadlines aren't penalised.Source: Intern Insider postings, Mar 2025 - Jul 2026 (n = 1,858)
Banking and utilities stand out for a reason beyond the headline number: only 9% of their postings closed early. When a bank names a date, the posting is there until that date. That fits what students already know about finance recruiting, which runs on fixed cycles with hard cutoffs.
Aerospace sits at the other end. Even with a stated deadline only two to four weeks out, 79% of aerospace postings were gone more than three days early. Whatever those dates mean internally, they do not describe when the posting comes down.
Company size matters more than we expected. On-time rates run 45% for employers over 1,000 people, 40% for mid-sized companies, and 31% for companies under 50, and the direction is consistent across all three tiers.
Smaller employers are likelier to name a date and then either fill the role early or leave the posting up long after. That is roughly what you would expect when nobody owns the job posting as a process.
A stated deadline tells you when a posting definitely will not be there anymore, and less than you would like about when it will.
The useful moves are unglamorous. Apply when you find the posting rather than when you have scheduled time for it, because a third of the time the deadline you are planning around has already stopped mattering. Give more weight to a deadline that is two weeks out than one that is three months out. And if you are looking at banking, utilities or pharma, you can plan around the date. In aerospace, you cannot.
We will rerun this in a year. One thing a single cycle cannot tell us is whether any of this is drifting, and we would want a second year of data before calling anything a trend.
